Google Gemini Strikes Back, China Hits NVIDIA on Antitrust | Bill Bishop, Pete Shadbolt, Sadi Khan, Lukas Czinger, Kyle Samani

Summary

TBPN Podcast Summary

Episode: Google Gemini Strikes Back, China Hits NVIDIA on Antitrust
Date: September 15, 2025
Hosts: John Coogan & Jordi Hays
Notable Guests: Bill Bishop (Sinocism/Sharp China), Pete Shadbolt (PsiQuantum), Saadi Khan (Aven), Lukas Czinger (Divergent), Kyle Samani (Multicoin/Ford Industries)

Episode Overview

This episode dives deep into the dramatic shifts in AI competition—highlighting Google Gemini’s surge past ChatGPT, China’s regulatory actions against Nvidia in the context of US-China tech tensions, and the explosive growth in open-source Chinese LLMs. The hosts break down the business models, market adoption, and implications of these shifts, with help from expert guests commenting on geopolitics, quantum computing, fintech, hardware innovation, and crypto treasury vehicles. The show also spotlights milestone fundraises and playful industry moments, capturing the frenetic energy at the AI-technology frontier in late 2025.

---

Key Topics & Insights

1. Google Gemini vs. ChatGPT — The Empire Strikes Back

• Gemini’s Surge:

  • Gemini overtakes ChatGPT in global Google searches following its "Nano Banana" feature release.
    • [05:15] “Gemini has crossed ChatGPT in worldwide search results, which is even ... an important metric to be tracking.” — John
  • Despite U.S. dominance remaining with ChatGPT, countries like India drive Gemini’s rise.
    • [07:54] “India is absolutely spiking ... Gemini is flatline and then just completely surged.” — John

• App Store Battle:

  • Gemini now #1 on App Store productivity; ChatGPT still leads in US total ratings.
  • The effect attributed to feature drops (VO3, Nano Banana), clever UX, and Google's marketing muscle.
    • [10:23] “VO3 got me to fork over $250 a month ... I feel like I’m getting access to a frontier model I can’t get anywhere else.” – John

• Interpretation:

  • Surge partly due to Google’s global reach and potential preferential promotion in international markets.
  • Discussion around possible "astrology effect” is debunked.
  • The rise signals a maturing, competitive duopoly—good for users, not necessarily bearish for OpenAI.
    • [21:10] “My read on Gemini doing well is not that it’s bearish for OpenAI at all. No, it means it’s an extremely important market ... a handful of players will win.” — Jordi

Notable Quote:

[09:25] “AI mode is within the Google search product Gemin, and to be clear, Gemini is doing really well. ... It’s been a good value trade.” — John

2. Open Source LLMs and China’s Flip

• China surging in open-source AI:
  • Chinese LLM models downloaded on Hugging Face are overtaking U.S. counterparts.
    • [21:46] “Open source models from China are set to overtake the US as downloads surge ... the flip is happening right now.” — John
  • Debate on whether open source LLMs will capture value like closed system platforms (Linux vs Android/iOS analogy).
• U.S. Tech Landscape:
  • Google releases Gemma line; Meta and OpenAI also play in open source, but China’s momentum is strong.
    • [24:48] “Google just has Gemma ... mostly like pretty small models ... Meta is really the only big player and open source in the US.” — Tyler

Notable Quote:

[22:47] “A lot of the closed source operating systems have captured like 99% of the value.” — John

3. Nvidia, China, and U.S. Trade Tensions (with Bill Bishop, [41:47-61:02])

• China’s Antitrust Probe:

  • Nvidia investigated in China for a years-old acquisition, timed alongside major US-China trade negotiations.
    • [43:02] “The timing of course is not coincidental ... it is part of a way to further encourage Nvidia and the US Government to pull back on export controls.” — Bill Bishop
  • Penalties could be severe—up to 10% of global revenue, possibly multiplied.

• Larger Context:

  • Chinese government playing multi-dimensional chess: possibly discouraging Nvidia chips to spur domestic industry, while Alibaba et al. want the best hardware regardless.
  • Huawei’s claims about domestic AI chip superiority examined skeptically.
    • [47:20] “The security forces ... may have decided to just rip off the Band-Aid and move to indigenous chips.” — Bill Bishop

• TikTok Divestiture Update:

  • U.S. & China reach a “framework” deal, but questions remain on whether it addresses real security concerns.
    • [51:12] “Only thing we didn’t contemplate was a president who didn’t enforce the law.” — Bill Bishop

Memorable Exchange

[50:23] “The narrative kind of turned to ... well, Trump kind of liked TikTok ... the cloud contracts angle ... it’s like, if you had a rival newspaper and you were given just the printing press but not editorial control, would you be satisfied?” — Jordi & John

4. Massive Fundraising & Guest Spotlights

Quantum Computing with PsiQuantum's Pete Shadbolt ([150:26–164:43])

• Announcement: $1B new funding round to build a million-qubit, photonics-based quantum computer, with anchor projects in Australia and Chicago.
  • [151:08] “We are now announcing a billion dollar fundraise for us to go ahead and actually build giant quantum computers.” — Pete Shadbolt
• Differentiation:
  • Uses semiconductors, high-volume photonics; candid about the technical risks and government due diligence.
• Use Cases:
  • Near-term focus on chemistry, material science, drug discovery—domains where conventional compute struggles.

Huge Fintech and Industry Rounds

• Aven (Saadi Khan): $110M Series E for a home equity-backed credit card offering instant, low-interest borrowing and cashback. Key: compressing a 30-day process to 15 minutes, including remote wet signature via robot arm.
• Divergent (Lukas Czinger): $290M Series E to scale manufacturing platform for defense and auto (notably their 21C hypercar), focusing on digital, flexible onshore production facilities.
• Ford Industries (Kyle Samani): $1.65B raised in 2 weeks for a Solana-based digital asset treasury—aiming for scale and M&A in the on-chain asset management space.
  • [181:14] “Spent two weeks fundraising and in two weeks we raised 1.65 billion ... definitely the most hectic fundraise I’ve ever done in my life.” — Kyle Samani

5. Industry Metacommentary, AI Podcast Farms, and Notable Startup Moments

• AI-generated Podcasts:

  • Inception Point AI launches 5,000 AI-hosted podcasts, highlights the proliferation of “slop” (low-value) content and the algorithmic podcast discovery war as LLMs flood channels.
    • [193:47] “Launched 5,000 podcasts, make 3,000 episodes a week ... break even if only 20 people listen. ... all for $1 per episode.” — John, reading from Trung Fan’s coverage

• Legacy vs. New Media:

  • Hosts muse on the acceleration of bot-created content and the coming necessity of institutional brands as quality filters (“search ‘founders podcast + name’ now”).
  • Notable “comments that aged like milk” from Hacker News, including dismissals of Dropbox, Bitcoin, Airbnb, and Cursor AI in their earliest moments.
    • [141:23] “When I was your age, Hacker News was the place. Every single day, opening up, reading threads.” — John
    • [145:05] “First real-world bitcoin transaction in 2010 ... two pizzas for 10,000 bitcoin.” — Jordi

• Miscellaneous Highlights

  • Lighthearted banter about quarterly vs. semi-annual financial reporting (Trump proposal [33:32]), personified LLM voices (Poke AI), collector Apple iPhones, and ticker symbol lore (“YOLO” cannabis ETF down 87%).
  • Tim Cook’s iPhone 17 launch sound bite and the “orange iPhone” as a nod to tech thought leaders.
    • [62:15] “We want a Daytime Emmy ... Tim out there flogging the phone like a guy selling pocket watches.” — John

• AI and the Broader Workforce:

  • Debate over real impacts—prompt engineering, jobs, and whether Americans really want to return to hunting gazelles for a living.
    • [132:32] “Sci-fi concerns have landed on deaf ears ... talk to me about, can I get a job tomorrow?” — John

---

Timestamps for Important Segments

| Time | Topic / Segment | |------------|--------------------------------------------------------------------------------| | 05:15 | Gemini’s surge on Google Trends/App Store | | 09:25 | Gemini vs. ChatGPT: product/feature comparison | | 21:46 | China’s open-source models surpassing US on Hugging Face | | 41:47 | Nvidia’s China antitrust headache — guest Bill Bishop | | 54:06 | TikTok divestiture deal, tech-trade complexities | | 150:26 | Pete Shadbolt announces $1B quantum round for PsiQuantum | | 165:02 | Saadi Khan/Aven: robot arms for notarization, Series E funding | | 171:31 | Lukas Czinger/Divergent: defense/auto manufacturing scale, Series E | | 180:54 | Kyle Samani/Ford: $1.65B digital treasury raise, Solana, M&A thesis | | 193:47 | AI podcast farms, the flood of auto-generated content |

---

Memorable Quotes

• “Gemini has crossed ChatGPT in worldwide search results ... important metric.” — John, [05:15]
• “My read on Gemini doing well is not that it’s bearish for OpenAI at all. ... a handful of players will win.” — Jordi, [21:10]
• “The timing ... is not coincidental. ... a way to further encourage Nvidia and the US Government to pull back on export controls.” — Bill Bishop, [43:02]
• “We are now announcing a billion dollar fundraise for us to ... actually build giant quantum computers.” — Pete Shadbolt, [151:08]
• “Only thing we didn’t contemplate was a president who didn’t enforce the law.” — Bill Bishop, on TikTok, [51:12]
• “Spent two weeks fundraising and in two weeks we raised $1.65B ... most hectic fundraise I’ve ever done.” — Kyle Samani, [181:14]
• “Launched 5,000 podcasts, make 3,000 episodes a week ... break even if only 20 people listen.” — John (from Trung Fan), [193:47]

---

Episode Highlights & Tone

• Rich banter and humor: Celebratory moments (e.g., no longer being in “quotes” in media, Wall Street Journal feature) mix with frank, technical market commentary.
• Insider takes: Critiques of viral LLM trends, startup war stories, and “slop” AI content, balanced by real admiration for technical execution and massive fundraising.
• Current events + business model depth: Real focus on how the AI and semiconductor battles will play out—globally and in the U.S.—with clear grasp of both technical and business stakes.

---

Conclusion

This episode captures a pivotal moment: Google’s Gemini mounts a serious challenge to OpenAI, the global LLM arms race accelerates (especially with China’s open source wave), and the geopolitics of chips reach a new fever pitch—with Nvidia as the prime target. Major industry guests break down what’s happening in quantum, fintech, hardware, and crypto, all while celebrating the playful, relentless, slightly absurd energy that defines the modern tech climate.

For a pulse on where AI, tech, and startup culture stands in late 2025, this is essential listening and a testament to how fast “the future” is arriving.

Transcript

[A]

You're watching TVPN.

[B]

Today's Monday, September 15, 2025. We are live from the TVPN Ultradome. The temple of technology, the fortress of finance, the capital of capital. Oh, we got some. We got some trust in the background. We're.

[C]

Trust up.

[B]

Trust up. We have made another improvement to the TVP and Ultradome. Gonna work on having more guests in person, more fun activities. We got a bunch more props. It's gonna be a great, great week. Great month, great year, great decade, great lifetime.

[C]

Fantastic to be back.

[B]

It is fantastic to be back.

[C]

It's so hard to be away from the ultradome.

[B]

It is good time. We have some breaking news. Of course. Gemini is at the top of the App store channel.

[C]

Well, we have some of our own breaking news.

[B]

Yeah, yeah.

[C]

Actually, I don't know if we're allowed to break this news yet. Dylan, are we good to break the news?

[B]

We are good to break the news.

[C]

We're good to break the news.

[B]

Let's tell everyone. So.

[C]

So we are going to be having our first Mag7 CEO this week, and he is none other than Mark Zuckerberg.

[B]

Zuckerberg.

[C]

You can see a totally real picture of him here. So we're going to be taking the show on the road Wednesday. We're going to be up in the bay live from, from Meta Connect. We're very excited. There's going to be a bunch of other guests tuning in. I will wait to share the exact names, but they'll be going out on our X and our substack over the next couple days. So very, very excited for this one. They have some crazy announcements and we'll be there on the ground reporting live.

[B]

It's all part of the long con to get meta on ramp. Ramp.com, time is money save. Both easy to use. Corporate cards, bill payments, accounting, and a whole lot more all in one place. Go to ramp.com to get started. So in other mag seven news, Gemini surging.

[C]

What about. What about the mag? Podcasting.

[B]

Okay, we got to take me through it.

[C]

Talk about the other deck.

[B]

Another.

[C]

No, no, no, no.

[B]

You're good doing whatever. Whatever.

[C]

This is more important.

[B]

This is amazing. They didn't put us in quotes.

[C]

They didn't put us in quotes.

[B]

They didn't put us in quotes. We made it. We made it.

[A]

We finally arrived.

[C]

A lot of people have so good EVPN in quotes. They're claiming that, like, you know.

[B]

Yeah, oh, tvpn, whatever that means. Is that really the name of the show or is it something else?

[C]

The journalists claimed historically that it was because we were a show. Yes, but it really we were reading between.

[B]

I learned that when you're writing in English class back in high school, I learned that if it's a book title, you put it in italics. I thought quotes were only reserved for, you know, if you're not sure and it's a quote from somebody else saying something and you don't want to put it in your own words, you don't trust the source necessarily. I always viewed quotes as something of a point of skepticism. So it is fantastic to see that we have escaped the jail of quotes in the media.

[C]

So very excited for this one.

[B]

So appropriate that it happened in the Wall Street Journal, one of our favorite publications.

[C]

Our favorite ever. This is. We will a lot of people like to make comments about legacy media. We will defend the Wall Street Journal with our lives.

[B]

Randy Jacoby says keep reading fellows. It's quote TBPN in paragraph one. I'm looking at screenshots. I see. No, no, no. Paragraph one, two, three. I don't know, I see. I don't see any.

[C]

You got a different copy of the Journal?

[B]

Yeah. What's going on? Kobe, Brian Jacoby, we're putting you in the truth zone. I don't know, maybe. Maybe they added quotes later or something, I don't know. But the copy that I have does not have it in quotes. So I'm happy and I'm sticking with it. I've screenshotted it and I'm taking a picture.

[C]

Anyways, for those that don't know, I hired Dylan at Party round back in 2021.

[B]

Yep.

[C]

He was my right hand man through the chaos of the ZIRP era. The joy. We did a bunch of work together. That was that prior to TPPN was really the highlight of my career. So very excited to have him on the team behind the scenes helping us make the show better and better every day.

[B]

Yeah, very excited. Anyways, bigger and bigger things.

[C]

You have some news you'd like to talk about.

[B]

The tech news. Anyway, the main news is restream one livestream, 30 plus destinations multistream. Reach your audience wherever they are only possible. This show is made possible because of restream. We are of course built on the back of restream. But the real news is Ajni Mihdha over at Andreessen Horowitz says the Empire Strikes Back and shows a screenshot of worldwide Google Trends results for the past 90 days for Gemini versus ChatGPT. And you can see that Nanobanana launched late August and Gemini has been surging in the worldwide results. And so I dug into this, I have a bunch of thoughts on what's going on with Google, DeepMind and Gemini in particular. And then Ajni actually followed up a week later on September 14th and said the empire goes bananas and shows that now Gemini has crossed ChatGPT in terms of worldwide search results, which is even, you know, I think it's an important metric to be tracking. Gemini's also at the top of the app store charts right now.

[C]

But it's such a funny dynamic. It's like Google runs the largest search engine in the world and then they're trying to get their alternative to search higher on the search rankings than their competitor running the technology that they created.

[B]

Yeah, it is a knockout drag out fight in the battle for the next AI platform. So I went and tried to get the even more up to date data from Google Trends and it's true. Gemini is absolutely surging worldwide wide. And as of today, Gemini is significantly higher than ChatGPT. Now I think some of this is momentum driven. I also needed to check is this related to astrology? Because Gemini of course could be Gemini season. But Gemini season is in May and June, so it doesn't have anything to do with Gemini season as far as I can tell.

[C]

I should ask Gemini what the best. Like if I were running Gemini, the AI application, what would be the best time of the year astrologically to send it on marketing.

[B]

Send it on marketing.

[C]

Yeah. Most likely to experience success.

[B]

Well, so I wanted to understand a little bit more about the search interest and what's actually going on here. So I went to Google Trends and I started to segment by different regions and different time periods. And if you go to the third, third slide, you'll see that in the United States, Gemini is seeing an uptick, but ChatGPT is still dominant. And so you can see. So that is the worldwide data there, Jordi, you'll see. And then the next slide is this is for the United States. And so for the United States, ChatGPT is still dominant. You see this like up and down of, of peaks and troughs. I think those are weekends. And I think that you see a lot more ChatGPT usage during the week and then people kind of chill on the weekends, which is interesting dynamic. There's also school stuff.

[C]

Yeah, that's school factor.

[B]

Back to work factor. Yes. And so.

[C]

I wonder what that data, I mean you can't necessarily get that data for Google search because itself, because people aren't googling Google, they're Not Google searching.

[B]

Google search. Yeah, that's a good point.

[C]

But you have to imagine, like, I probably use Google more on the weekdays.

[B]

Probably. Yeah, I think most people do. And so I wanted to know what was driving the worldwide surge in Gemini traffic Scrolled around to a couple different. A couple different areas. India is absolutely spiking. So if you go to the next slide, you will see right there. Look at that chart, Jordy. So Gemini is just kind of a flat line relative to ChatGPT and then just completely surged just in the last few days. And so I think that that's what's driving the overall. The overall trend in Gemini dominance. Because if you go to. On the next slide, I go to the United Kingdom. And the United Kingdom is similar. Similar to the US Seeing a little bit uptick in Gemini searches, but it's not completely overtaken ChatGPT. So there's a few different narratives here. One could just be Google has stronger penetration in certain international countries and they're pushing different things there with marketing or deals. There's a whole bunch of different dynamics that can drive even just a B tests on how quickly a. Like, how quickly are they routing you to that particular app? Like, you forget that, like, when you have a billion users, billions of users, just setting the default user experience flow can route a ton of traffic to one thing all of a sudden.

[C]

So I got prompted. I did a Google search this morning, and I got prompted to use the AI search overviews. No, no, no, no. I got prompted.

[B]

AI mode. AI mode does not count as Gemini. That's not what we're talking about here, which is interesting, but yes. I mean, it's all. And I'm sure you go through the flow. Yeah.

[C]

So they were counting that as no.

[B]

So AI mode is a. Is. Is within the Google search product Gemin. And to be clear, Gemini is doing really well. So qualitatively, VO3, incredible. I've made a bunch of fun videos, everything from monster truck videos for my son to promo ads for what we do here at tvpn.vo 3, I was not a daily Gemini user. I had not paid Google any money for Gemini's pro plan. I was paying for the ChatGPT. VO3 got me to fork over $250 a month, which I think is probably around 500amonth now, because I didn't cancel. And I'm very happy with that. Like, it's been a good value trade. And I feel like I'm getting access to a frontier model that I can't get anywhere else. And so it's worth every penny. Nano Banana has also produced incredible results and a lot of people are citing the massive growth as driven by Nano Banana. Now, I haven't been. I haven't been going bananas. Have you been going bananas lately? I've tried it a little bit. I just, I don't really have something in my workflow. It feels like nanobanana might almost be more useful if you're doing like professional level work. I've been sort of like interested in using it, but it hasn't become part of my daily workflow.

[C]

I think when we're traveling you have a little less time to make memes.

[B]

And yeah, maybe that's it.

[C]

You probably would have.

[B]

I did find that it was fantastic at that specific thing. You can. And I saw Dylan Patel post this, you can take a meme template and two images and just say nano Banana, like fill it in and it will just do it. It'll just basically do the Photoshop of like or like, you know, the layout to put the images over each other and it just does it flawlessly. And so normally I would be going to.

[C]

That seems to be enough to take Google from 3 to 4 trillion.

[B]

I would agree. Normally I go to something called like image flip and then I have to like type in the text and then put in the images. And it's this web UI that's like decent. It works pretty well. But being able to just do that all in a chat box seems really cool.

[C]

Yep.

[B]

Anyway, quantitatively, Gemini is crushing. It has 483,000 ratings. A little hot on the levels there. I love that. Oh, Bobby in the chat says it's Google's birthday. So happy birthday, Google. We are celebrating you today. Ben, it's like, yeah, it was a little hot. So in the iOS app store, Gemini is now number one on the productivity.

[C]

It celebrates its birthday on September 27, but September 15 was when the google.com domain name was registered. Domains are more important than anything else. So I agree. Today's Google's birthday in my book.

[B]

So to Compare Gemini and ChatGPT, Gemini about half a million ratings. ChatGPT has almost 4 million ratings. So ChatGPT is still the dominant app in the US iOS app store. But Gemini is certainly growing. And there's an open question I'm still noodling on. How much is the Gemini growth driven by Nano Banana versus back to school, back to work versus international efforts and stuff? Tyler, do you have a. Do you have a thesis? Yes. I don't. We don't have your mic Right now. Let's turn up. Can we turn up Tyler's levels? Hello? Hello, Hello. Just yell.

[D]

Okay. Okay.

[C]

Yeah.

[D]

Now. So I'm looking at ChatGPT when 4o Image came out and it was March 25th. If you look at the trends, there's like literally no spike in Google search.

[B]

Traffic for images in ChatGPT.

[D]

For ChatGPT.

[B]

Yeah.

[D]

There's no, like, you can't tell when the image came out, when, like. Very surprising.

[B]

And you're specifically referring to like the Studio Ghibli moment that went super viral.

[C]

Yes.

[B]

Got it. Interesting. Yeah.

[D]

So I like, it makes me think that, like, maybe the Gemini stuff is not actually that related to.

[B]

Yeah. What do you think is driving the massive growth on this particular chart on the Google trends?

[D]

It's probably just that, like, Google is pushing it really hard. I don't think there's really rate limits.

[B]

Yep.

[D]

The models are like, pretty cheap.

[B]

Yeah. Yeah, it might be. It might be free tier in India.

[C]

Pushing hard something somebody account. Hello in the chat says there's an Instagram trend in India with Gemini and girls wearing saris. That could be enough to send it.

[B]

Maybe. Maybe.

[C]

Yeah. There's so many ways that Google can drive adoption of the product. It always felt like they were holding back on, like really opening the floodgate.

[B]

Totally. Yeah. Yeah. And we saw some posts from. From folks who work on the Gemini app saying, like, the GPUs are on fire. The TPUs are on fire this. This weekend. They're trying to keep it up. I've always thought the gap in functionality was pretty narrow between ChatGPT and Gemini. Like, both have fast and free options, both have reasoning models, both have deep, deep research modes, both have image generation. I'm. ChatGPT is still sticky for a lot of people because it's just. It's just default. Like ChatGPT is on my home row on my phone, but Gemini is just one. One app above on the homepage. So I'm kind of going back and forth. But the. But I think. I think the wake up call of like, the empire strikes back is really just like, what does it actually mean to be an empire? And we saw a little bit of information come out from Demis, the founder and now the CEO of DeepMind. He was talking to David Friedberg at the all in Summit and he said something like, they have 5,000 PhDs on staff, which is just an insane amount of PhDs. They're like the DocuSign of hiring PhDs. Need to put them all in a football stadium. Just to talk about the plans.

[C]

So OpenAI has six, around 7,000 employees.

[B]

Yeah. And DeepMind has like it's a similar organization.

[C]

Yeah, 5,000 PhDs.

[B]

Yeah. And I think the reason why you see folks like Buco Capital Guy being so bullish on Google is that they're already vertically integrated, they're already a hyperscaler, so they have their own data centers.

[C]

You just had to look at their, look at all the different parts of their business. The data advantage, the distribution advantage. You know, you can go on and on and on and on.

[B]

Yeah. Ben Thompson had a great post about AI in the big five. So the five big tech companies that are taking AI really seriously for Google, they have the best infrastructure, they have a good model, they have no partners, they have the best data, they have distribution on Android devices. Search, Google cloud platform, the core business chat Chat bots are disruptive to search. This could be bad, but it also could be like, well it's, the whole organization needs to be awake because it's obviously a threat. You can't just sit back and be like, yeah, that'll probably benefit us. Like you have to move and scarcity risk, data feedback loops diminished, disruptive or sustaining. It could be disruptive new business potential in the cloud, says Ben Thompson. So there. So I do think that the Google Bulls might have over rotated a little bit against OpenAI. I saw one quote that accused ChatGPT of quote, losing at least $2 for every $1 it spends, which I just don't think makes sense from like an accounting perspective. You can lose $2 for every $1 you make. But if you spend a dollar as a startup and you lose two, like what, what, what, how does that work? Like can you imagine? Just think about like the actual financials of a business and think about like okay, our expenses were $1 million and our balance sheet went down by $2 million. Like what would have to happen? Like I don't even understand how that's possible. Can you think of a way?

[D]

Doesn't that just mean you're spending 3.

[C]

Million to generate a million dollars?

[B]

No, no, no, no, no. Not, not revenue they're saying so losing $2 for every $1 they spend. So expenses are $1 million, losses are 2 million. Yeah. How, how does that happen? Capex maybe, I don't know.

[C]

You're saying how can your costs be bigger than your revenue?

[B]

No, no, no, no, it's not revenue. How can your costs be bigger than your expenses? Oh like, like this quote says losing $2 for every $1 they spend they spend, not make. I understand. Lose $2 for every $1 you make. That makes sense. You have $3 in expenses. Very easy arithmetic.

[C]

Do you think. Do you that think they just.

[B]

I think it might have just been a typo. Yeah, I don't know. It might have been a typo or maybe somebody that doesn't understand accounting, but it just, like, mathematically, it's very, very confusing to me. And so I think people are sort of, like, running with this a little bit too much. Also, Brian Merchant wrote that OpenAI fails to have a discernible business model, and I think that's just crazy. Like, the business model is extremely discernible. It's freemium. Like, there's a free tier, and then there's a paid tier. Like, you might not think that the business model is super profitable or it's gonna work or they're gonna win, but, like, their business model is not some crazy. Oh, I don't understand how they make money. It's like, it's freemium. There's a free version and there's a paid version, and they make money off the paid version.

[C]

Yeah.

[B]

And they're investing a bunch and they're losing money. And, you know, this happens a lot in companies. They lose money and then they make.

[C]

Money and discovers that a business.

[B]

Yeah. And so. So I think people are, like. Are like, maybe getting over their skis. At least some people are getting over their skis. And like, oh, like, no one even knows how open a.

[C]

Or what their business model is about. Like, paying to, like, train the model. And they're paying all this money. They're spending money to train the model, and then they're still losing money on the model.

[B]

Yeah, this was the Dario thing where he was like, we, you know, we spend $1 billion to train it, and then that billion dollar training run makes 2 billion over the next few years. And so you add all those up. Those are all good. But then you wind up with this, like, endless pit of money in the short, like, in the overall business. But if each business individually was. Was very good and made a lot of sense. And then also, just on the business model side, it's like, it's like they hired Fiji Simo. They're doing ads. Like, semianalysis, broke it all down. They're going to take a cut of every. Of all the commerce that happens in OpenAI. ChatGPT queries.

[C]

That account, that supplement brand that we were talking about Friday, that's just like printing cash because they're ranked highest in ChatGPT. That will not continue forever.

[B]

And so it seems like my takeaway is like Google's executing well, they're seeing a lot of growth on Gemini. They're doing very well there. But it's not like OpenAI doesn't make any sense or ChatGPT is just like, oh, no one knows how it works. It's not this mysterious thing. OpenAI is still bigger on Google trends than Gemini in the US they're still making money from pro plans and plus plans and then they're also going to start taking a cut of commerce that happens on top of ChatGPT queries. So I don't know overall, I mean, it seems that Google is in a great position to stay in the game for the next technological wave. At the very least they'll be able to play in AI duopolies or oligopolies like they do with Android and iOS. They don't capture a ton of the profits there, but they capture a ton of the business value. And also with Google cloud platform, it's an oligopoly between.

[C]

My read on Gemini doing well is not that it's bearish for OpenAI at all. No, it means that it's an extremely important market and there's going to be a number of players that go after it.

[B]

Yep.

[C]

Do I think that there's going to be 10 different consumer AI, you know, multipurpose consumer AI applications that are, that are amazing businesses? No, there's going to be a handful.

[B]

Yep, totally. The. There was another chart that was going very, absolutely vertical and it's the adoption of Chinese LLMs. Did you see this? So we do.

[C]

We got to give them some credit.

[B]

We got to give them some credit. So open open source models from China are set to overtake the US as downloads surge in use across hugging face. The American ecosystem has stalled in growth and the flip is happening right now. And so you can see that hugging face downloads for USA based open source LLM models growing, I mean completely vertically, very exponential growth there. But then China caught up and the flip is happening right now. And it's interesting. I mean there's one take that's like open source AI will kind of cease to matter in the future and it's not going to be that big of a deal. It doesn't really matter.

[C]

And what's the argument there? Because I could see while 10, 20 years ago, companies making the same case around open source software broadly, in the long run open source is not going to matter and yet it's still super.

[B]

Foundational to it is, I think the argument goes something like just the value capture around it. So yes, Linux is dominant and Linux has a massive install base, but companies that have built their business around Linux, like Red Hat. Red Hat is like a unicorn, maybe like worth 10 billion. At various points in time it's been like taken private and taken public and whatnot. But Red hat is like 1/100 the size of Android or Microsoft or Mac or iOS. Right. Like the other even. I mean Mac built on Unix. So there's like a lot of lineage there. But a lot of the, like the closed source operating systems have captured like 99% of the value. And so I think the hot take on open source not being that valuable is just that there's not that much value capture that happens even if like there is pretty wide adoption. I don't know if there's more to it than that. Then there's just like the. Who had that take? That was like the open source. I think this was Tyler Cowan. He said like the open source models, like, well, if they're distilled from American models, even if they're like Chinese, they still believe all the American things and they wind up believing in free speech and believing in American values just because they've been trained on GPT5. Exactly. And trained on the American Internet. But my question is, how seriously is Google taking open source right now? They have the Open Weights Gemma line of open source models. Do you have an idea here, Tyler?

[D]

I mean, yeah, Google just has Gemma. Those are mostly like pretty small models. They're used a lot in like, like mechanistic interpretability stuff. So just like kind of like toy models. Sure, but. And then there's like OpenAI. They have their own. And I mean Meta is really the only like big player and open source in the us.

[B]

Yeah, my, my, my take on this is that it feels like there is a, there is a Crown in Washington D.C. that is waiting to land on some big tech executive's head for just dominating the American open source AI narrative. And obviously that was OpenAI's branding for a long time. Then they went closed source. They have an open source model now, but they're not known as like the leading open source lab. Right. The leading open source lab is arguably Deepseek or Alibaba with Quinn. But you could imagine that you could do a whole victory tour, whole victory lap in D.C. over saying like, hey, I'm the American open source champion. And so I wonder if Google will go for that or if Mark Zuckerberg at Meta will go for that with the next release of Llama and say, hey, I'm still taking this seriously. There's been a lot of rumblings about like, oh, is Meta backing off of open source?

[C]

I think John, it would already leaked. They had a conversation. There were conversations internally, allegedly around, hey, is this actually important or do we just want to deliver value for our own customers?

[B]

Yeah, exactly. So yeah, can they keep justifying the capex and then open sourcing the results and then handing out the freebies?

[C]

Anyway, did we want to read through some of this post from Zyphyr?

[B]

This format has been going like giga viral. Have you noticed that every third post is this?

[C]

It's like E Google in 2017 small team drops quote unquote Attention is all you need. Execs nod politely Go back to selling ads for socks. Let Transformer gather dust for five years like vintage beanie baby be Noam Shazir OG wizard quits builds AI boyfriend app character AI millions of lonely hearts pay $9.99 to eat 2B from near Google HR any chance you want to come back? Shazir busy. My EGF is calling.

[B]

I don't think this is what happened.

[C]

I don't think this is what happened at all. But it's funny. Funny wording history 2022 OpenAI drops ChatGPT World loses its mind Google stock does a bungee jump Sundar holds all hands Guys, we need A response by Q2 team ships Bard in three weeks, using 12 interns, 2000 TPU V4s in pure panic, Bard hallucinates the earth is flat. Did that. Did that actually happen?

[B]

I mean, I don't remember that. I remember there were some. I didn't use Bard.

[C]

I kind of missed the name Bard.

[B]

Bard was kind of a great name, but there were so many. There was Palm and Google was like sort of promoting like their internal research and their external stuff. So they had like product names and then also model names and the model naming convention. It's like way overstated to say that OpenAI is the only one with confusing model name architectures because Google had the exact same thing going on with like, with Palm and Bard and all these different.

[D]

Even now they have. It's like Gemini 2.0 flash light.

[B]

Yeah.

[D]

And then it's like, is Flash better than light? Or like Pro5 is worse than?

[C]

It's. It's a rite of passage to have naming.

[B]

You have to go through it before you actually just release the one good product, which would probably just be google.com. i remember a decade ago, people used to analyze like the updates to the google.com algorithm. There was one, there was one release that was called like Penguin or something. I forget what it was called, but it changed the way search results ranked and it was like disastrous for a bunch of companies that had built SEO around the previous architecture. And then when Google shipped this change, it like decimated a whole bunch of businesses. But it goes back and forth. But anyway, Bard was a good name. Maybe they'll bring it back at some point. But I think Gemini's cooking. I think Gemini is going to be around for a long time and I do think those, I do think they'll go model router and I think that we will not be seeing 0.5s very long. Realistically, all these products should probably go to an annual release cycle and it should just be. Yeah, I'm using Gemini 2025, 2026, 2027. Like a car. Anyway, you want to continue with this?

[C]

Shazir finally answers HR's 47th email. It's funny again, of course this is.

[B]

Not how it works.

[C]

Not, not how it works. Bard. Anyways, fine, but I want my own parking spot and unlimited Lacroix return waves hand Gemini 2 and 2 and a half drop context window so big it can summarize a bible and your group chat Google processing 980 trillion tokens a month equal to 40% of human SMS traffic.

[B]

That's crazy.

[C]

If that's real, that is wild. Someone sneezes Nano Banana into the Play store overnight. It's number one app turns your.

[B]

We're not going to read that last part.

[C]

We're not going to read that last part, folks.

[B]

Go find it yourself.

[C]

But this I asked how would Gemini the app do in September astrologically to Gemini from an astrological perspective, September is a complex and transformative month for Gemini. While it may not have seen the same effortless flow as in Gemini season, it presents a significant period of growth, learning and necessary recalibration. Yeah, it's just like the most classic like astrological, like generalization, like Gemini is generating this and then reading it being like, wow, that's. That's actually very funny.

[B]

Anyway, let me tell you about Figma. Figma.com Think bigger, build faster. Figma helps design and development teams build great products together.

[C]

Great to see John Exley. Back in the chat we were wondering where he was.

[B]

World famous. Now he's been featured in Business Insider. Thank you so much to John Exley.

[C]

TVPN was in Business Insider this morning. John got the quote.

[B]

I think everyone in this chat is.

[C]

Yeah, a lot of folks were featured in there. But Julia Hornstein broke down. She I think explained and described Aura farming and a number of others.

[B]

I think that that was overall great piece. I think she kind of missed the definition of Aura farming. We'll get Tyler to put it in the Truth Zone. That was the one thing where I read I was like, I don't know if that fully fits Tyler's definition of ore farming, but it was funny to see it printed in the this was Business Insider.

[C]

John's best quote was John Exley. That is I look at TBPN as an existence of the proof of heaven.

[B]

Thank you John.

[C]

Existence, proof of heaven.

[B]

Thank you John Exley. We appreciate you.

[C]

We appreciate it.

[B]

Anyway, Ajni Miha has been on absolute terror. He posted that AI is eating software across the handful of frontier AI startups that he's a board member in. Year to date, sales have gone from 1.3 billion to 8 billion. Meanwhile, publics are bleeding the most, the single most rapid reallocation of value from legacy companies to startups in history. And that's why we're partner with so many of the startups that are taking it to the to the public.

[C]

I wonder, I can't see what any of these companies are. So it's hard to give that creative analysis here.

[B]

One of them is Adobe. Adobe is up 33% this year. It's $150 billion company.

[C]

Yeah, I just wonder how much of these are selling off because of the overall narrative that the gen AI, the labs are basically going to be able to just generate these companies from scratch.

[B]

It's that. But it's also that there really is significant pricing pressure. This is what we were talking to Jim Cramer about with regard to Salesforce. Like Salesforce is a seat based model and they need to shift to an outcomes based value based pricing and that is something that is hard to do in the short term. Well, question of are you?

[C]

Here's something. So this morning Trump jumped on Truth Social. He said subject to SEC approval, companies and corporation should no longer be forced to report to report. He put it in quotes on a quarterly basis. He's a quote enthusiast. Quarterly reporting but rather to report on a six month basis. This will save money and allow managers to focus on properly running their companies. Did you ever hear the statement that quote China has 50 to 100 year view on management of a company whereas we run our companies on a quarterly basis? Not good. They still have quarterly reporting requirements, don't they?

[B]

I don't know.

[C]

It's just more of like a. Imagine if companies only had to report. Report every 50 to 100 years, though.

[B]

Could be pretty.

[C]

Imagine waiting. It's like you have your first Google, your first Palantir earnings as a young man. You know, average.

[B]

Maybe earnings.

[C]

I'm riding with Palantir.

[B]

We'll see. We'll see.

[C]

We'll see what you know, right? It's like, it's like you get that one earnings report decides whether you retire.

[B]

Yeah, yeah, yeah. Like at our last earnings, Google made.

[C]

At our last earnings before the Internet.

[B]

Google generated $100 billion in net income last year. Imagine going from like the IPO perspective. Yeah, yeah, we generated 20 million in net income to being like, yep, we're ready to update you. It's been 20 years, $100 billion in net income. We've done very well. It's a beat green candle. Anyway, regardless of what happens with the sec, you got to stay compliant. You got to get on Vanta. Vanta.com, automate compliance, manage risk, improve trust continuously. Vanta's trust management platform takes the manual work out of your security and compliance process and replaces with continuous automation whether you're pursuing your first framework.

[C]

It's so funny to think like, you know, DJT has his public company. He does Truth Social. He's just like, ah, I'm getting really sick of this. What if I, you know, what if I could cut the management teams? You know, sort of. I do, you know, one, you know, I can see there, there being some, if this actually gets rolled out, there could be some bad side effects. Maybe companies get sloppier, right? Like the quarterly cycles, like a forcing function for, for, you know, just making, you know, continuous progress and running a tight ship and, and all these things. But if the burden of being public goes down, maybe more companies go public earlier, right?

[B]

Yeah, maybe, maybe that's.

[C]

And maybe, yeah, CEOs, that would just have. But again, I just, I'm not sure it would be.

[B]

Since we cover earnings, we should be strongly against this because half as much content would be devastating to us.

[C]

I mean, there's, I mean, I don't.

[B]

Know there's plenty of COVID earnings that closely, but it does make for a fun news cycle, getting new data points and new earnings calls. But maybe the podcast world can fill in for that. Do a quarterly podcast tour and chat about your business. Less quiet periods for sure. Right? So be good on that. Anyway, in other news, the Wall street journalist is reporting that China has said that Nvidia violated antitrust law. Regulators move against the chip company comes during Washington Beijing trade talks. We're going to have Bill Bishop join the stream in seven minutes, China said an initial probe found Nvidia violated the country's anti monopoly law, heightening pressure on Washington during the latest round of U.S. china trade talks that ended Monday. Beijing's antitrust regulator cited the violations in connection with Nvidia's acquisition acquisition of an Israeli company that was completed in 2020. The regular. The regulator said the investigation was continuing and didn't elaborate. Beijing approved the deal after Nvidia agreed to conditions including guaranteeing the supply of its chips to China in But since 2020, the US government has blocked Nvidia and other American chip vendors from selling many of their top flight artificial intelligence chips to China. Nvidia said it complied with the law. We will continue to cooperate with all relevant government agencies as they evaluate the impact of export controls. Beijing's move came just hours before Treasury Secretary Scott Besant told reporters in Madrid that US and Chinese negotiators had reached a framework deal on TikTok following two days of trade talks. If you're not familiar or you haven't been following it that closely, the TikTok divestiture has been pushed a number of times. We're just reaching the end of the most recent 90 day hold. Yes, I believe it's Wednesday. We have two days left and so US and Chinese negotiators have been figuring out a framework for that. I believe polymarket has Oracle and Larry Ellison still in the lead. I don't know if you want to pull up the latest data there on what will happen to TikTok, but it is funny that we're hearing so much more about Larry ellison with the OpenAI deal, which we're going to go into.

[C]

Tomorrow, it's still sitting at a 43% chance that a TikTok sale is announced in 2025.

[B]

Okay, yeah, might happen. Of course there's way more to it. What does the sale actually mean? What will. So, oh, the chart is spiking with a sale. So yeah, unclear what exactly. What piece of the business would they sell? There's always been this nervousness around what if they sell the cloud infrastructure contract and the inference and they say, yeah, you guys can run the app, we decide the algorithm. And of course a lot of people who are nervous about the effect that TikTok is having on the American populace think it's actually the training of the model that matters a lot more.

[C]

If we were in a cold war with a country and they owned One of our most popular newspapers said, hey, we don't really want you doing this. And they were like, oh, it's fine. You can deliver the newspaper.

[B]

Yeah, you can take over the printing of it.

[C]

Yeah, you can print it.

[B]

We're still going to decide the words.

[C]

That go in it, but you can print.

[B]

Yeah, we're going to decide what's on the front page.

[C]

We'll let you.

[B]

Yeah, yeah. Anyone can post to this newspaper, but we're going to decide the sorting of the articles, like what makes it to the front page is important and so control over the algorithm should be relevant. And I've talked to some AI researchers who have said that it might be be possible to run like a secondary algorithm on top of what algorithm gets shipped over, essentially fine tuning it as it comes over. So the TikTok algorithm comes over and it's prioritizing one thing and then you run a secondary feed on top. And this happens a lot with different. You might have a secondary algorithm that's running for. Does this violate the terms of service of the app? Is this, is this adult content or something? It might still rank very high, but if it triggers the filter, it gets deprioritized after the fact. So there are different ways to go about it. And certainly not having the data immediately exfiltrated seems like a net win, but it is a complex topic, so we'll continue to follow the story as it develops. The two sides, US and China, are running up against a Wednesday deadline to do a deal to allow the popular video sharing app to continue operating in the United States. And it's crazy how big of a story this was. What, last year, maybe the year before, everyone was beating the drum on banned TikTok and it's kind of just melted in the background. People don't really think about it anymore. It's just kind of like become. It's fallen out of the news cycle anyway. Nvidia has become perhaps the highest profile business. Caught in the crossfire of the trade dispute between the world's two biggest economies. The company sells the most powerful chips. In December, China's antitrust regulator, the State Administration for Market Regulation, opened a probe into Nvidia's $7 billion acquisition of the in Israeli networking gear maker Mellanox Technologies, which I believe was a fantastic acquisition. I think they made a ton of money off that based on what they wound up selling because they bought this company before the AI boom and then network gear became incredibly important as they built out these massive data centers with Nvidia GPUs and so a week earlier, the Biden administration ratcheted up controls on China to China's access to high end chips. Antitrust lawyers familiar with the case said Nvidia was in a tough spot because it had to halt supply of its most advanced chips to China to comply with the US export controls. But that opened it up to criticism from Beijing. We will talk to Bill Bishop, who is in the Restream waiting room about everything that's going on in China with regard to Nvidia and elsewhere. So let's bring in Bill Bishop from the Restream waiting room. How you doing, Bill?

[A]

Good, how are you?

[B]

I'm fantastic. How are you? Have a good weekend?

[A]

I'm good. I had a great weekend. And I have to say thank you, of course. And I have to say Tashi's labubu.

[B]

Yes.

[A]

And I will say espn, you top ticked the labubu bubbles. Since you said you were sent this, I think Popmart stock is down like 20%.

[C]

Let's go.

[A]

I mean we on secondary markets, so it's awesome.

[C]

We did, we did say we were buying it so your dog could rip it apart.

[A]

Yeah, I actually examined it and decided there might be some toxic, toxic pieces. So I'm keeping it on.

[B]

Oh, that is, yeah, yeah.

[C]

That is so smart.

[B]

That's great. Anyway, we were just reading through this, the latest news on the China saying that Nvidia violated antitrust law. I was wondering if you had any thoughts and then we can kind of go into just generally give us the update on what you've been talking about on your fantastic blog cynicism. And then also Sharp China, which is your podcast which everyone should go subscribe to. It's in the shirtechery bundle. It's fantastic. But did you see this news? What was your take on the, on the Nvidia antitrust debate going on in China?

[A]

So it's interesting timing. It relates to this Mellanox acquisition from a few years ago and that Nvidia had agreed to certain terms. I think most were public. I think at least one term was not disclosed. And now, you know, the Smart Samr, which is the anti monopoly group, is saying that they violated one of the, at least one of the provisions. And the timing of course is not coincidental. Right. It dropped just as the US and China negotiators were meeting in Madrid. It comes two days or three days after the announcement of investigation into dumping of analog chips as well as discrimination in the sort of blocking of sales of certain other chips. And so I think this is one, I think Nvidia technically could be in a lot of trouble because if they are found to be violating their agreement, I believe they can be penalized up to 10% of their global revenue in addition, and then have that be multiplied two to five times. So they could be looking at several billions of dollars, several billion dollars of fines from the Chinese government if it proceeds to its conclusion and they're found to be in violation of the agreement. You know, my guess is that this, the timing is, again, not coincidental. And it is, it is part of a way to further encourage Nvidia and the US Government to pull back on export controls. And, you know, clearly Nvidia is already motivated to do that. This may motivate them more. And whether or not then the Chinese government uses this as part of this broader discussion around export controls, that's not clear. Now, in the previous segment, you were talking about TikTok. What was interesting in these discussions was from the statement so far from the US Side, they said that the Chinese tried to bring in export controls or other things, but the US side was solely focused on TikTok. And so, you know, but then they're going to keep talking and they're going to extend the deadline for the imposition of additional tariffs and they're just going to keep negotiating. And, and so I think that this is all part of the positioning on the Chinese side to push back on the US Government. The announcement last Friday of this investigation, the dumping of analog chips, is probably the Chinese getting ahead of what's to be expected of this 232 investigation from the US Trade Representative's office around legacy chips from China. And so there's just a lot of moving pieces. I don't think Nvidia is in any trouble immediately because this investigation, but it just makes their, makes their prospects in China a little murkier. And again, I think one of the, one of the sort of waypoints for investors to look at is assuming President Trump approves this modified Blackwell chip, though I think it's the B30 that Nvidia is trying to create to sell to China. It's much better than the H20. It's not as good as Blackwell, but it's much better than anything that the Chinese can make anytime soon. If the Trump administration says yes, Nvidia will go, will grant you licenses. Does the Chinese side actually start buying or do they do what they did with the H20 and basically say, we're not going to buy, and we still don't know why they don't want to buy the H20s but it certainly doesn't look good that that wasn't a good outcome for Nvidia.

[B]

Yeah, it feels like there's definitely like multi party interests here with the CCP maybe not wanting a bunch of chips to come in from Nvidia because they want to spur their, their domestic homegrown chip manufacturing business. And then, but then the companies, Alibaba and Deep Sea, like these companies just want the best chips to actually compete. And so there's, there's like a tension there. Is that kind of the correct narrative?

[A]

No, that certainly is how I look at it. And I think that, you know, ultimately, and that's why I said, you know, if Trump administration approves the P30 and China still says no, then I think the answer will be that the party, the sort of security forces, the party forces have decided that they are going to just rip off the band Aid and move, try and move faster towards indigenization of chips. If it turns out that actually they start buying B30s in bulk, then maybe the H20 rejection was effectively a ploy. Said no, no, we want the better stuff.

[B]

Right.

[A]

And so they set up this, sort of made it look like there's some confusion or, you know, but ultimately it, it's a benefit to Nvidia. We just don't know at this point. One of the big questions too is, you know, certainly a company that would want Nvidia to, in Chinese companies to not buy Nvidia would be Huawei. Right. Because Huawei has the Ascend series. They're the ones who are claim they have potentially the best potential ultimate competitor to Nvidia chips is whether or not Huawei has actually been over promising what they've been able to make, both in terms of actual capabilities and also in terms of production volumes.

[C]

Do you think?

[A]

So there are some whispers that the Huawei was sort of blowing smoke up towards the top of the system that they're further ahead than they are.

[C]

Do you think there are certain members of the CCP and the party broadly that are extremely AGI pilled and then others that are just thinking like, you know, this is an extending technology for the Internet and it's going to, we're going to develop valuable services and software with it, but we're not building machine God. So it's okay if we get sort of set back, if we set ourselves back, you know, five or so years because it's not necessarily critical to national security.

[A]

That's a great and key question. And I think you look at I think it was last week or the week before the Chinese issued this AI action plan, which really is more about how to bring AI into a whole bunch of different sectors. A big focus on industrial applications. Not at all. AGI pill. It's really basically how to.

[C]

Increase business efficiency.

[A]

Yeah, business efficiency. Increase governing efficiency, make it better for the police. Right. Get AI models for the police to do better predictive policing and better social governance. Improve education, improve the healthcare system. Not at all. They don't need the Ferrari of AI models. They don't need the Ferrari of chips. They need good enough. And the question then is, at least on the open source models in China, like Quinn, like the Deep Seq, are they good enough for what the Chinese need and are the chips that they're now starting to make in more capacity and more volume, are they going to be good enough for what they need? And that ultimately I think is, you know, if the answer is yes, then Nvidia is probably screwed in China. If the answer is no, then they still have a bit of Runway.

[B]

Give us the update on TikTok. It feels like everyone was. There was like an odd coalition on the left and the right. Like last year, everyone was like, Ban TikTok. It felt like it was a fever pitch. It was a rallying cry for tech execs. And it's like everyone had a good reason to want to ban it. Like the, you know, a lot of the tech leaders compete with TikTok, whether YouTube Reels or X or X or shorts on Instagra. And then there were the security and the China hawks. There were so many people that were saying like, yes. And then the narrative kind of turned to like, well, Trump kind of liked TikTok and it helped him get elected and he felt like there was a groundswell of support on TikTok and so maybe he wasn't worried about it. But now we've like kicked it out a couple.

[C]

Yeah. And then the new kind of potential emerged of the cloud contracts, which hands. And that felt like the example. I don't know if you heard, but the example we gave is like if you had a national rival, geopolitical rival and they owned a popular newspaper in your country and you didn't like that, and they said, well, it's okay, we'll let you print the newspaper and we'll let you do like physically distribute it, you'd be like, well, it's not really enough because I want to control, I don't want you controlling what's in the headline and what's on the front page and how the. And it feels like if we end up with a scenario where like, Oracle gets a big, even bigger cloud contract from bytedance and that's the end of it, that doesn't feel like it's resolved kind of the key issues that the real hawks were worried about.

[A]

Well, so a few months ago, I had a conversation with one of the drafters of the law that effectively banned TikTok if they didn't meet certain conditions. And this person just said, you know, the only thing we didn't contemplate was a president who didn't enforce the law. And so what's been happening is the extensions that President Trump has been giving, none of them are legal. Right. That they are completely blatantly in violation of the law. Doesn't matter. Right. And so now, though, the question is, in this particular, you know, today we got the news that they apparently have some framework deal, Trump and Xi Jinping are going to talk about it on Friday, is whether or not the framework deal involves both the ByteDance share going below 20% of NewCo or whatever they're going to call it, and that the Chinese side, the ByteDance, has no input, no connection to the algorithm. And so because the law is very clear, and actually, a few months ago, some of the investors in this mooted sort of consortium that was going to create this new code of buyout TikTok US, they were trying, they were lobbying on Capitol Hill to get a sentence inserted in any bill that would take it that effectively indemnified them against breaking the law, because they understood, you know, they're potentially putting billions of dollars at risk, they are smart enough that they don't want to actually do that in a deal that technically is in violation of a law on the books that survived a Supreme Court challenge. Right. Nine to zero. And so, you know, this is, I think if what we've, and we don't have really any details yet of what the Chinese and US Sides agree to today in Madrid, but if it turns out that the Chinese side, the ByteDance is going to go below 20%, and for example, the U.S. the new code is going to have to rewrite the algorithm, then it probably. And there's no data connection, there's no technical connection to China, it's complete, it's logically physically separated from anything that touches China, then that deal probably works. But then what is TikTok to who's going to write the new algorithm? I mean, last I heard, machine language engineers are, like, expensive and hard to find these days.

[C]

Well, yeah, and we have some of the smartest. Yeah, we have some of the smartest engineers in the world that YouTube and Instagram like, trying to make the algorithm as good as people report. TikTok is right. I think that, yeah.

[B]

How much of a, of an olive branch can be given on the cap table? Because there's a lot of American investors who invested in ByteDance or TikTok have huge stakes. Jeff Yass at Susquehanna is a good example. 15%.

[A]

Bill Ford at General Atlantic.

[B]

Yeah, General Atlantic. And so like there's a, there's a very like neutral outcome here where you say, oh, well, like you have this massive stake, we don't want you to take this huge write down. You can just have an even bigger stake in the US in the US entity. And so that's probably even more valuable. And you get your money out of China so you have more liquidity. You can IPO the company. There's a lot of things that you can do there. And it's not, it's not value destructive.

[A]

One of the things it probably does is it probably unlocks, if this deal gets done, it probably helps if ByteDance wants to, helps them sort of clear the path to an IPO in Hong Kong. Sure.

[B]

Right.

[A]

Which, which, and again, the US of all the investors by tense, I think in the last private round was valued at like 500 billion.

[B]

Right.

[A]

So there's a lot of money at stake here. You mentioned Oracle. You know, to assuage US regulators concerns, Oracle and ByteDance had gotten together. They've done this thing called Project Texas where they were going to segment out all the US user data. They proposed that they would actually review the code, the algorithm code, Even though the algorithm would be updated from China, they would have engineers who could go through the code and make sure it wasn't being manipulated. That didn't satisfy the national security. National security concerns. That's one of the reasons why the law was passed last April, in April 2024. There had been talk earlier this year of a project Texas 2.0 to kind of cut a deal. But ultimately, I think there were enough people in the GOP on the Hill who made it clear to the Trump administration that actually that wasn't going to fly because it still was in violation of the law that passed last April. And so really, I think again, we'll see what has come out of this. What's interesting too is how the US side made this TikTok the focus of the negotiations this time, not soybeans. There's an article in The Washington New York Times today about how soybean farmers are about to go bankrupt because China's not buying soybeans campaigns. That wasn't the focus. The focus was TikTok. And because, you know, think about it, there's only upside for the president, right? And the investors. Because now if, if TikTok is saved. TikTok is beholden to the president. It's another algorithm that probably, you know, has favorable feelings towards President Trump. He's helped a bunch of big investors, including Oracle. It's, it's a, it's an interesting moment when, like you said a year and a half ago, people were going nuts on, across the aisle, across the aisle about how dangerous, dangerous TikTok was. And now it's like, okay, whatever, as long as it abides by the law, whatever.

[B]

Give me the broader update on what's happening in Beijing and Xi Jinping's like, grip on power. There's been some rumors that maybe he's losing control. There's always rumors like, how are things just overall looking there? He was recently caught on a hot mic saying he's going to live to 150, which I think is like a very.

[C]

Watching a little too much. Brian Johnson.

[B]

So I'm wondering.

[A]

Honestly, I think those, I mean those rumors, from what we can see, there's nothing to them.

[B]

Yeah.

[A]

And, and what we saw at the, you know, the big parade on September 3, some other personnel stuff. There's just people. Every, every year there are these rumor cycles. They came a little earlier this year, but every year they turn out to be BS and so far they're looking like bs.

[C]

Oh, well, Trump said this morning that he thinks, that he thinks companies in the US shouldn't have to report quarterly. He referenced Chinese companies thinking on a 50 to 100 year time horizon. Yet Chinese public companies do have to report quarterly. Do you know if, do you know if. I mean, is it common for public company management teams in China to say like, you know, we just missed on everything this quarter. But let's look at this at 100 years. Let's zoom out a little bit.

[A]

Is that, that's one of those true social posts that just. Who knows where it came from?

[C]

Yeah, I was saying it's probably Trump. Trump has a public company, he's like, realizes how, how laborious quarterly reporting is for his team and he wants to free them up to work on product.

[A]

And now the light, the lobbyists for like the accounting firm and the lawyer, the law firms are going to be cycling up in D.C. like, wait a Minute, we make money on a quarterly basis.

[B]

Quarterly reporting is their business. Yeah, there's definitely some people that are indexed to that. We were joking about that. We cover earnings. We want as many earnings reports as possible. We like that.

[C]

Let's do daily. Daily earnings.

[B]

Daily earnings.

[C]

So CEOs can say, you know, take that day's revenue and say, we're at. You multiply it by 365 and that's your run rate.

[A]

Just have a public view of like your 0 or ramp or whatever to do. Your accounting system. System is.

[B]

Yeah, there you go.

[C]

What's the latest? Any updates on the export tip, donation, tax, whatever, whatever it was with Nvidia?

[A]

15%.

[B]

15% for the big man.

[A]

Yeah. No. So I think, but again, I think what, you know, a broader perspective there. Last week, a nominee for an important position at the Department of Commerce's bis, which oversees export controls. His nomination was withdrawn. He was someone who's quite hawkish on high technology exports to China. Right now, Nvidia, the chip firms are, I think, feeling like they're in a pretty good position here in D.C. and there's, I think, going to be a lot more pressure to try and roll back a bunch of these export controls. I think that the actions from the Chinese side, maybe potentially Nvidia today, that announcement, but then the announcements last Friday also may give ammunition here to the folks who are proponents of rolling back those, those chip export controls to further push the White House to say, this is not working, we need to take a different approach. And so that ultimately would be good for a lot of these chip companies. Again, back to the Nvidia conversation. The question is, will the Chinese still buy the better Nvidia chips? And I mean, I think they'd be crazy not to. But again, this is not necessarily an economic or technical decision. Solely, totally.

[C]

What are you. When, when will we have clarity or will we have clarity on the conversations in Madrid?

[A]

So I think the call Friday should, you know, supposedly she and she and Trump are talking Friday. After that, potentially we'll get more, more information. So far there have been, you know, both, both negotiating teams gave statements and there's been a little bit of leaking of the press, but there's nothing like specifically around TikTok. There's just very, very few, if any details other than there's some sort of a framework agreement. And you guys, I mean, you guys have been in the tech world for a long time. What does a framework agreement really mean and how is it legally binding and sort of where, yeah, it sort of Feels like you say framework when you still haven't figured out all the details.

[B]

Yeah, makes sense. Well, we'll be watching.

[C]

Do you have concepts of a plan?

[B]

I don't. Yeah. I don't like these private phone calls between Xi Jinping and Donald Trump. I would prefer a podcast. I want them to sit down together with a subset.

[C]

Live subset.

[B]

Yeah. Yeah. Open invitation to Xi Jinping and Donald Trump. If you want to come hash it out live on tvpn, we're happy to host you. You can discuss the plan, the concepts of a plan, the framework for the plan. We'd be happy to have you and it has been great having you, Bill. Thanks.

[C]

Always a pleasure.

[A]

Thank you.

[C]

Have a great rest of your day, Bill. Cheers.

[A]

You too. Thanks.

[B]

See you. Bye. Really quickly, let me tell you about graphite code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster. And Tim Cook was on the red carpet flogging iPhones like a guy selling knockoff pocket watches from under his jacket. Let's play this video phone 17. Tell me all about it. Oh, my God, I'm so excited. I'm so excited that you brought it up. This is the 17 Pro Max. And in this beautiful cosmic orange, it's a totally new design. It was re engineered from the inside out. It's the most pro phone we've ever designed. And so 16 is pretty good. The 16 is good, but the 17 is a must. I love it. That's why you got to give him a raise. That's.

[C]

Give him a raise.

[B]

Give him a raise. Give him a raise. He's not taking days off. He's not being. He's not underselling.

[C]

The market loves it.

[B]

The market loves it. I love it. He's a. What is this? The Emmys? The Emmys, which I believe is for. What is the Jemmys? It's for TV movies. I think it's an award show. You're in the film award show. I don't know. I don't follow his.

[C]

Good question.

[B]

Oh, yeah. It's tv. Because we're going for a Daytime Emmy. We want. For your consideration, that is Dylan Abrascato's number one KPI. Get us a daytime.

[C]

I love to see Tim out there selling every single person. I love it at a company should be selling.

[B]

Yes.

[C]

And it starts at the top.

[B]

Yeah. And you don't soft play it. You're not like, oh, yeah, you know, it's just another year. No, it's the best iPhone ever. It's the Thinnest, lightest, best, biggest, fastest, awesomest iPhone in history. It's a must, so you gotta have it. It's pretty good. The last one, pretty good. But this one must. It's good. But I mean, he is right. It is re engineered. There's that whole. You've heard about the vapor chamber thing. A single drop of water inside the iPhone. Did you hear this? Oh, it's amazing. So inside the iPhone there's a single drop of water in a chamber and it heats up, evaporates and moves to a different part, away from the hot internals to cool the phone. And this has been one of the complaints for the last few years. I haven't had any problems with heat on this phone on the 16, but that's cool. But yeah, my phone overheats.

[C]

Overheats all the time.

[B]

Is it. Do you. Do you find that it overheats when you're like using software on the phone or camera. Oh, filming or video.

[C]

Photos.

[B]

Photos. Just taking photos and you. You notice it overheat. Wow, that's wild.

[C]

It doesn't turn off.

[B]

Yeah, yeah, definitely Just heats up a little bit and it feels hot and. Well, that should be a thing of the past with your new iPhone. 17 Pro Max in orange.

[C]

I'm excited to.

[B]

Do you think that the orange color is a nod to YC or Stratecheri? What were they going. Or bitcoin? Which one do you think they were going for?

[C]

Probably nod to Ben Thompson over at.

[B]

Probably Ben Thompson.

[C]

So a hat tip.

[B]

Yeah. Ben Thompson's been covering the stock for what, almost two decades. He's done a lot of great reporting on Apple and why not give him a little nod with the Stratechary orange iPhone.

[C]

I like it.

[B]

Or the McLaren orange. What else is orange? I don't know. There's a lot.

[C]

Anyway, Cote says I propose a technology brother pilgrimage to Starbase to look at rockets, feel better about the world and maybe throw a giant warehouse party or something. Maybe go pull this off.

[B]

He runs a hard tech conference in San Francisco. I didn't have a chance to go last year, but they. Wait, was it. It was on like a, like a. On an aircraft carrier, something. He really pulled out all the tech bro. Little touches. I believe there were cyber trucks on top of a aircraft carrier and there were sponsors and stuff and everything.

[C]

I would like to have a house party in a nuclear submarine.

[B]

House party. Some flip cup or some beer pong. Yeah, in the nuclear sub. The nuclear sub. It's pretty tight quarters. Have you ever been on a Navy ship?

[C]

No, but you know, it could be a really. You light it properly, it could be a really fun vibe, you know.

[B]

But going to Starbase makes a lot of sense, I'm pretty sure.

[C]

I mean imagine you can just. 24 hour trip Starbase in the morning.

[B]

Yeah.

[C]

Abilene, Texas in the afternoon.

[B]

Oh yeah.

[C]

Stargate or Stargate House within a Crusoe data center. Yeah, something there.

[B]

It'll be fun. Well, get on. Julius, what analysis do you want to run? Chat with your data and get expert level insights in seconds. Ask Julius to analyze your data. Over 2 million users trusted by individuals.

[C]

It's got to be 3 million.

[B]

Yeah, we need to update that for sure.

[C]

Anyway, Rahul posted yesterday. This is a great post on a Sunday. Bad day to be a manual workflow.

[B]

I don't even know what inspired that. I guess he was just automating things.

[C]

It's a Sunday. He's grinding, he's automating workflows.

[B]

That's great. What more do you need? Base is beginning to explore a network token. It we're in the early phases. This is from the official Base account. That's Coinbase's new product. We're in the early phases of exploration and don't have any specifics to share around timing, design or governance. We're committed to bridging the community, bringing the community along with us and building in the open Brian. Brian Armstrong, CEO and founder of Coinbase says we're exploring a base network token. It would be it could be a great tool for accelerating decentralization.

[C]

Expanding creator and developer growth in the base is interesting. It's a blockchain without a native token.

[B]

Yes, and we asked Brian Armstrong about that at the BASE launch event and he said no comments right now, no news to share. But that was a question that everyone was asking us to ask because it did seem like there might be a logical conclusion of this project is to eventually decentralize the project so there's no definitive plans. Stay safe out there. Don't go buy the wrong base token. Getting get too excited. But interesting development from Brian Armstrong over at Coinbase. So congrats to them on the progress and congrats to fall. Generative media platform for developers. The world's best generative image, video and audio models all in one place. Develop and fine tune models. Go with serverless GPUs, not demand clusters. Bill Gates. Someone resurfaced Bill Gates resume as a freshman at Harvard. The resume and experience is still probably better than 99% of college students in tech out There. And so it says his objective was to become a systems analyst and systems programmer. On his resume, he puts his Height and weight 510, 130. What a hilarious thing to put on your resume.

[C]

Interesting. Is that to make your. Is that to make you more interesting from a health insurance standpoint is like, hey, my carrying cost is not going to be that high.

[B]

Yeah, free lunches. I'm not going to be doing double protein. I'm just 100. I'm just a buck 30 when I'm wet and wearing boots. He says no dependents. He wants a $12,000 salary, but he's open to whatever and he's open to relocating.

[C]

Needs at least 12 grand. He's open to 120 billion.

[B]

Look at this resume. His experience, he knows Fortran, he knows Cobol, he knows Basic, he knows Machine languages for most of the above computers. List.

[C]

He even built a payroll product.

[B]

Oh yeah. Traffic flow analysis system, school scheduling, payroll. He's a systems programmer and in partnership with Paul Allen, of course the co founder, designed and put together a system for traffic engineers to study traffic flow. The system is built around Intel's MCS8008 microcomputer. The software and hardware setup has been fully tested using a prototype. Demonstrations to customers are planned for May 1974. Wow. Information Sciences. Fun. I wonder how that surfaced. Like, when did he publish this? Did someone else, did someone else publish this? Leak this? I don't know.

[C]

It's interesting. He was like working with a recruiting services firm. Probably a good client for them. Seems like he would have had a lot of options.

[B]

Indeed.

[C]

Junk bond analyst. We already covered this screenshot of the truth social post about subject to SEC approval.

[B]

Yep.

[C]

This seems. But the sec. Yeah, I mean we're gonna need to see the art. Like why report? It's funny to say, like put it in quotes because like it very much is definitely a report.

[B]

It's a real reporting on a quarterly basis and then also in parentheses. So subject to SEC approval, companies and corporations should no longer be forced to report on a quarterly basis. Parentheses, quarterly reporting, exclamation point. Like why are you putting this in parentheses but rather report on a six month, six and then six in parentheses month basis. In case you couldn't read S I X. I really don't understand the texture of this. This is definitely not LLM generated.

[C]

This is the desk of the president. It ends up being like the exact same amount of work. Right. Like the report. It's the six month report. Just like doubles, you know, doubles in terms of how Intensive. It is. And the function is that the lawyers and the accountants get paid. Exactly.

[B]

This is a wild post. Did you ever hear the statement that quote, China has a 50 to 100 year view on management of a company, whereas we run our companies on a quarterly basis. Question mark, question mark, question mark, not good. Exclamation mark, exclamation mark, exclamation mark. And it doesn't close the quote. Like there's no, there's no end to that quote. And also, since when have we been looking to China for best practices on anything? I thought, I thought we were like near peer adversaries. Very, very odd. Anyway, Elon Musk has purged over 2.56 million shares of Tesla worth $1 billion. The purchase happened on September 12th with an average price of $389.22. This is Elon's first open market stock purchase since February of 2020.

[C]

He is squarely in the 1:1 trillion club. Tesla, the market cap is currently 1.28 trillion.

[B]

Wow.

[C]

It's crazy that only a little while ago it feels like Meta was at like one.

[B]

Yeah.

[C]

One like you know, just, just north of that. But. But yeah, I feel like there's, there's going to be news here, especially around Robo Taxi. Yeah. There was also news Xai terminated like 500 employees over the weekend.

[B]

Yeah.

[C]

Which again I don't want to read too much.

[B]

Put on the 104 hub, bro, go crazy. Give me the theory. He buys the stock, I think stock goes up and then use that to merge in Xia. That would be the 3D chess. If something crazy happens like that. Yeah. I mean obviously Tesla fantastic AI team built self driving like one of the best, if not the best self driving.

[C]

The employees that were laid off were. We're doing like data labeling.

[B]

Right.

[C]

Which, which can probably non equity comped.

[B]

Yeah, yeah, yeah, yeah. And increasingly there are tons. Yeah. It's interesting because there's been tons of competition in that space between scale. Mercur handshake. What's that other one? It's not Volt. It's like Surge.

[C]

So there's a, there's a few Turing Mercour. Surge.

[B]

So there's a few companies that are outside contractors.

[C]

Label box.

[B]

Label box. Yeah, I just heard a label box add on Dwarkesh. It's fantastic. So this is a service that you can very much buy from a highly competitive market and effectively outsource that role. A few companies have brought it in house. It seems like XAI is going from in house to outsourced but I think not too much to read into. On the XAI side, there was a post in here. Maybe we can pull it up. But let me see. It's deeper, deeper, deeper. Yes. So another one of those posts by Zephyr. Zephyr's really been on a tear. 1.4 million views on this post. Be Elon Co found OpenAI for Humanity. Watch Altman flip it into Microsoft's loot box quietly exe launch XAI poach the best GPU whiskers from DeepMind.

[C]

OpenAI Yes. Meta bottomed at 1.23 trillion this year.

[B]

This year.

[C]

And now this year. Okay, and now Zuck is. Sorry, sorry. Elon is like, is north of that today.

[B]

Okay.

[C]

Says, yeah. Considering the revenue and earnings difference between those two companies, it says a lot about Elon's ability to make people believe and give them optionality on so many different tech trends. Yeah, I mean he's making Transformers, they're making chips.

[B]

Yeah, it's a very, I feel like the market still very much believes that self driving cars will be a thing and it'll be big and Tesla's leading in that market and or, you know, at the top of it. And same thing with robotics, with robots. Like, you know, we grew up watching sci fi films or some of us did. I know if you did. But watching sci fi films of robots, if robots are a thing, Elon is going to figure out a way to make money on it. So that's always been the story in recent history of Tesla's bull case and certainly Elon's betting on that. But the bull, the bull case for Xai. This is the data that came out from OpenRouter which I was talking to Tyler about a little bit over the weekend. So Grok code fast one is now the number one ranked model on OpenRouter with 1.06 trillion tokens generated. Claude Sonnet four is at 343 billion tokens. GPT five is down at 72 billion tokens. And I sent this to Tyler and was like, I don't believe that Grok is 4 times bigger than Claude and 10 times bigger than GPT5. Like this just doesn't seem right. And Tyler kind of broke it down for me. What's actually happening on open router. So can you explain when people use open router, what type of data would be included in this particular data set versus the broader LLM inference market?

[D]

Yeah. So is my mic on?

[B]

Yeah, you're good.

[D]

My thesis is basically like you use OpenRider when you're doing kind of low intelligence but like high volume things when price matters a lot. Because if you look at the top models, a lot of them there is like Claude, there is Gemini, but then you see a lot of open source models which are like super cheap to inference. So at least for me, whenever I use like I do a bunch of inference on different models, whenever I use like OpenAI or anthropic or Gemini, I'm usually hitting their like actual API. I'm not using open router directly. Yeah, because like maybe there's some sense where like there's different params that aren't actually exposed via the open router API or something like that. I think that's probably generally true for a lot of people. But if I'm using, if I just need like super cheap intelligence then I just go on open router. I see. Okay, this is like a pretty good model. It's like super cheap. I'll just hit that. And then sometimes if the inference is like super slow, I can really easily switch to a different open source model.

[B]

Sure.

[D]

So that's kind of my thesis. So it's also why if you look at the kind of total tokens inferenced it's like okay, Cloud Sonnet 4 is at 500 billion of the past week which if you kind of map out what the actual revenue is from that. Yeah, it's like a super tiny amount of what their revenue like is probably. If you kind of like think there's some numbers leaked.

[B]

The leaked numbers for Anthropic are it's around $4 billion a year in business, probably more now. So you work backwards and you get to like they should be making like 250 million a month and, and this data only accounts for something like $12 million of that 250. So the actual frame is like 95% of their business or something like that. 90% of their business is happening directly on their API and then 5% of their business is happening in open router.

[C]

Yeah.

[D]

Or something like that. I didn't include the input tokens which is about, it's like 10x cheaper. So it's like at least in that order of magnitude, but it's significantly less than what it should be. Where I've used GROK a fair amount via API. I've never used it through Groq's API. I've only used it through open router.

[B]

Got it.

[D]

So I think that's probably true. Also one of the reasons you might see GROK code fast one being super high is that at least since its release, I think it's still free in cursor and Klein. And I don't know if they're using open router. They're using some kind of router in the back.

[B]

Right.

[D]

Because you can.

[B]

Cursor is.

[D]

Yeah, obviously. Because, because you can select like Grok or Anthropic or.

[B]

Well, isn't that just their own router?

[D]

It could be okay, but they could.

[B]

Be sitting on top of them.

[D]

Yeah, I mean it's like very easy to use it. So yeah, those are some of the reasons why you might see.

[B]

Yeah. So I think performing here, I think this, this post, this is, is sort of like, you know, meme incendiary, written for virality. The default story it's telling is like, oh yeah, like Grok is, you know, 10 times bigger than OpenAI. And that's probably not right.

[C]

But charts are great for marketing.

[B]

Yes.

[C]

And whether you're hitting number one globally in the app Store here, where you're charting matters a lot. Right. And. And it's natural for companies to share the charts that are good.

[B]

When they're good.

[C]

When they're good.

[B]

Exactly.

[C]

And they're just snapshots too. And charts. Yeah, charts are.

[B]

And so what is this actually? What is the actual interpretation here? Like they're winning on open router, which means that Grok has successfully delivered a level of intelligence that is affordable. Yeah.

[D]

I would say like, I would almost put this model, comp it to just open source models like Coin or something rather than kind of the Frontier reasoning. Frontier reasoning models.

[B]

Got it.

[D]

So you could say they're winning kind of in that field. Even though it's not open source, it's still in that similar price range.

[B]

There does seem to be value in soaking up some of the demand. That's further down the cost curve for maybe more price sensitive users. Yeah.

[D]

It could also be like, I mean the model is still good enough where like even if it's. If it's worse on a single shot compared to like cheap D5, if you just run it 10 times, it still will be a little bit cheaper and you're like kind of at a similar spot. So where you can just kind of max out on volume instead of like your intelligence per token is less, but if you just run way more tokens then you're at similar intelligence, it'll still be cheaper.

[B]

Yeah.

[D]

So maybe there's some kind of factor there.

[B]

So like lightly bullish for the XAI team. They've clearly built something that people want at a specific price. Point.

[D]

It's definitely good.

[C]

Yeah.

[D]

I mean, I haven't seen like, I don't really use like the newest Groq model, like their frontier 1. Is it 3? Yes, 3.

[B]

3.

[D]

I almost never use it.

[B]

Yeah.

[D]

So this is maybe like one of the first examples of Grok models, like really being used a lot.

[B]

Yeah. It'll be interesting to see what happens with the company. Yeah. The roll up theory is kind of interesting to toy with. There's a bunch of Imagine Axe the Everything app integrated with Tesla. That's a funny, funny outcome.

[C]

If that would happen, I would be basically forced to buy a Tesla.

[B]

I would definitely buy a Tesla. I mean, buy one and then take it to the shop and manual swap it and naturally aspirated V12 in there.

[C]

But what if they just came out and said, you, you can't use X when you're moving fast unless you're inside of a Tesla. Force people to transition out of traditional cars. That's the next catalyst for the next 5 million.

[B]

I really hope Tesla ships the Roadster. I saw someone on the timeline maybe a week or two ago talking very, very sad, like a super Tesla. Bull, Owns multiple Teslas, owns the stock, saying that they canceled their Roadster 2 reservation after maybe eight years of waiting, something like that, and didn't have to.

[C]

Put down like 100 bucks.

[B]

No, no, no, no. The Roadster was like, was like over ten grand.

[C]

So real deposit.

[B]

But also you hear this crazy stat where people are like, well, if you'd put it in Tesla stock instead, it would be worth 5 million. And it's like, well, that's not really like a fair comparison. The Cybertruck reservation was like 100 bucks, but the Roadster 2 was. I mean, it's. You know, I think people knew that it was going to be several hundred thousand dollars supercar territory, but it just hasn't moved. But I think that they should do it. I think it'd be good. They need a halo car. The Cybertruck's kind of working as a halo car, but the design language is a little bit too disparate from the core portfolio, where if you see, if you see an Audi R8, you're like, oh, Audi's cool. Maybe I'll get an A4. Right. Like, because there's some styling cues that are the same, but if you walk into the Tesla dealership and you see a Cybertruck, you're not like, yeah, Model three really, like, speaks that language. Right. It's like they're wildly different designs. So maybe if over time the three and the S and the X and the Y, they all update to have a little bit more of that cybertruck language. The cybertruck could work as a halo car, but. But right now it feels like a very different car from like a very different brand world, essentially. It's cool, but I want to see. In other Elon news, Perfrock 4 is their smartest. The Boring Company posts the smartest, fastest, safest tunnel boring machine in the next iterations. 5, 6 and 7 are already being built at the Boring Company factory in Texas. Goal is continuous mining and 0 people in tunnels. Zone pit. And they're. They're hiring exceptional mechanical engineers, electrical engineers, and software engineers.

[C]

The Boring Company has just not gotten really any attention for the last few years.

[B]

And we've been grinding.

[C]

It's so funny to think about a world where you're. You look up and you see, you know, starships being launched. And if you could look down, there'd just be, you know, tunnel boring machines everywhere. But. But this one's a sleeper. I hope to see more out of them.

[B]

Yeah, I mean, it's tough. It's heavily regulated to build stuff underground.

[C]

Not going deep enough takes a long time. But deep enough, the regulators won't know.

[B]

Yeah, I heard one interesting thesis that was like. It's just like the underground domain is like, it is a resource. It's essentially real estate. If you can build stuff down there, even if it's just a tunnel, and that's extremely valuable because it's. It's effectively free. Like, it's a free resource. Yeah, just like space.

[C]

I'm not gonna, I'm not gonna pretend like you're supposed to know this, but can you try to figure out like, who regulates once you get. Because I would assume like 100ft below your property is like, probably like there's some traditional property law.

[B]

Yeah, I would think that there's nothing.

[C]

If I could dig a tunnel down there, I would just assume that I have some. Some I can imagine what happens there. But I think if you go deep enough, I mean, is that not like the play?

[B]

What would you be okay with if. If some random person was like, jordy, hey, I got a. I'm. I'm. I'm building a tunnel under your house. Don't worry, it's 500ft. Would you be like, yeah, cool. Or would you be like, I would prefer not.

[C]

500. 500 is fair game.

[B]

500.

[C]

As long as. If they can prove out safety. Right. I don't. I don't want it Any type of collapse.

[B]

Okay.

[C]

Like the level my home is at.

[B]

Yeah.

[C]

And the old in the ultra dome.

[B]

Yep.

[C]

I don't want to suddenly feel a jolt fall through the earth.

[B]

Yeah, Yeah. I don't know. I have yet to tour the. The. The. The. The boring company tunnel. They built one in Las Vegas as kind of a demo. There's been a few that have. That have built. But, I mean, this is one of those businesses that I think has just been building. Building similar to neuralink. Like working, chopping wood, grinding stone in the earth, and. Yeah, maybe something happens here. Tyler, did you get any closer to an answer on how unregulated the underground domain the subterranean domain is?

[D]

So this is ChatGPT. It says courts generally say that you can go. You own the ground beneath your land only to the extent that you can reasonably use it.

[C]

There you go.

[D]

So there was a court case in 1946. The Supreme Court rejected the to the center of the earth doctrine. So you don't just own everything below you. So usually it's like maybe a thousand feet is kind of the extent, and then below that is just. I guess it's just free. Free reign.

[C]

Do you know you're not using. You're not using more than a 10ft below your. I'm sorry, Tyler. The second you have property, I'm burrowing a tunnel underneath it. I'm gonna be hanging out under there, and you're gonna have to prove that you can get down there.

[B]

I do think that there's a massive market for bat caves. You know, it's very hard to do ADUs. ADUs are booming now in California, but it's still hard to actually go and add real estate to a house. We have a housing crisis. Why not just have the. The tunnel company come build and just be like, hey, we. We built you the perfect man cave underground, your house, under your house. If you want us to build a little tunnel up to your. Your bedroom that you can climb down and go enjoy your new movie theater.

[C]

Can you imagine being like a angry teenage getting in a fight with your parents about, I'm going to the main cave. I don't want to. I'm going to the bat dinner. And then you just go in the tube and you go.

[B]

And then you're down in the Batcave. Huge, huge, untapped market. If they can figure it out, they.

[C]

Don'T understand me above.

[B]

Do you know the story of the subway that goes out to Harvard University in Cambridge?

[C]

I do not.

[B]

This is fascinating. So the mta, the Massachusetts MBTA Massachusetts Bay Transportation Authority built a subway that goes from Boston out to Cambridge. And when they built it, they wanted it to go through Harvard, Harvard's territory, Harvard's land. And it would have been a straight shot to get from Boston to Harvard. But when you're on the actual subway, it slows to a crawl at one point and starts. The wheels screech and go as you're going around this weird curve. And I was very unclear on why the Metro, why the subway needed to take a curve when you can clearly draw a straight line between the logical places to go from Boston to Cambridge, to get off and go to Harvard University. And the reason is that Harvard does own the ground underneath. And the reason that they. They did a deal. And do you know who they did a deal with?

[C]

Probably some baron.

[B]

Close. He wasn't a king. It was our first president, George Washington.

[C]

Wow.

[B]

So they got in a. They got in a legal battle. And the Harvard administration fights the Boston Transportation Authority and the government, the Boston Transport and the Boston. Yeah. The Boston Transportation Authority says, we have eminent domain. We. We own the land. We govern this land. You need to let us build this subway in a straight line. It'll be better for everyone. Harvard says no. And we have proof that we actually control the land all the way down to the bedrock. And our evidence is a letter that is signed by George Washington, who said, I grant you this land. And it held up. And so the subway had to go around mogged and honestly, super inconvenient to Harvard students. And I don't know why Harvard wasn't just like, yeah, obviously build it in a convenient way. But there are probably other considerations. Anyway, turbo puffer search, every byte, serverless vector and full text search built from first principles on object storage. Fast 10x cheaper and extremely scalable. Happy birthday to Casey Hanmer, friend of the show. It's been 15 years in this incredible country. He says it's his American birthday on September 14th. Fifteen years ago, 2010, he became a US citizen. He says he feels like he's still getting started. Incredible addition to the squad. Thank you to Casey Hanmer for all that you do for America. Let's play that American eagle. Sound happy, brother.

[C]

Casey.

[B]

Casey. In other news.

[C]

To 150 more.

[B]

Yeah, to 150 more. In other news, debate continues to rage between Salesforce CEO Mark Beniot.

[C]

This account we're featuring here, just calling balls and strikes. Yeah. This is media.

[B]

This is journalism.

[C]

Not conflicted at all. So Palantir hotline with the target says CEOs will do the math. And will and would rather pay Palantir 10 million to make save 100 million instead of pay CRM 5 million to break even or worse value creators greater than value extractors. So not conflicted at all.

[B]

Not conflicted at all. And this is a post from John, who is excellent at clipping our Alex Karp interview.

[C]

This is such a great quote.

[B]

Yes.

[C]

He says, I'm so inspired. So I listen to Benioff tried talking smack with Kramer and saying we're winning deals. Karp fired back on tvpn and now Benioff is taking the high road. He says, I'm so inspired by that company. Not just because they have 100x multiple on revenue, which I would love to have. Are you pulling up the video?

[B]

I'm pulling up the video.

[C]

Pull it up.

[B]

Let's go. It's in the chat now. They're going back and forth. I mean, there is a real discussion. We talked about it earlier in the show about how Salesforce positions its revenue model, how the market will treat that in the short term. Benioff and Jim Cramer had a back and forth on Mad Money talking about how Salesforce, you know, actually makes money and charges the customer. And yeah, there is a question about where both these companies go. It's a big battle. Anyway, let's pull up the clip of Marc Benioff talking to CNBC about Palantir. An idea out there. And you see it in the valuation of the company. Palantir, this idea that that's the future of software. It's disrupting software, and it's doing so without a marketing team.

[E]

Oh, my gosh.

[B]

I am so inspired by that company. I mean, not just because they have 100 times, you know, multiple revenue, which.

[C]

I would love to have that too, maybe.

[B]

Well on their revenue soon, but the prices that they charge to their customers. I mean, there. It's got to be the most expensive enterprise software I've ever seen. I've been, like, looking at their demos this week, saying, my gosh, how do.

[F]

They get these prices?

[B]

Because I thought I had analytics, I thought I had data management. I maybe I'm not charging enough.

[C]

And you're competing with them.

[B]

We compete with them. In fact, we just won a huge.

[F]

Deal at the US army against them.

[C]

The positioning is insane. Insane. Benioff's a legend.

[B]

Yep.

[C]

We're hoping to have him on the show next month. We're working on it, but that is insane levels of Coke.

[B]

You think so?

[C]

Yeah, I mean, I think this is guy notorious for selling expensive Software Salesforce is angry that someone else is more expensive software.

[B]

That's a good point. Yeah.

[C]

When. Yeah. On a long enough time horizon. This is about the value that you deliver to customers. You need to deliver more value than you capture. And Palantir's got a fantastic pitch. They have fantastic. I mean, we were at aipcon. We talked on air and off air with people that are using Palantir Stack. And it's not. I think they obviously compete in a real way, but there's also plenty of companies on Palantir Stack that are not even in conversations with Salesforce around a bunch of different use cases.

[B]

Yeah, I was about to say that it does feel like Palantir's position themselves completely differently from a product perspective in terms of what they do. People think sales. I mean, the ticker is CRM. And so when I hear about like an army deal, I'm like, are they. Are they getting a CRM? Like, but. But I guess that they're just so big at this point.

[C]

So you can do a lot of different stuff. The company C3.

[B]

Yeah, C3. AI.

[C]

They have the ticker. AI.

[B]

AI.

[C]

I was gonna say Marc Benioff should buy them just to get the AI ticker to come into the next era.

[B]

Yeah, just rebrand your ticker. There's been a lot of ticker fu people. I mean, Mark Zuckerberg with the meta rebrand. Getting your core brand on the theme or defining the theme is somewhat of an underrated strategy. It's like a good dot com, a good ticker that people can just go and say, I want, I want exposure to AI. I'm getting C3. I. I want exposure to CRMs. I'm getting $sign CRM. That's.

[C]

That's a potentially new fund with the strategy of just investing based on the narrative surrounding the narrative that you construct around the ticker. Yes, I need an AI. I need an AI company. Maybe, Maybe I need a CRM company.

[B]

Well, whatever you're building, you need to get your brand mentioned on Chatbot. You need to go to Profound reach. Millions of consumers who are using AI to discover new products and brands. Should we go through the Wall Street Journal article? We were featured in the Wall Street Journal. If you're just tuning in, Dylan Abrascato says, on X, I'm joining TVPN as president. John Jordy and team have built something incredibly special. And after starting my career at cnbc, this feels full circle.

[C]

Dylan also ran partnerships for HQ Trivia during their heyday, which is a crazy, crazy Journey. He'll have to jump on the show sometime to talk about. Dylan also knows about sports, which is.

[B]

Oh, yeah, this is a major unlock for us now.

[C]

Have expertise on sports. He's also very into collectibles, so he's helping us build out.

[B]

We should sit down with him and ask him, like, what are all the sports? Like what. How do they work?

[C]

Like, what are the sports? What are the leagues?

[B]

Yeah, what are the leagues?

[C]

What are the plays?

[B]

Yeah, the plays. I would love for him to name every play, get up to speed on the plays. That would be very cool.

[C]

I don't know that we have to go through this. Maybe we can just. Maybe we can pull out some of. Some of the key. Some of the key quotes.

[B]

Yeah. What else is here? They call this podcast, but they didn't put us in quotes.

[C]

At least not total, total victory.

[B]

Big, big shout out to the.

[C]

Total victory to the Wall Street Journal. I, I wanted to highlight one quote in here. Let me find it. I. This, this was my favorite part.

[B]

Yes.

[C]

And I hope that that doesn't sound important.

[B]

Who said this quote? That's your favorite quote that I have.

[C]

Said when talking with Katie, the journalist over. I said, well, we said, this is Katie's reporting. She said the duo has a laundry list of moves they have zero interest in making. Forming a content studio, creating a network of shows, pivoting to political coverage, raising capital, setting up a venture fund and selling. And I said, we're perfectly fine with there being somewhat of a ceiling on the size of the business. Hayes said, we're not building for an exit. We want to retire on the mics, on the air.

[B]

I love it.

[C]

So I told Katie that, you know, some at some point, decades into the future, there will be. There will be a final show. I hope it. I hope it is many decades from now, but we hope to retire here on the mics. I hope to see John. John Exley's name in the chat for one final salute. But anyways, absolutely love the Journal and fun to be in there for the first time.

[B]

Very fun.

[C]

Did you see Mike Tyson drop Mr. Beast?

[B]

I did. I just dropped in the timeline.

[C]

Let's pull this up.

[B]

Mike Isaac said TYSON Nearly Houdini'd Mr. Beast on camera. Are you familiar with this story?

[C]

Didn't Houdini. Didn't he. Houdini himself? I don't actually know what is being Houdini, Tyler.

[B]

What happened with Houdini?

[D]

I mean, this is like from when I was in, like, fourth grade, but I was told that basically he got Punched in the stomach just like this. But he didn't clench or anything.

[C]

He prepared for it. So he had a. I remember now. So he had a whole bit of letting any, any guy could just punch him.

[B]

Whoa.

[C]

And if you, if you flex in the right way, pretty much take a punch. You can take a punch, sure. But a guy I guess caught him when he wasn't, like, ready for it.

[B]

It just broke a rib or something. Internal bleeding and Houdini passed away. Rip to a real one.

[C]

Let's pull this video.

[B]

Let's play the video of Mike Tyson.

[A]

Punch for Mike Tyson.

[C]

Whenever you're ready. Straight to the liver.

[B]

Oh, that is intense.

[C]

Whenever you're ready.

[B]

He really. Whenever you're ready. Okay, okay, that's enough. That's enough of that.

[C]

Is that. Do you think, do you think he was doing a bit of acting there because it was such a viral moment?

[B]

I don't know.

[C]

It wasn't like a liver. I don't think it was a liver shot.

[B]

Yeah.

[D]

There's like a longer version I saw. And he's like, still down there for like another 10 seconds.

[C]

Yeah, yeah, that's. That's just.

[B]

Yeah, but Izzy just acting.

[C]

He's an entertainer.

[B]

He's an entertainer. I think he knows. I think he knows to entertain, but at the same time, I think he knows that it is like he can talk to Tyson and say, hit me hard, but don't actually take me out entirely. And Tyson can dial it into a point where Tyson knows. Yeah, I'm not going to break your rib, but I will knock the wind out of you.

[C]

The thing seeing the, the, the Tyson Jake Paul fight.

[B]

Yep.

[C]

Is watching Tyson all these years later with just like this incredible form and like movement despite being like long retired was. Yeah, it was cool. It was cool to see. It was literally like riding a bike for him. I pulled up. We were talking about tickers earlier. People that wanted to. To invest in CRM could buy hashtag CRM or you could buy C3 by.

[B]

Buying AI on public.com investing. For those that take it seriously, of course they have multi asset industry leading yields.

[C]

Trusted by millions, trusted by billions. Soon. But I pulled up some tickers here and I wanted to. I wanted to help you build a portfolio.

[B]

Oh, you have, you have more funny tickers.

[C]

Okay, start. We can do ticker cake. Cake for the Cheesecake Factory.

[B]

No way. That's good.

[C]

It's hard not to go Taco Taco. This was Del Taco. It used to trade under that before they got acquired. Bud Anheuser Busch.

[B]

Okay. Yeah.

[C]

Boston Beer company traded under Sam.

[B]

I'm still Samuel Adam. Yeah, that makes sense. But I would expect the ticker being beer. Like, I want the ticker to be generic, you know, for. For. For Budweiser. I would expect the ticker just to be beer. I would expect a ticker.

[C]

I got another one. I got it.

[B]

It needs to be generic. Totally generic.

[C]

I got a. Okay, next up, Hog. Guess. Hog.

[B]

A baking company makes bacon.

[C]

Close. Harley Davidson.

[B]

That's still. It's still.

[C]

That's good.

[B]

It's good.

[C]

You want to invest in hogs.

[B]

That is a very funny ticker.

[C]

I like you. If you want to have broad exposure to hogs.

[B]

Yes. No, no, no. It doesn't actually give you exposure to hogs. What else? Anything else?

[C]

There's.

[B]

There's a few.

[C]

There's Crzy. Crazy Woman Creek.

[B]

Crzy. Cor. What is that?

[C]

That's a tiny bank in Wyoming. Otc. Somebody made a cannabis ETF called yolo. I wonder how that's doing.

[B]

I don't know.

[C]

Probably not great.

[B]

Yeah, there does seem to be some alpha in having a unique ticker. Like a unique dog. Yeah.

[C]

YOLO is down 87% all time.

[B]

Hopefully you didn't actually YOLO into YOLO. Be careful.

[C]

Just absolutely brutal.

[B]

If it looks like you only lose.

[C]

You only live once.

[B]

You only lose once.

[C]

Go down 87%.

[B]

You only lose everything once.

[C]

You're only gonna be able to. If you don't lose everything once, you're never gonna be able to make it all back.

[B]

Yep. Hey, maybe. Maybe the thesis for YOLO was this is a generational opportunity to go short. You could be up 80% if you went short with size.

[D]

Guess so. I have one.

[B]

Play. Play.

[D]

Yeah, Dave and Buster's.

[B]

Okay, that's pretty good. Yeah. Long playing. It's not as much of a pure play as AI or CRM.

[C]

There's also love, which is LoveSac, the beanbag furniture retailer.

[B]

It's a rival to Moon Pods. Don't like it.

[C]

Yeah, don't like it.

[B]

Good luck to everyone over there. Anyway, there was an incredible exchange between Secretary Wright and Jason Calacanis about solar and Elon Musk at the all in Summit. So Elon has a complete wrong. Yeah, I have the video pulled up. Let's play this. Let's play the.

[A]

Has a complete, completely wrong.

[B]

He has a wildly exaggerated view of where solar and batteries will go. And I'd like if we could make.

[A]

A bet 50 years out.

[B]

I'll make a bet.

[G]

Solar never gets to 10% of global energy.

[A]

So Elon Musk has completely wrong.

[B]

So this is the US Secretary of Energy very clearly like long nuclear relative to solar and it's a hot debate. I still feel like nuclear is growing and there's so much like how many startups have we talked to in the atomic world. There's something there Elon, he seems to be a purist about solar and purist about not using lidar on the self driving cars and Elon likes the purely elegant solutions of just the most efficient thing but if the, if energy capacity is like messy and you need it at night there's only so much that you can do. What do you think?

[C]

Yeah, I just want more energy. Just give me more.

[B]

I think that's what as much as possible. I think that's what secretary Chris Wright is saying. It is interesting. Doug from Radiant worked with Elon on a plan to power Mars Mars and ran all the numbers on what it would take to generate enough power to create fuel out of the ice on Mars to bring a rocket back and concluded that nuclear was the best option left. SpaceX started a nuclear company and Elon somehow was unconvinced and went and is still super long solar. What's also interesting is that for all of his Elon for as much of a a solar bull as Elon is he doesn't seem that indexed to the trend. Right. Like he owns SolarCity but he's not, he's, he doesn't have like a mega production facility.

[C]

He feels longer energy storage than he is totally like yeah solar.

[B]

Yeah yeah yeah certainly makes a ton of batteries makes a ton of cars and stuff but it makes sense that you know solar is important too as important is a story that But I mean if you charge your Tesla in 10 years off of a nuclear power plant like he's still a winner. Yeah so it's not like he's, he's bet his entire career or his entire.

[C]

I've never, I don't remember any. I'm sure he's spoken about nuclear but I don't, I never remember Elon saying that he was anti nuclear.

[B]

Yeah I think he's just, just he's, he's, he's been amplifying a bunch of posts that have been basically from like Solar Maxis. But anyway if you're planning your build out of a nuclear power plant, do it on linear. LINEAR is a purpose built tool for planning and building products. Need the system for modern software development, streamline issues, projects and product roadmaps. Internal tech emails has some Larry Ellison Data information from 2002 March 22 we cannot remain in denial if we talk with our with our users every day. Larry Ellison says our top priority is to get Oracle users delighted with the quality and function of our CRM products. Until that happens, I doubt we will achieve success in the market. The biggest problem is missing features as opposed to Showstopper bugs. There are a few of those as well, but they are much easier to fix. Every applications meeting now centers around representatives from our user community who report back on application quality and set priorities for development. We cannot remain in denial if we talk with our users every day. We need to work very hard to achieve a high degree of customer satisfaction for our sales and marketing products as soon as possible. We will have major improvements by June, but we are unlikely to deliver all we need until September. Doing it by September will require unrelenting focus and a Herculean effort. It will be done, but not unless we change our development testing process and do a better job of listening to our user community. Larry Pretty, pretty interesting that they were taking such a run at CRM. I feel like Oracle is not particularly known for that.

[C]

Yeah, I'm trying to figure out how what their, their revenue for the CRM product is today, but they bundle it into cloud application revenue, which is, yeah, huge. Still significant, you know, 3.8, 3.8 billion a quarter. They're already reporting their first quarter of fiscal year 2026.

[B]

I just hope we get Larry taking some shots at Mark Benioff and some putting the timeline in turmoil. That'd be fun. Anyway, did you see this? Did you see this app Poke? Do you see this?

[C]

Yeah, we got to get, we should actually get the team on.

[B]

Yeah, I was, I was texting with the founder. We should definitely bring him on. But Austin Allred interacted with their chat. It looks like he'd signed up, given their phone number, and wound up texting with their AI support rapper, their AI sales rep. And Austin Allred just says, WTF is going on? And so Poke identified that Austin is the founder of Bloom Tech and says, what brings you to my digital doorsteps? Trying to see how the real AI companies do it. And it's like this super sassy fine tune of an LLM. And Austin says, yeah. And the Poke bot says, quotes, yeah, that's all the effort you're going to put in. I expected more from someone who tweets about AI productivity all day. Maybe you should enroll in one of your own courses on how to convince AI bouncers to let you in. Austin just says Too much sass from AI SaaS is literally what I'm here for, Austin. But fine, let me be more direct. Why should you get Poke access when you're already busy running Bloom Tech, managing angel investments, tweeting about AI productivity. But here's the thing. Access to Poke isn't just about follower count. We're looking for people who will actually use it, Provide feedback and help us improve your work. Training devs to leverage AI is interesting. Tell me more about what you'd actually do with Poke if I let you in. And yes, tweeting about it would help your case. I think that's me. Depending on how good you are. Depending on how good you are. And so he has this like, sassy exchange with Bokeh.

[C]

I don't know if this is intentional, but making the screenshots of the product marketing for the user's business is pretty smart. If it was intentional because he has an incentive to share this because it's.

[B]

Saying it's so ridiculous.

[C]

It's funny. And it's also, look, you built an entire business around teaching devs to use AI effectively.

[B]

Yeah.

[C]

Respectively that.

[B]

Yeah. And so Austin kind of closes out. Look, let's get one thing straight. I know you're an AI model. You're a bunch of linear algebra running on a machine. If you can give me productivity gains, then show me how. Otherwise I'll stop using you. I'm not here to impress you. You need to impress me. And so he has this interaction, but it wound up working, right? Like he posted about Poke, he got a bunch of views and likes on the post. It definitely drew our attention to it, So I guess mission accomplished. I still don't fully understand what Poke does, so. So maybe there's a little bit more to do there. But it is somewhat impressive that during the onboarding flow, the system was able to identify who Austin Allred is. I mean, I guess he's enough of a public figure at this point that if you just hit any LLM and asked, who is Austin Allred? It would give you the high level summary of, he's an angel investor, he's the founder of Bloom Tech, and he tweets about AI productivity. And so maybe that's baked in. But if you could bring this level of personalization. I don't know that I agree with like the sassy tone necessarily, but if you could bring this level of personalization to interactions, maybe that's good. I don't know. You're.

[C]

I saw somebody. I saw somebody posting that that Apple should acquire Poke I mean, which is the Interaction Company of California.

[B]

Oh, really? That, yeah, that's their sub brand.

[C]

That's their parent. The parent company.

[B]

No, no, no. Really?

[C]

Yeah, yeah. And they've been that. They've been that way for a while. So remember we gave a cutoff.

[B]

Yeah, yeah, yeah. No more. But then they were the last one through the gate right before the cutoff. Oh well, yeah. I would be interested to hear from the founder, like what this is at least like a teaser for what the product does. Like clearly it's a friend.comsassy fine tune on AI.

[C]

I think it's more. So the difference between this and friend Friend is not trying to be functional at all other than being a friend. And this is trend to be a functional kind of tool assistant type thing that has more friend code.

[B]

I just mean that there is a world where you're a company and you want to bring an LLM into your product. And if you're trying to bring in any other tool off the technological shelf, whether it's a UI component, HTML5, the way a link looks, you will express your brand through css, right? Like you will. You will. You might have. Everyone has a button on their website, but every button looks different. It's an expression of their brand. Right. And so if you're going to bring LMS or chat interfaces to bear, it makes sense to actually try and bring your brand through in some way. And this is a more extreme brand than I think most people would go with. But there is something to the idea that even if you're just a large enterprise company, you might want your LLM voice to actually be an extension of your brand. Whether that's just being slightly friendlier or slightly more corporate or slightly use more or less jargon. You might not want to just show up with the default ChatGPT or Gemini or whatever flavor of LLM that people are know people are used to interacting with. You might want something that's a little bit something.

[C]

Something tells me that the okay team could help the Apple team figure out how to summarize messages without botching them 90% of the time.

[B]

That's a good take. Yeah.

[C]

Anyway, interesting 400 episodes of the greatest podcast of all time in history. Future hall of Famer Founders Podcast Coloss. Warm it up and give it a good hit. Great contact. John Colossus had the write up this morning. This week, David Senner has just released his 400th episode of Founders, a milestone that represents far more than a number. It's a testament to one of the most extraordinary learning projects in podcasting history. Over nine years, David has read and reread 400 biographies and autobiographies of history's greatest entrepreneurs. Distilling their wisdom into a treasure trove of timeless business principles. From episodes one to episode 400 founders has become the ultimate masterclass in entrepreneurial thinking. David just doesn't just summarize books. I know what you're thinking. He excavates the core ideas that built empires, the mental models that created fortunes, and the philosophies that turned ordinary people into. Into business legends.

[B]

Little syntax. That's not bad form.

[C]

Do you think that.

[B]

Maybe. Who knows?

[C]

Who knows? It's well written.

[B]

It is funny how ChatGPT has just ruined that entire stylistical flourish, which was very, very useful for a long period of time. And now it's like. Like there was a big movement on X you probably saw where it was like, bring back the em dash. Make em dashes legal again. Because ChatGPT is so clearly stolen the EM dash from writing that you can't use one without getting accused of using.

[C]

The OpenAI team doesn't seem to be.

[B]

Laying off the em dashes, you see, more than ever.

[C]

I think the same amount.

[B]

Yeah, I thought that it's not this, it's that using the EM dash that was going to be like a stylistical flourish that was going to be left behind with 4.0 and that 5 would have a different flavor, different structure.

[C]

Over the weekend, I kept seeing accounts commenting that were clearly running software to basically like look at the original post and write back a response automatically to it. These are people that build bot accounts up, or maybe it's their own account, just want to grow. And I kept wanting to just like break, you know, my phone and just slam it from the ground because it's so annoying to start reading something and then halfway through you realize it's. But yeah, anyways, this is a very nicely written, nicely written piece and a.

[B]

Huge congratulations to David Senro.

[C]

They say.

[B]

Awesome.

[C]

That is 400 episodes down, a thousand to go. I would put it more in the range of 400,000 episodes, but, you know, if you love David Senra as much as we do, the best thing you can do to make his job his life and his job better is to become one of history's greatest entrepreneurs.

[B]

Yes.

[C]

That is go on a generation career, get a book written about you, get multiple books written about you, maybe die.

[B]

That helps.

[C]

Some of the founders featured are no longer with us and you'll give him content that he can immortalize.

[B]

No, he does cover alive Entrepreneurs. He made three episodes about Larry Ellison in 2020. He was on a serious run. Went on a full Larry Ellison kick. Read three books, wrote three episodes, made three episodes.

[C]

David's story. Maybe we'll have to do. Someday we will make the founders episode on David. We will just fully. We'll study for. For months. Will just emulate his entire process.

[B]

Yeah.

[C]

And because his story is truly incredible. Right.

[B]

It's also.

[C]

He was toiling. He was toiling away for years. Nobody knew his name.

[B]

Yeah.

[C]

When I discovered the show, I knew from, like, I just Knew from. From 20 minutes listening that he was special talent. And I don't know if I was. I think I was one of the first few. I had to be one of the first three sponsors or something. You knew this.

[B]

Yeah, I think you did mention that.

[C]

And I said something that was totally, in hindsight, so offensive. I was like, would you consider, like, selling? Because he didn't. He wasn't generating, like, you know, a ton of ad revenue at the time. I was like, look, you probably make more. And he was just like, no. He said very nicely. He was like, no, I'm not. Not. Not interested, but thank you for the offer. And knowing David now, he was thinking, like, I would not sell foundry for $100 billion.

[B]

It is funny how. How Senra's impact and his, like, his cachet in Silicon Valley and business broadly is, like, massive. And he's never written about. Like, if you were to actually try and do a founder's episode about David Senra, it would be very hard because he's not someone that people just file a lot.

[C]

He's never really done press, doesn't do.

[B]

A lot of press, doesn't try to do press, and also gets left out of, like, articles because he just exists in this. In this, like, in this completely insider space where he doesn't really break out all the time, he doesn't cause a lot of controversy, doesn't get canceled by rain, random people. He's not out there, like, duking it out. And so I think the mainstream media just doesn't realize how big he is in the podcasting world. And so you have this. You have this weird dynamic where it would be extremely hard because he wouldn't just be like, oh, yeah, just pull up, like, one of the many profiles on him. Like, there are profiles about a lot of people. There aren't that many about him.

[C]

I think by this time next year, people have.

[B]

I think they'll pick up on it. Yeah.

[C]

Woken up and Written the proper profile.

[B]

Numeral hq.com sales tax on autopilot. Spend less than 5 minutes per month on sales tax compliance. We're getting a little bit of an eras tour going on. Andrew Ross Sorkin, another friend of the show, a book tour. This is going to be. This is going to be close to eras tour level, fervor and revenue. It's going to be listimania level. I think people will be shaking, crying. If you're big into financial journalism, fantastic research, business histories, you're going to want to attend this. Andrew Ross Sorkin is on tour for his new book 1929. It's a like minute by minute account of the Great Depression, how it happened, who was in the room at a very concrete level. I talked to him a little bit about it. Seems like a fantastic book. I'm very excited for this.

[C]

He's been working on it for years.

[B]

Yes. And just a completely different take on the story that you've heard many times about but you've heard it at an abstract level of like there was a financial bubble. The person who was shining the shoes was giving stock tips and there were all these different accounts and you've heard them at a very high level about the 1929 bubble. But I don't think the story's ever been told at this level of detail. So very excited for this. Obviously it's one for the finance heads more than the.

[C]

I just think anybody that's.

[B]

I think this will be exciting period will be. And if you are in any of these cities on these dates, I highly recommend going and checking it out and joining Andrew Ross Sorkin for a tour of 1929. His latest book.

[C]

Matt Palmer hit the timeline last week and said there is insane demand for people who can understand and explain technology in a compelling way. Zeffie went viral by saying sales. It's called sales. Kind of, kind of a. Yeah, kind of, kind of a.

[B]

Do you think Matt was getting at that? Like what was Matt. Because Matt is saying.

[C]

Gabby Goldberg was also saying this is another word for comms. Like comms in its, in its purest form is not about just connecting you to like media and managing those relationships. It's about like figuring out how to position your position your product and generate attention in your market and drive revenue and these other things.

[B]

Listomania is a real phenomenon during when Fran's list would play, people would shake and cry and be so excited and faint is a popular thing. Tyler, are you familiar with List o Mania? Oh, you got to get up to date on Fran's list. It's great music. 2.6 million in the chat asks when will TBPN go on tour? That's a great question. We have a lot more to build here in the studio, a lot more to do with the show as it is. But live events would be fun. We want to put our own twist on them, of course, so we will be figuring that out in the future. Until then, we will be telling you about Fin AI AI, the number one AI agent for customer service. Number one performance benchmarks number one in competitive Bake off number one ranking on G2. You've seen their billboards all around Silicon Valley. Go check them out. Trunk Fan Shares the R Ipod subreddit is full of Gen Z and Gen Alpha kids recommending which MP3 player to buy and bring to school after majority of US states have banned smartphones from the classrooms. Oh, that's interesting. People are reselling old iPods for $500 plus on eBay. Old is new. This is interesting. We had a viral Instagram short talking about the fact that the ipod might be coming the older ipod might be coming off of patent and so there might be an opportunity to get a new version of it.

[C]

Like the play. If you just wanted to generate return. Clearly a play would have been to go back and just buy. If you could go back in time, just buy, buy Apple products, hold them, don't box them, etc.

[B]

Yeah.

[C]

The question is, are people going to attach incredible value to the iPhone 15? Yeah, yeah, 16. The 17. Are there going to be specific generations?

[B]

I think the original iPhone trades above its value, but I don't think the iPhone 6 does for example, even really? Yeah, yeah, yeah. I'm pretty sure that's that most of them lose value and even something like the original iPhone, I don't think it's outperformed the stock market. So it's very much like a collectible car market where even if you're in unless you're in the absolutely best performing car, even if you're in a car that held its value and went up in value, I believe the Carrera GT has not outperformed the stock market even though it's maybe doubled or tripled in value. It's mid $1.5 million car.

[C]

Yes, you can buy a brand new iPhone 6, 16 gigabytes silver unlocked.

[B]

How much?

[C]

Actually okay, this is interesting. So the first result is just new. You can tell the box is open. Second one is the iPhone 6 factory sealed, unopened investment grade, going for 5,000.

[B]

5,000 but is it going to sell? I wonder. But 5,000, that's a lot for an iPhone 6, you said?

[C]

Yeah.

[B]

Okay. That's a good return. Actually that was a $500 product at the time.

[C]

Okay. But there, here's another one selling for. That's sealed for $88.

[B]

Okay. Okay. I think the market might.

[C]

Which one am I going to buy? I mean. Yeah, probably.

[B]

Great. Well, is one of them signed by Tim Cook or Steve Jobs?

[C]

I mean, a signed iPhone.

[B]

Signed iPhone. Anyway, Antonio Garcia Martinez has an interesting anecdote here. Hyperliquid is generating more than $1.2 billion in net income, surpassing Nasdaq's 2024 net income of 1.13 billion. Remarkably, it achieves this with a workforce 832 times smaller while directing 98% of revenue to hype token buybacks. And Antonio says, funny that the proverbial four person unicorn was achieved faster in crypto than AI.

[C]

Well, and this is happening happened before there was a. There's a Y Combinator company called like ax.

[B]

I remember you. Yeah. You were talking about this. Were they a unicorn out of a.

[C]

Batch that we covered? I don't know if they've been valued at a.

[B]

Sure.

[C]

Doing like hundreds of millions of revenue.

[B]

Yeah, yeah.

[C]

I think, I think comparing hyper liquid, which is like a, is a, is more of like an exchange product to an actual stock exchange is like that is that.

[B]

I mean, it's not the craziest comparison.

[C]

I mean it still still sounds crazy to be like, you know, tiny company.

[B]

Yeah.

[C]

Generating.

[B]

No, no, no, no. It is remarkable. I mean you could go further. Bitcoin, it's a trillion dollar asset. It's a one person unicorn. Right. Or maybe a couple contributors. But mostly, you know, extremely tiny, extremely high revenue per employee. But yeah. Wild. I still think we will probably see those. The four person unicorn in AI eventually. But we'll need to keep tracking it.

[C]

Can we pull up this video in the timeline tab?

[B]

And while we do that, let me tell you about adeo Customer relationship Magic. ADEO is the AI native CRM that builds scales and grows your company to the next level. They need a stock ticker. CRM2CRMA. That could be AIC.

[C]

Five letter stock ticker.

[B]

I don't think you can do five letter stock ticker yet.

[C]

We, we, we know some people. Nice.

[B]

That would be a whole. It's like when the new like dot AI domains came out or dot X, Y, Z domains came out. There's like this flourishing of, of new domains that flood the market. If, if one of the exchanges went to five letter tickers. That could be bullish. Anyway, you wanted to pull up a video. What do you got for us?

[C]

Roon has a post here highlighting this new humanoid demo.

[B]

Okay, let's play it.

[C]

Let's pull it up. We can't see it.

[B]

Here we go. Whoa.

[C]

Daddy has some feature. Gets it to.

[B]

It just pops right back up. That's remarkable. Wow, that slow mo is really insane.

[C]

Gets kicked over.

[B]

That's so fast. That's so fast.

[C]

I think. I think a use case for humanoids that.

[B]

This is. People.

[C]

Aren't people always talk about humanoids? Oh, they're going to fold. Fold your laundry.

[B]

Yes.

[C]

Do your dishes. I just want to be able to box one huge market for shadow boxing right now.

[B]

Yeah, yeah, yeah.

[C]

Under monetized. When you can just kind of be, you know, throwing, you know, haymakers.

[B]

I mean, you can go as hard as you want, I guess.

[C]

Go as hard as you want. And you can train the robot. Robot to. All right, you can punch back, but just like don't take my head off kind of thing.

[B]

That is.

[C]

Hopefully there's no bugs.

[B]

That is wild video. We keep getting those. Do we know what company produces that? Is that unitree?

[C]

I think this is pretty cool. Let them go public in the US.

[B]

And then in five years we're like, we have seven delays in the divestiture. There's going to be. There's concepts of a plan to get unitree to divest in the United States. Who knows? Anyway, speaking of tech in Washington, Will Menidis has a little thread for us. The cultures of tech in Washington are fundamentally incompatible, says Will Menidis. In tech, the foundation of a good career, the foundations are reckless. Sharing of favors, expecting nothing in return, endorsing young ideas and people and public networks of influence in Washington. Any of these will end you, says Will. In the fullness of time, tech will end up having dramatically less influence on American politics than finance or any other industry that was once a similar size. We are used to infinite positive sum games. Tech came in like a conquering army and we will be run out by the midterms. It will be a record year for lobbying firm profits, certainly a record year for state House lobbyists as tech learns their ABCs of government. And you should expect a lot of consolidation in retirements in this space. I love that. But the number of bills will be at an all time low. This too will fade. If I had to make a trade, I would be willing to bet that we will have two vice presidents that spent Some of their career as venture capitalists before we have a tech native lobbying firm with over $10 million in durable bookings. And Tae Kim says, I think that this is generally true for small to medium sized startups growing rapidly. But as companies grow larger, the bureaucracy, empire building, power warfare, gaming of metrics and internal politics creep in. Tech is not immune. So interesting to see where tech is going. In Washington. Will was on a tear talking about.

[C]

What'S your major takeaway from this?

[B]

I think Will's really breakaway take this weekend was that, oh, this is an older post from August 3rd. I don't know how this resurfaced, but I think it was because Will was saying that the tech industry needs to focus away from the sci fi stuff. The leas like tech kind of and AI, the technology AI, the foundation, model labs. It does feel like they let Eleazar Yudakowski set the agenda for what was discussed. It was like all paper clipping debunks and fast takeoffs. And Will's discussion of this has centered around this idea that the discussion needs to. If you are engaging in American politics, if you are engaging in Main street politics, people that actually vote in elections, you have to put it in terms that are relevant to them. Which means what will the impact of AI be on your job, on your community? Will it be positive? Will it be negative? There's a lot of work being done right now on what AI means for Main street, for the average American. And it's definitely an open question. The models are at a level that can displace labor for sure. And the impact of that can be positive if it's handled appropriately. It could be very negative if it was handled inappropriately. And so I think that's where I think the sci fi concerns have landed on deaf ears broadly in America. And they've been written off just as like, yeah, of course I'll put that in the bucket of maybe there's unlimited fusion energy and we'll be on Mars and we'll have faster than light travel and we'll have AI super intelligence and all that sci fi stuff. Cool. Keep it in Hollywood, folks. Like, talk to me about like, can I get a job tomorrow?

[C]

Yeah.

[B]

And for a lot of technology, there's been a debate about like, what has the effect on the labor market been? Unemployment in America is still really low, but are the jobs as satisfying as they used to be? Did people prefer being subsistence farmers or are they cool with only 3% of the population doing farming? Like, I think that that's a net Gain. I think people are having. Happy with that people. I think people actually do like their email jobs more than being on the farm. But, you know, I don't know, some people, you know, it's not. It's not that crazy to say, like, yeah, it was actually really rewarding to just be a hunter in a cave and go out and hunt a gazelle.

[C]

I mean, look at the. Look at the trad movement. Right?

[B]

Yeah.

[C]

People want to return. To return. Just going out.

[B]

This was the.

[C]

This is off the land.

[B]

Yeah. This is a transposition of the. Of the Unabomber. Take that. The power process of having the work that you do directly relate to how you survive. So if you go out and you hunt the gazelle and you bring back the gazelle and you eat, that provides for survival. And the further you get away from that, once the point that you're doing an email job to buy food at the grocery store, that is an abstraction that makes it harder to satisfy the fundamental power process that every. That is innate and every human. And so I could wind up benefiting everyone, but making everyone's job more abstract because it's like I don't even paint the painting anymore. I just prompt it. And even if I'm making the same amount of money.

[C]

Did you see two people now? Like, if you wanted. The models are actually better at prompt engineering than humans.

[B]

I saw this.

[C]

They used to be like, oh, don't worry, you're going to be a prompt engineer. You're going to be fine.

[B]

Yeah, yeah, yeah.

[C]

It's like, oh, actually just saying, asking the model, what's the best way to prompt if I want the model, if I want this kind of output.

[B]

Yeah, yeah, yeah. I've done that a lot. I actually go to ChatGPT and say, Build me a VO3 prompt that will describe this particular scene and flesh it all out with the level of detail that will get me consistent video output. But there's still obviously a ton of value in creativity and coming up with interesting ideas. There's still a ton of value all over the place that hasn't been eroded. And at the same time, one day we'll be saying, okay, well, the models are plateauing and the enterprise adoption isn't that high and labor productivity isn't spiking. So the actual rollout, we're looking at what, $4 billion in anthropic, $12 billion at OpenAI. Like, there isn't that much value that's being generated.

[C]

China's AI action plan is we're going to integrate AI into manufacturing workplace. We're going to drive productivity and efficiency across the enterprise.

[B]

Yeah.

[C]

Soham. Not Soham Parikh. Soham. ETW says my Twitter account has more MRR than most YC startups. And I think that is entirely untrue in itself.

[B]

It's true in two ways or it's untrue in two ways. There are two flaws.

[C]

There are two flaws.

[B]

The average MRR at YC startups is certainly higher than the X creator payouts. The X creator payouts are usually. I would be shocked if it was over a few thousand dollars for some. He's kind of just taking a shot at YC startups for having like low MRRs. But when we talk to these companies, the MRRs are usually in the tens of thousands, hundreds of thousands of dollars.

[C]

Oh, so this guy created an open source version of Clulee and called Cheating Daddy, a free and open open source app that lets you gain an unfair advantage.

[B]

What you were looking for what everyone wanted. Well, Casey Flint put Soham in the Truth Zone, said one day people will learn what the middle R in MRR and ARR stands for. Not today though. Tyler, do you know what the middle R in MRR stands for?

[D]

Isn't it recurring?

[B]

It is recurring.

[C]

Why did you have to take so long? Did you think it was Millie? And the reply says R for random monthly random revenue. Yeah, we had better revenue here, better revenue there. We counted, you know, this revenue that was over these four months in this month. No.

[B]

Yeah, there was someone.

[C]

No. Some people say I am at a hundred million ARR. And you can say that can be also seen as annualized run rate.

[B]

Yes. ARR.

[C]

Somewhat disingenuous if you're trying to market it as like annual revenue contracted.

[B]

Yes. But I mean, to be fair, like there is a wild, there is a wild like gap between the quality of ARR. Like there are some, there are some companies that have contracts that are so ironclad they would bankrupt the customer to try and get out of them. Like it is. It is written in stone that you're paying that bill every month for the next five years.

[C]

And yeah, Elon went to war with a bunch of SaaS providers during the Twitter. When he got in there at Twitter and he was like, wait, we're spending.

[B]

How much on this, how much on that?

[C]

And just, yeah, I'm not paying.

[B]

And he probably was able to pull out of a lot of those that were booked as ARR as true annual recurring revenue at the same time.

[C]

But again, it's like if there's a. There's a degree. Even if something is contracted, if they have to sue Elon in order to get the money back, it's like. Or get what they're owed. It's like at a certain point they'll spend more on the lawsuit than the actual value. And Elon probably ran the numbers.

[B]

Yes. Don't misstate your mrr. Don't misstate your ARR. Just be honest until your investors. And folks, what's really going on with your business? How'd you sleep last night?

[C]

I slept fantastic.

[B]

Me too. Me too.

[C]

That must be why we're feeling so good. I only got an 86, but I got.

[B]

So I got. I spent not the full night in the eighth sleep, so I got a 76. It says I only slept six hours, but I got another three in the kid's bed before. So I got like nine hours. I'm feeling. And fantastic. I need to get an eight sleep in his room so I can like merge all the data together, give it the real, the real picture. Anyway, you can go get eight [e-mail], get a pod 5, 5 year warranty, 30 night risk free trial free returns, free shipping, Netcap girl.

[C]

We must throwback post from 2022.

[B]

Yes. This is what started it all.

[C]

We must return to an era of naming companies like American Computing Corporation or US Semiconductor. The days of cute and silly startup names are over. No longer is it sufficient to just drop a vowel from a word and call it a day. We must build serious companies with serious names.

[B]

Wow.

[C]

And again, this was the browser Company of New York already existed. They had kind of done it.

[B]

It was the start of the trend.

[C]

It was the start of the trend.

[B]

Yeah. But then we got.

[C]

But then we got Amco.

[B]

American Manufacturing Company of America. No, Advanced Manufacturing 2Americas. That would be the real. The real, the real final boss. The New York Browsing Company of New York. The American Browsing Company of America. People have been having a lot of.

[C]

Fun with this, this meta. This meta.

[B]

I think meta's over.

[C]

Should be. No. The American Computing Corporation or US Semiconductor. I still think this is why I don't.

[B]

I think it's. I think it's over. I think going forward, I want. The thing that I want is Coogan and Sons.

[C]

Yeah.

[B]

Last names Coogan and Sons.

[C]

And Chris Amadon did this.

[B]

Oh yeah, he did.

[C]

He did this. But it's an evergreen meta because no one can say, oh, you name the.

[B]

Company, you can take it further. Amazon sounds like the name of a startup a little bit. It kind of works like it's not Amazon and Sons, it's not Amazon and Amadon. That's what I want. I want someone to come in with the, with the Cosgrove, Cosgrove and Cosgrove LLC and build an AI ERP solution or something. Business process outsourcing. Something like that. Anyway, whatever your startup name is, throw it on a billboard, get on adquick.com out of home advertising Made Easy and Measurable say goodbye to the headaches of out of advertising. Only Add Quick combines technology out of home expertise and data to enable efficient seamless ad buying across the board.

[C]

I'm going to let you run the run through these legendary Hacker News comments.

[B]

So Hacker News is legendary. Tyler, are you a Hacker News user?

[C]

Occasionally.

[D]

I'll go on, but I never go into the comments anymore.

[B]

Yeah, I've only gotten one really solid post on Hacker News. I think I have like 154 karma. But I mean when I was your age Hacker News was like the place. Like it was, it was the spot. Like every single day. Opening up, reading threads. That was where news broke, where the discussion happened.

[D]

I used to go on there a fair amount. I think I've kind of replaced it with Marginal Revolution.

[B]

Oh really?

[D]

Tyler posts like whatever, five or six links every day.

[B]

Yeah, yeah, generally. Are you in the comment section?

[D]

Even those? I don't, I don't.

[B]

You don't comment. You're just a lurker. Yeah, okay. Well Hacker News is legendary for not enough lurking. These people really have some all time bad takes. These are legendary Hacker News comments that age like milk. So the first one Dropbox, one of the greatest YC companies to come out of the early batches. Posted. This is from Drew Houston, the founder. He posts on Hacker News, my YC application Dropbox. Throw away your USB drive. Of course. Dropbox is the way to sync and share files on the Internet. And Brandon M. Chimes in in 2007 right when the company was getting started, I have a few questions. I have a few qualms with this app. One For a Linux user, you can already build such a system yourself quite trivially by getting an FTP account, mounting it locally with curl FTPFs and then using SVN or CVS on the mounted file system system from Windows or Mac. This FTP account could be accessed through built in software. Why don't you just do that? Why don't you just build your own Dropbox? It's so obvious and I love this because obviously the take was wildly incorrect. Dropbox became a multi billion Dollar company. But it's a beautiful bubble to live in where you assume that everyone is comfortable booting up Linux and building a Dropbox competitor on their own. Maybe with Vibe coding it's possible, but. But still, I think people are still using Dropbox. The company's actually been doing quite well and has not been really affected anyway. Second, hacker news posts that age like milk. Bitcoin. Peer to peer network based on autonomous anonymous digital currency. This is from 2009 and the top comment here says, well, this is an exceptionally cute idea, but there is absolutely no way that anyone is going to have any faith in this currency. That is a wild thing. I mean it was reasonable at the time. It was a crazy idea. And the faith in the distributed system took years to build up. It was seen as a toy and then a joke and then a meme and then a bubble and then a crash and then another bubble, then a crash and then another bubble. But eventually it became accepted and now people have lots of faith in bitcoin. And even with all the turmoil the that we've seen in the market and turmoil in international geopolitics, bitcoin is now at $115,000 a coin from what was probably a few cents when this post went live. What was the price of Bitcoin 5-8-2009? That is an interesting question. I imagine it was very low and now it has appreciated quite significant significantly. Probably enough to buy a luxury watch on Bezel. Go to getbezel.com your Bezel Concierge is available now to source you any watch on the planet. Seriously, any watch. Tyler, do you have the price of bitcoin?

[D]

Everywhere I go it just says zero.

[C]

Yeah, zero. No, but here's the story.

[B]

Yes.

[C]

So it didn't have a market value at the time. I'm sure they were free. Yeah, it was basically free.

[B]

It was free.

[C]

The first recorded real world transaction for bitcoin occurred over a year later on May 22, 2010.

[B]

No way.

[C]

When a guy bought two pizzas.

[B]

That was the pizza transaction.

[C]

Two pizzas for 10,000.

[B]

Wow. I didn't realize it was that old. That is, that, that is crazy. Somebody was posting about this on hacker news in 2009, a year before the pizza transaction. That is crazy. I mean the pizza transaction is famous. Famous. How many bitcoins was it? It was like 10,000. 10,000 bitcoins.

[C]

I wonder if there'll be another for two pizzas. I wonder if there will ever be another trade like that where you could just pick up.

[B]

They should make a whole documentary about that trade where that bitcoin went. Did the next person hold? Did the next person hold how the value accrued? Fantastic. Anyway, in 2023 this is a more recent Hacker News comment that age like Milk Michael Truel from Cursor now worth I think in the secondary market trading almost north of 20 billion, certainly north of 10 billion. A massive company used by probably millions of developers at this point. Posts on Hacker News Cursor, a code editor built for programming with AI and the top comment from Yak account4 on March 24th of 2023 there's nothing on the official website or GitHub that indicates what this software is, other than a cropped screenshot that looks like VS code with a prompt cover with a prompt pop up over it. And I still can't figure out if this is some sarcastic joke software or something. Wow. Age like milk not great. And the final Y Combinator Hacker News comment that age like Milk Y Combinator's airbed and breakfast casts a wired A wider net for housing rentals as Airbnb from TechCrunch in 2009 and a hacker News commenter chimes in and says Airbnb isn't just casting a wider net, they're changing their entire business model. This is. This is what two decades before OpenAI started using or ChatGPT started using this, this syntax. Airbnb isn't just casting a wider net, they're changing their entire business model because the original one is severely flawed. Renting a room or bed out of your home like a hotel most likely violates multiple tax and business laws in just about every US city, let alone the severe liability that a guest slips in your shower or breaks their collarbone. I've worked at a resort hotel. Cities have specific bed taxes. Hotels have to be inspected. They have liability insurance. Vacation rental owners pay taxes. Some cities don't even allow short term rentals. Of course, none of that mattered. Airbnb figured it all out. They paid the taxes, they fought the local governments and created new legal structures to satisfy all parties involved. And now the business is worth tens of thousands.

[C]

They say if I wasn't a vc, I wouldn't touch. If I was a vc, I wouldn't touch this site.

[B]

That's what it says. Yeah, well, if you're looking to go upmarket and find a vacation home with inspiring views, hotel graded amenities, dreamy beds, top tier cleaning and 24. 7 concierge service. Get on, wander and find your happy place. Good call cutting to Tyler on that one. It's a vacation home. But better Hayden identified exactly the trend that you were feeling. This is who you're arguing online with. And it's just a picture of a neural network, of course. And 31,000 people enjoyed it. It really is. The bots are getting wild, wild, wild, wild. Well, I have some tips for those of you who are using X the everything app. I recommend going into the notification filters and you can turn off notifications from accounts that don't have a verified phone number, don't follow you, you don't follow them. And after you do that, you will basically only see notifications in the notifications tab from people who you're mutuals with. People that you can evaluate their full account to see that it is in fact a real person. Maybe someone you know in real life, maybe someone who has enough exhaust on the Internet that you can. They've established themselves as an interesting poster who's clearly not just a bot. And then you will only see those folks in your replies. And then you can always see who's replying to your posts if you click on the actual post. But for the most part, that has dramatically upgraded my experience. I don't get, I really don't get any notifications from just random accounts. I only see. I only see stuff from people I'm mutuals with, people who I actually care about. And it feels, the actual experience of being on X feels like being in a very small chat room. Even if I have a post that gets thousands of likes, I'm really only seeing. Oh, oh, Tyler liked it. Oh, Jordy liked it. Cool. I made them smile. They're my friends. It creates a much tighter, small hometown feel.

[C]

If I don't like one of your posts, it's gotta be really bad.

[B]

Yeah, because if. What's the quote? If there's one person liking one, like it's me, it's me. Fantastic. Anyway, we have our first guest of the show or second guest of the show because we already had Bill Bishop on the show and the restream waiting on room. We'll bring in Pete from Psiquantum. Pete, how are you doing?

[C]

What's happening?

[F]

Hey guys, how you doing?

[B]

Doing fantastic. How are you? I hear you have some big news. Break it down for us. What's new? Introduce yourself at the company, please.

[F]

Oh yeah, thanks. Thanks a lot for having us on. I really appreciate it. I'm Pete. I'm a co founder of Psi Quantum. We are a quantum computing company based in the Bay Area. And yeah, I moved here 10 years ago to build a Quantum computer and thought I'd be on the boat back six months later and we're still here. And we're really just like humbled to be now announcing a billion dollar fundraise for us to go ahead and actually build giant quantum computers.

[C]

Incredible.

[F]

There you go.

[B]

It's massive. $1 billion.

[C]

Okay, give us the 10 years in two minutes.

[F]

Well, I'll give you the the.

[C]

Prior.

[F]

To 10 years, that 10 years, myself and my co founders spent about a decade in the university system doing what people think of when they think of quantum computing, which is science experiments in a research lab with single photons and chips and so on and so on. Published a ton of papers and then got fed up with that. And some of my colleagues made a bunch of breakthroughs in the architecture for optical quantum computing. And so we sort of developed a conviction that we could actually build a million qubit system. You know, Google and IBM and others are all in hot pursuit of useful quantum computing. Everyone today can make systems of tens or hundreds of qubits, but you need about a million qubits, you need a huge system. And so we sort of convinced ourselves that we could actually build that quickly through leverage of the semiconductor industry. And we moved to Silicon Valley basically to build the organization that would do that. That meant taking weird devices like single photon sources and single photon detectors that you know, you can't find in a regular manufacturing process and putting them into a commercial fab. So we spent north of $100 million and many, many years putting superconductors and single photon devices into GlobalFoundries Fab 8, which is very big, very mature commercial semiconductor foundry in upstate New York, last remaining bastion of pure play semiconductor manufacturing in the United States. Big emphasis on semiconductors basically. For many years we were doing other stuff as well, quantum architecture and algorithms and systems and so on, but really spending money on semiconductors. Now we have these wafers we can make like high volume. We can manufacture in pretty high volume the devices that we need. And so our emphasis is switching towards building data center like machines. And so last year we announced about a billion dollar Australian deal with the Australian government for us to build our first system just outside Brisbane. A few months later we announced that we'd be the anchor tenant of a new $500 million quantum computing campus on the south side of Chicago. Former US Steel site used to make a million tons of steel a year. We are honored to be breaking ground there in a very, very short period of time. And yeah, just thrilled to be at this level of maturity and sort of readiness now to actually realize these systems.

[B]

How does the Australian government even assess something like this? I feel like the government is not known. I mean at least the US government's not really known for having like tons of technical expertise. Internally famous examples of like the government struggling to build a CRUD app for, you know, registering your health care information. It feels like the technical due diligence to understand how to underwrite that investment and then actually get any sort of value out of it. I mean you have to rewrite the systems to actually write something that's going to run on a quantum computer. Correct. What was that deal like? What is Australia optimistic about in having a billion dollar facility?

[F]

Yeah, I mean governments struggle terribly to build CRUD apps, but the entire might of Silicon Valley also struggles to build a good to do list app. There is still no good to do list as far as I'm concerned. I have a sort of provocative thesis that building to do lists is actually harder than building quantum computers. But that one will get me in trouble. But you know, governments have excelled obviously in Manhattan Projects and James Webb Space Telescope and you know, Space Shuttle and you name it, like semiconductors, the Internet. And you can also look at companies like SpaceX and Tesla and TSMC and ASML, like the pillars of our economy and our advanced society. Those companies wouldn't be the companies that they are today without literally billions of non diluted government support in every case. And so I don't think it's unusual that a government supports this. I think governments increasingly understand that hard technology and this kind of frontier technology is sovereign and strategic and dangerous if you don't pay attention to it. We see that very acutely with semiconductors and AI supercomputers. And then you asked about the sort of diligence. So the chief scientist of Australia spent about 18 months together with a pretty big team evaluating what we're doing. They came to our labs and held us by the scruff of the neck and made us reproduce measurements that we were claiming. There was a Freedom of Information act request that revealed her emails early in that process. And her emails revealed that she was highly skeptical of us when she first met us. And the Australian press wrote about this in the frame of like, you know, chief scientist highly skeptical of psych quantum claims. Which I'm like, what do you want there to be? Do you want to be, to be like highly fragile?

[C]

It's her job to be skeptical.

[B]

Yeah.

[F]

And, and you know, we were just like super gratified by the end of that process. That Obviously after speaking to a lot of different, different outfits, they chose us.

[C]

And then in the US So sorry to cut you off, but what is. So the Australian government, you guys are partnering with them, what is the value that they're hoping to get out of this project?

[F]

Yeah, so I mean I think like if you look at semiconductor fabs, if you look at the current effort to get TSMC into Arizona, if you look at some of the force trading that's going on around the location of AI supercomputers, I think nations now understand that it is strategically important where these systems are built. And you know, when the US incentivizes TSMC to build in Arizona, the US government isn't buying chips from tsmc. You know, they're not owning the fabs, they're just making sure that that infrastructure is built on a piece of land that they care about. And to first order, the Australian thing is structured the same way. It's worth saying that Australia as a nation has been historically forward thinking in investing in the academic foundations of quantum computing. So a lot of the basic research, two of my co founders, a lot of the early work on optical quantum computing and error correction all came out of Australia. And so it's pretty poetic and touching actually that it's heading back to Australia where it all started.

[B]

What are some of the use cases that you're excited about? People talk about like oh, you'll be able to break Bitcoin, that'll be like disruptive or cryptography.

[C]

But is there or quite profitable?

[B]

Yeah. If you guys do it, it's the same as asteroid mining gold. If you get, if you bring $10 trillion of gold to America, what happens? But in the context of AI and supercomputing actually delivering business value, is there sort of like a near term scenario where quantum computing can drop the cost of inference or unlock a different scale of training? Or will the architecture of what we are doing with AI fundamentally change once a million qubit quantum computer comes online?

[F]

Yeah, so we're not going to replace GPUs, we're not going to replace AI supercomputers. And although there is some papers that talk about quantum machine learning, that stuff is really early at the moment. If someone tells you they know how to do quantum machine learning today, you should be pretty skeptical.

[B]

Sure.

[F]

What we're excited about is, you know, the way I think about this is that language models now I visualize them as this sort of expanding frontier that's just like eating through everything we do with computers. And by the way, everything we do in our lives and so on. There's a known limit to that beyond which we don't expect conventional computers will go. And on the other side of that threshold is chemistry, material science, drug discovery, fuels, catalysts, fertilizers, semiconductors, like the microscopic foundations of our physical world, where conventional computers have a really, really hard time doing their simulations, and where we don't have the training data in general to train AI models that will reach far into that virgin territory. And that's what we're really excited about with a quantum computer. A quantum computer calculates from first principles rather than making approximations from data. And through that kind of first principles calculation, we think we can reach much, much further into unexplored territory. The other thing that's maybe interesting is that at Psych 1, we make the highest performing photonics in the world against a whole bunch of figures of merit. And of course, photonics is increasingly being viewed as a potential solution to what is otherwise maybe a difficult situation in the scale up of AI supercomputers where we run out of power, run out of space, run out of money, et cetera.

[C]

What were the key unlocks that allowed you guys to put together this billion dollar round? It's certainly larger than most IPOs.

[F]

Yeah, yeah, it's pretty surreal to be clear. Right. This is the only job I've ever had. The only other real job I've had is working in a garden center stacking compost. So yeah, I should be clear. This, this isn't, this doesn't feel normal. It feels pretty crazy to me. I'd put it also in context with other things that have happened in Silicon Valley where people have spent a way larger amount of money before they get to a product on things that are debatably more or less valuable than a quantum computer. But the unlock. One of the biggest things that we did when we raised our series D with Blackrock and Temasek and others, Bailey Giffords, we told them that we were going to unlock one of the hardest problems for us, which is a new material for optical switching. We had to build an enormous crazy machine here in the South Bay. Biggest molecular beam epitaxy tool in the world. Molten titanium inside, liquid helium inside. Huge risk for us. And without that, there's no way we're building an optical quantum computer. The shipping company dropped that into the Alameda bay and broke it. So we had some trouble during COVID but we got it done on time, we delivered, we got it built. And then the other thing that we've.

[B]

Done is it still at the bottom of the ocean or did you get.

[F]

It then it just like fell and cracked.

[B]

Okay.

[F]

We were like late on the schedule.

[B]

Scuba diving equipment go down, get. Not quite.

[C]

John's scuba certified.

[B]

I like scuba diving. I like treasure.

[C]

Willing to help out.

[B]

That's amazing.

[F]

I mean there's gotta be some amazing stuff down there, like out on the, on the port at Alameda. It's gotta be some beautiful stuff. But then the other thing we did, you know the chandelier?

[B]

Yeah.

[F]

Like the golden chandelier that politicians love to take photographs with. We got rid of that.

[B]

Okay, got it.

[F]

That's in the trash.

[B]

What do the energy requirements look like over the next like decade? If we assume that there's a million qubit quantum computer running, are we going to see a similar race for just like intense energy production that we're seeing in the AI super computer build out?

[F]

No, it's like, it's like the same story but told backwards is what you should expect with quantum computing. So like, yeah, AI and conventional data center. It's pretty helpful for us given that we're talking about building big machines.

[B]

Yeah.

[F]

Ten years ago in Silicon Valley, if you were talking about supercomputers or HPC or like building a big computer, people would fall asleep in the meeting and throw you out of their office. Right. It's nice that we're now surrounded by people who are sort of brave about building big machines. But we are going to build a big machine. And then the good thing is that three months into building it, it's already going to be out of date.

[B]

Right.

[F]

Like we, we're at the beginning of that sort of densification, miniaturization, power efficiency journey. Whereas AI and conventional supercomputers I think are coming to the end of a multi decade journey of more and more efficiency and so on. So it's actually pretty exciting that we should be able to further miniaturize and densify these systems over time.

[B]

It's fascinating stuff. Thank you for hopping on the stream. I learned a lot.

[F]

Thanks for having us.

[B]

This is such a complex industry, but it's been really fun tracking all the progress.

[C]

I don't think we've had any, any.

[B]

This is the very first quantum company on this show. I haven't put, I haven't thrown out a lot of invites because I'm so out of the loop on it. But thank you so much for taking us through it and actually explaining it in terms that even I can understand.

[F]

Really appreciate it. Thanks.

[B]

Have a great rest of your Day. We'll talk to you.

[C]

You're the man.

[B]

Our next guest is already in the restream waiting room. We'll bring them in and I will let you take the introduction.

[C]

What's going on? Welcome to the show. It's please, please pronounce your name before I butcher it.

[H]

Saadi Khan.

[C]

Saadi, great to meet you. Great to meet you and welcome to the show. Introduce yourself and the company.

[H]

Yeah, I'm Saadi. I'm the co founder and CEO of Avenue Aven is a home equity backed credit card company. We launched our home equity card about four or five years ago and it's been on a tier since then. In 15 minutes or less, you can save 50% or more on your monthly interest rate as a homeowner in America. We just announced our series E last week and we're excited to do the Goblin.

[B]

Thank you.

[C]

Series E. Hard to get into those numbers across four years.

[A]

What?

[C]

Walk us through? I guess. Yeah, I'd love to. I think it's, I think our audience will generally understand how if you have a credit card backed by your home equity, it'd be easier to be competitive from a rate standpoint. But break down the actual way that you, you make this work.

[H]

Yeah. So as you know, there's about, traditionally a home equity credit card, home equity in general takes about 30 days to originate and cost thousands of dollars. And as you know, a credit card takes a few minutes to get from Capital One or Bank of America, any great bank. And so we started with this mission of how do we make it as fast as a credit card to get heloc? And so we got it down to as fast as 15 minutes by machining down this process for maximal efficiency. And so whether it be like robot arms that do signatures to automatic income verification, the IRS to everything in between, we're able to compress this thing down to 15 minutes. Sort of feels as convenient as a credit card. But you get the rates of a HELOC and the rewards and the convenience of a credit card. And so we offer 2% unlimited cash back on every single transaction that you do on a HELOC for the first time ever. And you get it in the form factor that you can go to Home depot and spend $1,000 on lumber, or you could go to Whole foods and spend $5 on a banana. And our customers have been able to do both.

[C]

That's, that's pretty incredible. What, I guess what were kind of the key unlocks in the journey was there. How did you navigate? Like, how has like the interest rate environment kind of impacted the business as well. I mean, ultimately there's always going to be a lot of home equity that's just locked up that people aren't able to leverage, regardless of the rate environment. But I'm curious what trends you've seen.

[H]

Yes, I think at a very high level, our goal is to build a product which has the lowest cost of capital for consumers after your primary mortgage. So regardless of the interest rate environment, if it's high or if it's low, that's something we don't control. Jeremy Powell controls that. So regardless of what the interest rate does, our goal is to come in and offer every single consumer the lowest possible, lowest cost of capital after your primary mortgage. And we've generally historically done that. And when interest rates go up, there's generally two or three things that happen and interest rates go down roughly the inverse of those two or three things happen. So I think we're all expecting interest rates hopefully to come down a little bit for consumers over the next few months on the next Fed meeting. And the three things that I think will happen are a, we will expect more demand for capital.

[B]

Right.

[H]

More people will use more credit, generally speaking, across not just our product, across all products, we become actually more competitive in some ways because if you kind of think about it, the interest rates dropping drops to a larger proportion. The drop affects a HELOC or a mortgage by a larger proportion than a traditional credit card, which is also a variable rate product. And then finally there is some kind of trace effects of how interest rates and mortgages kind of interact with one another, where due to the very low historical interest rates, people are sitting on extremely low mortgage rates. But because of the increase in interest rates, people are not able to refinance their mortgage and get a cash out, which they've done historically. And so they're tapping HELOCs to do home improvement projects, make their home better. Because it's getting harder to move to a new home and it's getting harder to do a cash out refinance.

[C]

Yeah, you said something about robotic arms doing signatures. Anything more to that story?

[B]

What?

[H]

Yeah. So I mean, I don't know if you guys, when you guys bought your home in San Francisco or elsewhere, when you closed your mortgage deed of trust, you had to go to a notary or you had to go to a title closing station and you have to sign a stack of papers, yay thick. And we looked at that, and if you wanted to get this thing down to 15 minutes, there was just zero way we can do it by having a notary show up at your house or you going to a title company. So he went, well, how can we figure out a way to be able to do a remote wet signature? And so we invented this robot arm that literally, you control from your phone and you can execute what is something called an original wet signature that is acceptable in every single country. So we can close this thing down in 15 minutes. We do some crazy shit.

[C]

Is that. Did you guys ever get. I don't know if many people.

[B]

I'm sure regulators hate this one weird trick.

[C]

Well, it feels like potentially. It feels like potentially a workaround. That. How different is it for me to move a pen than it is to move an arm that. That signs a dock? Right? Why do I need to be in the room, you know, why do I love this?

[B]

This is great. Probably.

[A]

Yeah.

[H]

We spent months going through every single. I mean, people have tried this before, like, trying to build, like, a robot arm to do wet signatures. There's a lot of detail in our patent that goes down to, like, the exact amount of memory box we have to keep on, you know, the device versus the wire, versus your phone to ensure it is always an original signature and not just a remote signature.

[C]

My thought was, like, you could spin this out, and it's probably like a billion dollar business by itself.

[H]

We've been told that before.

[B]

You've been told that before? There we go.

[H]

We've been told that before. But we're just very focused on making sure homeowners can get access to the lowest cost of capital today.

[C]

That's amazing. Well, congratulations on the funding milestone. And, yeah, I'm sure you guys are just getting started.

[B]

John, how much was it?

[C]

$110 million.

[B]

It's the week of the Series E overnight success. Overnight success.

[C]

Love to see it.

[B]

Thank you so much for coming on.

[C]

Great to meet you, Saudi.

[B]

We'll talk to you soon.

[C]

Cheers.

[H]

Awesome.

[B]

And our next guest is in the Restream waiting room. We have Lucas Singer again from Divergent.

[C]

I wanted him to drive one of his supercars into the Ultra.

[B]

I know, I know. And I've heard unsubstantiated rumors that he might be down for that at some point. Couldn't make it happen today. Welcome to the show. How are you doing? Lucas? Good to see you.

[G]

Great to see you. Doing great today.

[B]

Kick us off. Did you raise a series what now?

[C]

Series C. Series E. Series E. There we go.

[B]

How much?

[G]

Late stage, $290 million.

[C]

Series E. Boom.

[B]

Congratulations. Triple. It's a hat Trick of series E's today on the show. Congratulations. What does that unlock? What's the next phase of Divergent accelerating.

[G]

This platform that we've built. So really taking this into scale manufacturing with a focus on defense work.

[B]

Work.

[G]

As you might remember, we started an automotive or deep technology business. We built this over a decade, but now it's really unlocking additional capacity. Capacity to make the most meaningful, frankly, weapon systems primarily in the world and to supply those to our great prime customers and to the US government directly to add over 50,000 square feet to this campus in Los Angeles. But then also to fund the build out of our future facilities across the.

[B]

U.S. how are you thinking about the footprint of that build out across the US? We've heard that famous story about there's an F35 part that's made in every single congressional district. Maybe it's coincidence. It does seem like it's important to have a balanced footprint across the US what's the courtship process like? Is it based on tax incentives or local talent? How do you think about picking a new territory to expand to?

[G]

Yeah, we think about where we can have the most impact for the nation. So where can we be closest to the customer that's making a product that we can truly be helpful with? And what is the greatest concentration of products and customers?

[B]

Where is that?

[G]

In the U.S. and that's where we want to be. And internationally we think along those same lines. But also strategically, if we're going to be near theater, where does it really make sense to have a facility? And the beauty of these factories is they are digitally defined. Right. You've got hardware that can go from making auto parts to defense parts back to back with no downtime in between. So you can be driving a capacity based model and you can make almost any part, any assembly, one factory, any product. That's our marketing tagline. Because that quite literally is what this system unlocks. And I think we're not ready to say exactly what states we're going to. We've got conversations with multiple. But it's really about the government customer and the prime customer. And of course, yes, the economics and the incentive that we can get.

[C]

Yeah, imagine, I imagine people are competing to win, win you over in terms of off, you know, trying to create the best environment for, for diversion. Is that. Is that correct?

[G]

Oh yeah, yes, absolutely. And the states get competitive. You also see the congressional side of this. You see the various stakeholders getting involved. And you want to be first as well. You want to show that your state or your exact county is on the forefront of this and they're bringing digital manufacturing, they're bringing the next generation of that to their location. So yes, we're, we're seeing thankfully a lot of demand, a lot of interest and a lot of support for these facilities.

[C]

Fantastic.

[B]

Talk about the actual integration with the customer. You said you want to be, you want to deliver value to them. What are the considerations of actually being near a customer? Is that like they're going to have forward deployed engineers in your organization? You're going to have four deployed engineers in their organization. Is it actually beneficial to be like walk across the street or is it drive across the town? Or is it like I need to be near a big airport that's a hub that everyone can fly in and out of? I'm not trying to do 20 questions to figure out where you're going. I'm actually just interested in like the business of how you, you decide where to build.

[G]

You made my job easy. You gave half the answer there. So I appreciate that. But no, it's really partially logistics as well for our customers. So if you think about one of the large planes, they're going to end up doing the integration, say of a missile system integration, meaning they're going to put the avionics in it, they're going to put the payload in it, they're going to do whatever their end of line testing procedure is. Right. And if we, as the primary structure supplier that's making that full airframe for them can be right adjacent to their general assembly facility, that's a huge win for them. Right. And that's almost obvious in how you want to architect your production process for a complex system like that. So as being adjacent to these primes.

[B]

Right.

[G]

Like walking distance or very easy from a manufacturing process, potentially even being on the campus of these primes is in our minds. And then on the government side, you look at these depot centers and sustainment and maintenance and can we be right adjacent to these bases? So if you look at Oklahoma, for example, Tinker Air base, can we be adjacent to that air base, partner with their air force sustainment organization and supply parts directly to the base from that factory at the same time that that factory could flex use some of its capacity to do work for the primes that are in that locality as well. And that's also the beauty here is no one is burdened with the full cost because this is a platform and you can share the capacity. You're not asking the government to pay for 100% of it. 100% of the time you can actually scale and flex and have elasticity to the capital and the risk associated with that capital.

[C]

Jordyn, any updates on Singer on the automotive side, specifically the supercar that you guys built?

[G]

Yeah, no. It's been an amazing few weeks for us. We've got more cars on the road now, so we did our 10th delivery. As you know, we started deliveries this year. So there's now multiple cars on the road in la. Multiple across the us. We actually have our first couple cars in the UK as well. So we've made our way onto greater European turf with America's Hypercar. So very excited to keep building that business out, deliver more cars this year and then look into actually developing variants of that hypercar. Won't announce those now and then also our next vehicle, our next model altogether. And that feedback loop, that integration with Divergent remains very, very attractive. And frankly, when the government or a customer comes and they come to this campus and they see the Divergent facility, then they walk right next door and they see the general assembly of the 21C, the hypercar, that shows them almost the relationship between us as a subcontractor and the prime as the prime integrator. Right. We're actually putting our own dollars in front of our statements there on how you can integrate for full product development.

[B]

What's the. What's the shape of the workforce going forward? Obviously, new capital comes in. You're probably hiring, you're probably always hiring, but give me some advice for young people that want to get into doing work that might still be around in 30 years.

[G]

Yeah, we truly are software and hardware. Right. We built our own software design engine, we've got our own manufacturing software, but we also build hardware, we build our 3D printers, we build our assembly cell, we actually engineer those materials. So in terms of scope within one business and amount of vertical depth in each of those pillars is really tremendous here when you're starting out, we're mostly hiring, or I was mostly looking for talent that could bring a lot to the table day one, both in terms of their connectivity, but also their knowledge and their ability to operate. We're now at a point where we're actually able to hire on some of the brightest and youngest engineers and production staff, which is an amazing position to be in somewhere where I've wanted to be for a few years. And what we look for in that sort of more entry level role is really how scalable is that person going to become? What are they already showing today that I can tell is going to make them a great learner and eventually a great operator. And then we have the individuals here that can add to their expertise that they're bringing in. And within one, two, three years, I want to see that entry level engineer have progressed to a lead engineer or have found his management track and pick one of the two. And that's what we're screening for, is individuals that are going to scale to that next step.

[B]

Fantastic. Well, congratulations on. Great to get that massive fundraising round. Always good catching up with you. Have a great rest of your day.

[C]

Good to see you.

[G]

Great to see you both.

[B]

We'll talk to you soon. And we have our next guest in the restream waiting room. Kyle Simone coming back on the show for the third time, I believe, to.

[C]

Talk about creating the world's leading Solana treasury company.

[B]

Kyle, can you hear us? How you doing?

[E]

What's going on? John, man, Good to be here. Hey, Jordan.

[C]

John, we're back.

[B]

We're back.

[C]

Great to see you.

[B]

Good to see you. Always great to you.

[C]

You've been busy.

[B]

You have been busy.

[E]

Oh, dude, have I been busy. It's been crazy.

[B]

Break it down. What's new?

[E]

Well, I spent two weeks fundraising and in two weeks we raised 1.65 billion.

[C]

Is that good? Is that good? Two weeks, 1.6 billion.

[B]

Congratulations.

[C]

I'm kidding. That's. It's hard to do. It's hard to do.

[B]

Wow.

[E]

Definitely the most hectic fundraise I've ever done in my life, but has been super, super fun.

[C]

Were you on the. Were you on the road or were you holding court just like this?

[E]

Did you zoom, zoom in my. In my office?

[C]

There you go. That's the modern, the modern roadshow. It's just zoom link.

[B]

Yeah, Zoom out. Give us the. Give us the pitch that you gave the investors.

[E]

Yeah. So, you know, we think, look, there's obviously a lot of digital asset treasury companies out there and we want to be not only the top Solana based digital asset treasury, we want to be the top digital asset treasury in the world. Our aspirations is that this is a publicly traded permanent capital vehicle and we have aspirations for this thing to become a $50 billion plus publicly traded company over the next half, five to 10 years. And we think the key to that is going to be the team. The folks I'm working with, Jump and Galaxy, are incredible. I think we have the absolute all star dream team. Not only have salon of that, but of all that's out there. We bring together that mix of deep trading and technical experience, that deep experience across Capital markets function and being a very large counterparty that trades with almost everyone in crypto and the deep VC experience in Solana Core and Solana ecosystem.

[B]

How do people think about digital asset treasury companies these days? There's one world where you're like this is a hedge fund and I'm backing a trader who's going to pick good entry and exit prices and there. And it's almost like investing in hedge fund. Then there's another which is just because of where my money is locked up for a variety of reasons. Maybe it's in 401k. I don't have access to this particular asset class. And wrapping a particular digital asset in a public ticker gives you access to that. Which are those reasonable narratives? Are there other narratives that people are talking about? Like what, what. How do people justify underwriting an investment?

[E]

Yeah, there's a bunch of reasons you might want to invest in Ford Industries. And by the way, the ticker is Ford. It's already like the car company NASDAQ listed. But I think there's a few. The first is like you have to believe that we as the sponsors of this vehicle can do interesting things to deliver increasing solar per share for our shareholders. And there's a number of things we can do to that effect. I think the, probably the most interesting is actually M and A. There's, there's probably going to be 30, maybe 40 of these digital asset treasuries out there. Not only salon related but all the other assets too.

[B]

Yeah.

[E]

And I don't think the market's going to sustain 30 or 40 of these things. And so there's going to be a ripe opportunity for M and A. I think given the caliber of our team and just the knowledge we have the banking across especially galaxy, like we're in a prime position to execute on that. The second, I think really lucrative opportunity is in actually arbitraging capital market structures. Meaning we're working on getting loans from banks, you know, hopefully in the 5%, maybe 6% range. And then we can deploy that in the salon of Defi at today probably close to 15%.

[B]

Sure.

[E]

And arbitrage that spread. That's obviously a very sexy story. Shareholders love hearing stuff like that.

[B]

Yep.

[E]

The particularly cool thing about that is also we can even use that to fund preferred share issuance. So Microstrategy or I guess strategy now pioneered this concept and they're issuing these preferred professionals and you know the problem strategy has is they are paying a 9% coupon but they have no cash flow off of their bitcoin, we can actually produce cash flow both in sole terms and in dollar terms by tapping into the native salon of staking and defi ecosystems. So it actually makes the vehicle as a whole a lot more financially congruent than something like strategy. So those are, like, the best reasons why. And look, in the long term, we're going to be capital allocators across the slot ecosystem and crypto, and in the long term is obviously a bet on our ability to do capital allocation.

[C]

Was there an alternate timeline where the groups that kind of came together into. Into Ford went and did their own treasury? Like, did you really have to wrangle people and say, like, look, fall in line or let's do this together, or what? What did that. What did putting the team in the. In the investor group look like?

[E]

Yeah, the Jump in the Galaxy guys started talking together before Multicorn got involved, and they kind of realized they were like, hey, we really need kind of a third leg of the stool. So they called me, and it was pretty natural fit. I didn't have to spend too much time thinking about it. There were some legal questions that we had to work through in terms of compliance for multicoin, but we worked through those and we said, all right, guys, let's do this. And they've been absolutely phenomenal partners.

[B]

Take me through some of the other stories that are happening in the crypto world broadly that might be bullish or bearish or something to keep an eye on from your perspective across stablecoins. There's that new Stripe project. There's a few other, like, chains that are launching L1s, L2s. Like, what do you think is worth keeping an eye on? And what's your interpretation of the news?

[E]

Yeah, I think there's a handful. So One is Solana. ETFs are coming. I think they're going to be here in the next few weeks.

[B]

Okay.

[E]

That's obviously like, a watershed moment for the rest of the industry. Bitcoin and ETH have always been kind of, like, blessed as special assets, and I think Solana is about to get that same. That same treatment. So that's super exciting.

[B]

Yeah.

[E]

The second one, this one's really in the weeds, but the team at Pump has been. They've had a livestream product for about a year now, maybe. Oh, yeah, this past weekend it really blew up.

[B]

Yeah.

[E]

Handful of creators started making tons and tons of money this past weekend, and a lot of creators from Rumble and Twitch are now publicly experimenting on the Pump platform. That could really snowball and gain momentum and really kind of change the dynamics around creator monetization. That's actually funny. They this happened over the weekend. I was at the All In Summit last week and I gave a presentation there on Internet Capital Markets. I think that video is going to drop in the next couple of days. And one of the examples I actually highlighted in the presentation was the Next Generation. I talked about Entertainment Finance and you know, Jim Cramer was Entertainment Finance 1.0.

[B]

Yeah.

[E]

And in my presentation I highlighted Roaring Kitty and Dave. Dave Portnoy as being emblematic of Entertainment Finance 2.0.

[B]

Yeah.

[E]

And I think we're about to have this just onslaught of streamers who are going to be Entertainment Finance 3.0.

[B]

Interesting.

[E]

And it looks like that's a lot of that's going to happen on Pump and all of Pump is built on Solana.

[B]

Yeah, yeah, play that out for me. I mean that's fascinating. Since we're in the live stream world, we don't give financial advice, talk about specific stock prices, but I could imagine someone comes out and does something that's much more priceless.

[C]

We once joked about somebody was like how do I invest? And TVPN and we said well LVMH of course because it's the best proxy. And then somebody immediately while we were live created Solana Token.

[B]

We had to kind of disavow it. But, but I am interested. Like do you think that there's a near term world where stocks go on chain, they get wrapped and then. And then that livestream pump ecosystem you have like a true Jim Cramer type who's like talking about public companies but has some sort of like wrapped exposure to the, to the coin that's underlyingly tied to the public company or something that is that something that could happen.

[E]

Yeah. So it's already a fairly common behavior. You'll see a lot of the crypto traders, they'll just pull up their binance screen or whatever they're trading on. They'll just live stream and trade and talk with their hands and stuff. There's never been a great way to follow along like with buying a sell button like built into the stream kind of a thing. I do think that, I mean that now exists with Pumpkin for the native token of that that stream. So that's already a thing. I think you're going to see more of that kind of streaming products continue to evolve in the pretty near future as it pertains to tokenized equities. Yes, the team the Galaxy actually announced a deal with Superstate to Tokenize Galaxy's equity on Solana. It went live a week or two ago and this is obviously something relevant for Forward Industries for us with our Ford ticker. And we're working on that. I hope to talk about it very soon. And once it's up, we have this tokenized equity, then you'll be able to kind of do that same configuration that you just described.

[C]

So what. I'm not sure this is correct. So Ford is already trading.

[E]

That's correct.

[C]

That's correct. But what is the. Is it. Is it what? Like why. Why is it trading at you guys? That saying. Is this correct? It's saying the current market cap is somewhere. It's 67 million. Is that. Is that correct? How does.

[E]

Yeah, so.

[C]

So yeah.

[E]

So we closed on the pipe transaction last week, but it hasn't.

[B]

Okay.

[C]

Yeah.

[A]

The. The.

[E]

The SEC has. So the company has issued shares these to all of the investors who came into the pipe, including Multicoin Jump Galaxy and many others. And the SEC has a process by which you have to register new shares that are created and that process takes roughly 30 days. So the Yahoo finances of the world and the other kind of major stock screeners only show registered shares.

[B]

Got it updated? Got it. Interesting.

[C]

Awesome.

[B]

What a fascinating world.

[C]

Well, you're our resident Solana expert, so.

[B]

Thanks so much for hopping on.

[C]

Anytime. Congratulations on putting this together. It's got to be up there in. In maybe, I don't know, you know, top, top 1,000 roadshows ever. The one however much you raised in a couple billion.

[B]

Two weeks. Not bad.

[C]

Definitely top. Maybe probably top five pure Zoom road shows. Anyways, great stuff, Kyle. Good to catch up.

[E]

Hey guys, last thing. We are hiring at Ford Industries. You know, we're obviously building out this permanent capital salon strategy. If you're watching the show and you're looking for a cool new role and opportunity and work with us, please get in touch. We're looking to hire awesome people.

[B]

Fantastic. Well, good luck. We'll talk to you soon.

[C]

Cheers.

[B]

Thanks so much for having on. Thanks guys. Talk to you soon.

[C]

The team they put together is pretty, pretty wild.

[B]

I haven't had a chance to dig into it. We'll have to.

[C]

Some heavy, heavy hitters.

[B]

In other news, Kimball Musk, he came on the show a few weeks ago. He posted that he's deeply honored to have entrusted with Vatican as a stage to bring grace to the world. With his dear friend Pharrell Williams, Andrea Bocelli and Jennifer Hudson, we made a miracle. You probably saw some of these photos. It's a drone show that happened over the Vatican. He says grace for the world. And some of the images and videos are. I mean, they look like AI, they look like cgi, but they are actually realized.

[C]

He was behind this.

[B]

Yes. No, this is. This is his project.

[C]

We what I was saying, imagine being the tribe in the Amazon. Kimball Musk starts flies over. That is a euphoric.

[B]

Yeah. Pretty wild. And what a fantastic, like launch strategy to go to the Vatican. Like the biggest possible. Like, it's such a striking. Everyone has seen this building because when the Pope is elected and the Pope comes out and pope's chosen, gives a speech and whenever there is news in the world of the Catholic Church, the TV cameras descend upon the Vatican, of course. And to put your product right there is remarkable. And so I'm sure the full video is out there. We'll have to watch the full thing and see. And I'd love to have Kimbal and the team on later. I think that there's probably more news. The company's growing and fantastic.

[C]

We should cap off on this post from Barely AI, which is stay safe out there, folks. Yeah, but this is trunk fan company. This is an AI startup called Inception Point. AI is for flooding podcast apps.

[B]

Yes.

[C]

They've launched 5,000 podcasts, make 3,000 episodes a week. They spend $1 to create an episode and monetize by getting programmatic ads break even on most episodes if only 20 people listen. Types of podcasts go from lowest effort, like weather reports, simple biographies, to higher effort. It created 50 AI personalities, including food expert Claire Delish, gardener and nature expert Nigel Tissel down Thistledown and Ollie Bennett, who covers offbeat sports. The company has built 184 custom AI agents. Let's give it up for AI agents using various LLMs to create the podcast and occasionally spot checks the output. A CEO, a former exec at podcast studio Wondery says she wants to make thousands more AI personalities, including an AI Oprah. Now, wouldn't you like. Wouldn't you like to create an AI Oprah? Interesting.

[B]

Closes by saying, so basically an article content farm for podcasts. Well, Trung, it's not an article content farm. It's a radically new AI podcast strategy. It's sloppy.

[C]

Put it in GPT speak.

[B]

This is actually extremely annoying. I search for biographies on the Apple podcast app every once in a while to try and get up to speed on things. And the amount of just AI generated content is overwhelming. And then I'm like, well, I could just go to ChatGPT for that. You know, if I wanted that that's what I would get. I want something.

[C]

There was a beautiful time where when you wanted to do some research on something about a topic and have and know that somebody like painstakingly spent hours, hours trying to handcraft a story.

[B]

I think that, I think that the long term outcome here is like institutions get stronger. And so now I've noticed that if I'm searching for the history of an entrepreneur, I just won't search for the name. I search founders podcast and then the name because I just know that if it's not that it's not worth listening to.

[C]

Yeah, I wonder where the article content farms like how they've done. Obviously they're getting like, you know, if I'm sure some have done quite well but are actually getting hurt by language models now that are just collecting the information and not actually hitting the site as a human. But perfectly compared. They don't have like any sort of like real technical advantage. And so on a long enough time horizon consumers might, that they might today go, I want to go and get, I'm going to go to Spotify to get the weather report from this inception point AI company or there's a world in the future where they're just, they tell Gemini, make me a podcast every morning about the weather today, tomorrow and the next day and just like have it ready for me in the morning. And that just kind of runs well.

[B]

That's kind of a Notebook LLM.

[C]

Yeah, that's.

[B]

What is that?

[C]

It's like every, every consumer AI app will be able to make a podcast.

[B]

What is crazy is that the CEO gave an interview to the Hollywood Reporter which is basically just like a massive target compete with me. Like how many startups do you think launched after this article went out? Like, oh, it's, it's easy to just.

[C]

Wait, you're trading $1, you're profitable after 20.

[B]

Great. Okay, I'll do the same thing.

[C]

It's like great classic.

[B]

This is going to be the new hustle. This is something where anyone with an.

[C]

Open source loop this is the new WAP for sure.

[B]

This is new dropshipping new which will.

[C]

Of course get competed away, destroy the opportunity.

[B]

It will. But good luck in the short term and maybe this grows into something better.

[C]

Yeah. The funny thing is if out of this they get one true hit, like if they were to make an AI personality that is, that goes on an Oprah like run, that'd be fantastic to pay for all the slop.

[B]

Yeah, I would imagine that it's got to be crazy fine tuned. Like what we're seeing with Friend. Like what we're seeing with that other, that other LLM where you're like, oh, this is, this is an AI. But it's not just the default. So you can't get this level of reporting or this level of content out of NotebookLM because it's so. So NotebookLM has a very Notebook LM flavor to it, which is good. It's default. It's information. Tyler, you have a take what you got.

[D]

You're absolutely right, John.

[B]

Thank you, Tyler. The chat was asking for more Tyler cutaways and we're working on it. Also. I love that when Lucas Zinger came on the show the chat was just series E serie. He did the meme. He did the meme. Thank you chat. You guys have been fantastic today. Great to see you all. Great to see Raghav and Bobby Dak and John Exley of course. Ilhan. So many good folks, so many names we know and love in the chat.

[C]

And on that note, have a wonderful afternoon. Yeah, we will see you tomorrow.

[B]

We will see you tomorrow.

[C]

Thank you for being with us today. Go subscribe to our newsletter brandon tvpn.substack.com very hard on it. Or you can just go to tvpn.com you can, you can subscribe there. Wherever you want. Will be in your inbox and leave us five stars on Spotify. Apple podcasts goes a long way. Remember we still have our arrangement where if you leave a review on Apple and you put an advertisement for your company in the review, we will read it out on stream as your business. So thank you to everyone for tuning in and we'll see you tomorrow.

[B]

See you all. Goodbye.