Loading summary
Transcript620 lines
- [00:00]
A
You're watching TVPN.
- [00:02]
B
Today's Wednesday, June 3, 2026. We are live from the TVPN Ultradome, the Temple of technology, the fortress of finance, the capital of capital. Wow, that's loud. Ramp.com Time is money safe.
- [00:16]
C
Both these use corporate cards.
- [00:17]
D
Go pay accounting a whole lot more
- [00:18]
B
all in one place. We're having fun today.
- [00:22]
A
I got the air horn.
- [00:23]
B
Yeah, we got a fistful air horn. And when Nick fired that thing off unexpectedly, it really did some damage. But once you're prepared for it, it'll grow on you. Satisfying. Pretty satisfying. Pretty satisfying. Anyway, Microsoft build is satisfying for a number of reasons. They're in the foundation model game. They train a bunch of models. Mai code 1, flash Mai thinking 1. The company's first coding and reasoning models respectively, which several speakers played up as super efficient on a cost per token basis. In the ROI race, you gotta be efficient. Microsoft Scout is an agent. They're OpenClaw pilled now. Powered by OpenClaw open source technology that operates across cloud, desktop and web, connecting to Teams, Outlook, OneDrive and SharePoint and to the data that powers your day including chats, emails, calendar, contacts. Good news if you're all in on the Microsoft ecosystem. And then we talked about this a little bit with the Jensen announcements from Nvidia. They're going into the PC market more. The Surface RTX Spark dev box is sort of the answer to Apple's Mac Mini Custom silicon designed for agentic AI. There's also a new Android based OS operating system designed to run agents instead of apps called Project Solara. And there's a pretty cool demo so we should play the video. The Verge always does a cut down of these keynotes. They take you through Microsoft build in 25 minutes. But we're only going to play a couple minutes of this because it's a long presentation.
- [01:56]
A
I'm really excited in order to tap
- [01:58]
E
into all this compute power is to
- [02:01]
B
expand the scope of Windows ML and Windows AI. We are also announcing two very cool
- [02:08]
A
new models that are all going to run on Windows inbox.
- [02:13]
B
Okay, let's jump to eight minutes because this is where Satya introduces Project Solara, which Ben Thompson said Project Solar is, to be very clear, vaporware at this point. Although the company did show real devices and has signed up Qualcomm and MediaTek as chip partners. It's also extremely compelling. So Ben Thompson likes it. Let's listen to Satya Nadella introduce it. Very broad categories. The first is stationary and the second is portable. The first device is designed for your desk. And it's built on MediaTek Silic concept car with hello for business. Walking up to the device securely signs you in giving you direct access to to your agents. And Amazon and Google Home have similar products at this point with screens for more smart home. You think, plan and even act by delegating tasks to your agents with a simple tap or just using your voice. It even supports experiences like handoff between devices acting as a little tricky to imagine when you wouldn't want to use your phone for for this. Since people carry their phones everywhere. Firing off an agent isn't the most cumbersome thing. But I do love consumer hardware so I'm excited to see if there's any unique things that you can do only with this product. Lightweight form factor designed for agent interactions. This is a very interesting thing. It's not a phone. It looks more like a smart key card or badge.
- [03:40]
F
Yeah.
- [03:40]
B
It even has its face on it like it's a badge that you'd wear. Tap to unlock the device and I have access now to all my agents in a secured manner. And would you look at that. I already have a task and it says gather content for your social media posts.
- [03:55]
A
Sorry if I missed it. Why? Why would you want this over Having an application on your phone?
- [04:01]
B
Camera is recording. I'm gonna pan across. I'm not sure. Yes, thank you. That's a good question. Does anyone have an answer?
- [04:12]
A
There's a new meme, the two phone meme. People feel left out. They want a second device.
- [04:18]
B
There's like the dumb phone route where you don't want everything in the phone but you do want to kick off agents that go do things for you. You want to be more like delegating instead of consuming. You're not going to be scrolling TikTok on that thing, but you might be firing off work tasks that you can come to later on your desktop and sort of lock in on.
- [04:40]
G
I don't know.
- [04:40]
A
Yeah. So I could prompt it and say write a 20,000 word message. John telling him that I would like to hang out on Saturday afternoon. Make sure it's easy enough to digest so his agent can summarize it into a few bullet points.
- [04:57]
B
Yeah, ideally just one sentence. Want to hang out? Let's see. Ben Thompson broke it down a little bit. He said first off, note the framing. The PC is old tech with agents. What about new tech uniquely enabled by agents? And note the classic Microsoft hook. Could that new tech sit on top of the new platform? So he says there was one brief moment in the promotional video that preceded his appearance that made the concept click for him. The problem with wearable devices is the interaction model. They are only useful when you are interacting with them, when the human is in the loop. But being in the loop with a wearable is annoying and inefficient. What is being demonstrated here, however, is a brief interaction and then the agent doing work in the background. In other words, the usefulness happens in the cloud without the human needing to be involved because an agent is doing work. That's what Ben Thompson finds compelling. On one hand, you can make the case that of course, Microsoft would be interested in a device model that uses the cloud as a platform, given that Microsoft doesn't control a mobile device like the iPhone. What occurs to Ben Thompson, however, is that even if Microsoft doesn't succeed with Project Solara, this model, where the cloud is the hub and multiple devices are the spoke instead of the phone being at the center is clearly a better one for agents. Agents work best in the cloud and across apps and devices. Yes, the phone might be one of those devices, but when it comes to agents, it shouldn't be the hub because it's too locked down, he says. Again, this is vaporware and very much in Microsoft's interest, so take Project Solara with an appropriate grain of salt. It's a vision of the future. However, that does make a lot of sense, particularly in an enterprise scenario where all of the context and compute is already in the cloud and Project Solara is focused on enterprise, not consumer. So you can mandate effectively that all of your employees carry these as their badges and then they have a sort of like secure on ramp to their enterprise agents in the cloud that all run in the Microsoft Azure ecosystem, in the Microsoft 365 ecosystem, he says. It's also something completely different from the past and fits Ben Thompson's thesis that in the age of AI, thin is in because the compute is so constrained to the data center on device, compute is maybe going to happen, but there's a lot that you can do on in the cloud that you can't do locally. So just have a very thin client and that little key card device is basically the thinnest client you can have. Its sole purpose is to just interface. It looks a lot like the rabbit R1 sort of that form factor which a lot of people were taking little victory laps on behalf of the founder of the rabbit R1, saying he was just a little bit too early because the actual decision to offload all the compute to the cloud do something useful up there have a very minimal device that maybe you could take out with you that could do some basic stuff. But it's always hard because people like the phone. They like being able to just watch a full movie on their phone if they want and you never know.
- [07:48]
A
Has anybody been running down making a smart AI enabled cowboy hat? Cowboy hat? You potentially have a lot of room up here for onboard some local, right? Maybe you could build a Starlink into it.
- [08:03]
B
I feel like you would not wear the smart cowboy hat. Just frying your brain with the Mac Mini on your head, you're going to be against that. There's no way you're picking that up.
- [08:13]
A
Under explored, underexplored Any hardware platform, whether
- [08:17]
B
you're bullish or bearish on Microsoft, you got to get on public.com investing for those who take it seriously. Stocks, options, bonds, crypto, treasuries and more with great customer service. Alex Heath, friend of the show, summed up the embrace of openclaw in a post Scout is what it's called at Microsoft's first proactive AI agent for Copilot buried the news that Satya Nadella is fully embracing openclaw. When Scout is released more widely this summer, it will be powered by OpenClaw, and Microsoft will contribute its security guardrails back to the project's open source ecosystem. A lot of people that got excited by openclaw sort of saw the rough edges and we saw this with the meta AI, the Instagram account theft that was going on. You can imagine that if you have something that's as powerful as openclaw but still constrained within your Microsoft ecosystem, all your cloud accounts, you could do things that are useful, pull together spreadsheets, PowerPoints, databases, all this different stuff, but not run roughshod over everything. And so if you're all in on one walled garden, the walls are actually somewhat safe.
- [09:25]
C
I don't know.
- [09:26]
B
So Microsoft getting in bed with openclaw makes a lot of sense, says Alex Heath. You only welcome a growing open framework onto your turf when you're confident you can control the ground it stands on. In this case, Microsoft is doing what it does best, being a platform company rather than trying to own too much of the stack. Microsoft also gets to ride the agent wave in a way its main hardware rival won't. Even with OpenClaw's initial buzz driving a surge in Mac Mini purchases, it's highly unlikely Apple will create a white glove experience for Microsoft for OpenClaw, like Microsoft has with Scout. And so Apple, you know you're in
- [10:00]
A
the Mac Mini, you're in the one of the primary beneficiaries of OpenClaw Boom.
- [10:05]
B
In terms of they really did sell out. They really have sold out all over the place. But.
- [10:10]
A
But not gonna embrace it. Not. Yeah, just not at all.
- [10:13]
B
They just don't have the enterprise motion necessarily.
- [10:15]
A
Well that but also the.
- [10:18]
B
Yeah but WWC is next week and who knows, maybe we will see an open claw competitor. Maybe we will see something that's, that's you know, a great leap forward for Apple in the AI and the Apple Intelligence feature set. I don't know I'm. My predictions are something that are look a lot more just like okay, Siri works now. It can do the basic shortcut integrations. It can answer questions at a near frontier level. It's running Gemini under the hood and so it's probably going to be pretty good at just answering basic questions, doing basic things on your phone. I'm not expecting it to go and like worm its way into every other app like openshaw has. So there's a few other observations that Alex Heath had from Microsoft Build Nadella is trying to tamp down the data center backlash. You know it's getting bad out there when a Mag 7 CEO is debunking water usage fears. The quote was in fact the daily water usage over the course of the entire year is roughly equival to what a single restaurant would use. So he's giving more context there. The Copilot super app is not ready. Alex Heath showed off what the new Autopilot tab with Scout looks like, but it wasn't shown on stage. So that is delayed as it rolls out to become the Copilot super app that will sit alongside all of the Microsoft apps. Microsoft is still not at the frontier according to Alex Heath. He says it didn't make a big deal out of the benchmarks for its new family of models. There were other comps they were comping it to older models from Anthropic and OpenAI. There were some obviously some great cost benefit trade offs. It's important to be in the game. Do you have more context on how these models are performing?
- [12:00]
F
Yeah, I think the interesting comparison here
- [12:02]
A
is not with Frontier Labs, but I think with Meta because they had musespark pretty recently and the thinking model like Mai Thinking one is actually quite competitive with Meta, which is interesting because I feel like I just have not heard that much about the Mai team. Like obviously they have, they have Mustafa and that's kind of like the big name at Microsoft. But like Meta you hear like over
- [12:22]
B
and over they went through all this crazy. Nat Friedman, Daniel Gross, Alex Wong. You know you have so many people that have like, you know, have done podcasts to have aura and have clout and the industry coming together. Not to mention like the actual researchers that they recruited.
- [12:37]
A
Yeah. So it's interesting that they're actually like these are pretty good models. The researchers would actually more important than.
- [12:42]
B
Yeah but in terms of like building. Building hype around whatever you're working on like Meta definitely built the aura of MSL and TBD labs. There's a whole article about Alex Wang and in the Journal or the Financial Times today. We can get to that later. But Alex, he says agents are the new os. I'm not sure if an AI access badge with a screen will do it. He's skeptical. Are you putting on the badge?
- [13:09]
A
I'll throw on the badge.
- [13:10]
B
I would throw on the badge for a little bit. I wonder how often I would actually reach for it versus a phone. Also I'm not although I love PCs and I love Windows for gaming. I've never been like a run my whole life in Outlook and Microsoft Teams and all of that. So I feel like I would get 1% of the benefit of that. You got to be all in on that ecosystem.
- [13:32]
A
Be all in. You got to pushing your chips in for sure.
- [13:36]
B
For sure. Well, we will keep tracking that. What else is going on? Ok. Live reading reacting thread from Stoke There's a ton to go through on Mai thinking was there anything interesting in this thread? So Chasm says I also have to wonder how much of this report is informed by what they know about IP OpenAI's IP. So of course they have access to intellectual property from OpenAI for training and they have access to the models. But one of the things that kept coming up in this presentation is that Microsoft is sort of touting like a very clean pre training data set so that you as a company who are using this model can be very confident that it's not going to get you in any trouble down the road. Because if the New York Times doesn't want to be in there or Harry Potter books don't want to be in there, it's not in there because Microsoft has done all the hard work to sanitize all the training data. They also made a very big point that they did not distill on another lab, which has been accusations that have been thrown around a bunch during the Elon Musk OpenAI lawsuit. Elon was on the stand and mentioned that he might have or Xai Folks might have done some distillation on either anthropic or OpenAI models at one point. And so that was sort of like not the cleanest thing and might lead to problems down the road. Microsoft saying, hey, we're getting out in front of this, there's no distillation involved at all. And then they also launched a lot of features for companies to be able to fine tune these models. Slightly different than what Amazon does. Amazon does mid training. They give you a checkpoint of the pre train and then you can add data that is relevant to your business and fine tune it from a mid training checkpoint. Microsoft is offering more of a reinforcement learning RLE post training step. But all of these lend themselves to. There is a model that has a bunch of great capabilities at the baseline and good price per performance and it runs on Azure and they've already optimized it for the systems. And then you can take it, tweak it, fine tune it and then you can deploy it on the same hardware and you know it's going to run, you know how much it's going to cost. Bertoken, you're good to go, but it's going to answer your particular questions, work for your particular business a little bit better. At least that is the pitch, that is the hope. We will see what adoption is like. Microsoft clearly has a strong go to market team, strong enterprise sales team. So we will hear how this is being deployed in the near future. Before we bring in our first guest of the show, let me tell you about CrowdStrike. Your business is AI. Their business is security. I don't know if you can actually hear that. The filter might be a little bit too much. But anyway, we can deal with that later. We are very, very fortunate to be joined by Mikey from Suno. Welcome to the show. How are you doing on a massive day? He's here live with us in the TVP in ultradome. Amazing that you could be here. How you doing?
- [16:29]
F
I'm doing great. I'm excited to be here. This place is awesome.
- [16:31]
B
Thank you. Give us the news. What happened today? Hit the gong. Hit the gong.
- [16:36]
A
Tell us what happened and then come
- [16:37]
B
back and tell us and then come back and tell. I like how it shakes the entire camera.
- [16:45]
A
Anyway, that was by far the strongest hit from a guest.
- [16:48]
B
Yeah, a lot of people come in and they don't, they, they, they don't want to break the gong. It's might mention you told me to go at it.
- [16:55]
A
I went. I know, that was perfect.
- [16:56]
B
Well, it's a good day for that,
- [16:57]
A
that's the new bar.
- [16:58]
B
So take us through. Why are we hitting the gong?
- [16:59]
F
We're hitting the gong. We raised a little over $400 million announced today. Very exciting. Led by Bond. Fantastic. A history of investing in really category defining consumer businesses which is really, really cool. Lots of new investors joining all of our old investors participating. So really, really excited about that.
- [17:18]
B
IDP4Runner Union Square Matrix. Quiet Lightspeed Manlo. Wow. Murderer's Row. Good job.
- [17:25]
A
Great.
- [17:26]
B
What unlocked this round? Why do this round? Are you spending a ton on inference? Are you spending a ton on training? Is there a huge cost or is there just so much traction that that unlocked the round? Like what's the, the thesis?
- [17:36]
F
Really, really more the latter. There's been a ton of traction and a ton of actually progress since the last time I was on. And so just to zoom out here,
- [17:46]
B
where is the company right now in terms of the products?
- [17:49]
F
So every time we release new products, we release new models, more people come to the product but more people stick around. And so the engagement metrics that we track say it's usage, retention, what fraction of our users come three or more days a week session time. All of these are getting really, really good. All of these are up about 50% in the last six months which is huge. And this is just like constant making the product better, making more people fall in love with, being creative. And so it's been incredible to see. And yeah, it's like a lot of hard work from the team but it's
- [18:22]
B
your North Star metric on retention like Dao Mao Monthly Churn. How are you thinking about actually what is the correct metric to measure for your particular business?
- [18:34]
F
You know. Yeah, different businesses are going to be different. We like retention. I think all of these are kind of very, very similar. The thing that's like really important here is just to remember, yeah like revenue is a trailing metric for how good is your product and retention is really the best way to think about. People are coming to my product over and over again. They're falling in love with it. There's no real hacking that basically.
- [18:57]
B
Yeah. And are you early or are you far enough along to see smiling curves in the retention?
- [19:04]
F
Yeah, for sure. So it's been really exciting. So also there's two retentions. There's usage retention, there's paid retention. So remember we have a pretty generous free tier and so our free users are also still coming back over and over again. So all signs point to, I think the thing that we've thought about for a long time, which is everybody's creative. Everybody will enjoy being creative if we can make the experience right for them. And we've known this deep down because when you demo the product for somebody, you take out your phone and you guide them through having that magical moment, their eyes light up and so it's really around. How do we let people do that themselves and really feel creative by themselves?
- [19:37]
B
Walk me through the paid tier philosophy. Single tier, are you thinking or do you already have ultramate pro plans where someone can spend a lot of money if they're using this as more of a prosumer tool Enterprise? How, how striated is the monetization at this point?
- [19:59]
F
It's a great question. I think the short answer is we haven't really changed our pricing all that much in a long time. And so it might not even be appropriate. But what we have now is there's a free tier, there's a $10 a month tier, there's a $30 a month tier. $30 a month tier gets you not only more usage, but some different power features. And then AI is this weird thing. It doesn't really fit neatly into SaaS pricing. And so a lot of companies are trying different things a lot like everyone else. When you reach your cap, you can buy overages. And so we've got lots of people who, they spend hours and hours a day on it and they hit their caps and they keep paying for more.
- [20:34]
B
Yeah, it's such a weird dynamic in the modern era where you will have customers that are effectively gross margin negative, potentially depending on exactly where you set the thresholds. And then you will have customers that are, they want the best product, they're happy to pay, but they come once a month with something and, and they're probably much higher gross margin. And there's just this blend because we're in this like, you know, real marginal cost era. No longer the zero marginal cost error.
- [20:59]
F
You know, a few things there. Like actually our margins are pretty okay and music models are small.
- [21:03]
B
Yeah.
- [21:04]
F
And so it's not really the same cost structure. But the other thing is, it's really interesting. There's not really a lot of consumer businesses being built right now and certainly not a lot of consumer entertainment businesses being built right now. And there's a real different, I think, interaction that the average user has when I'm sitting down at Cowork and my boss is paying for it. And so I don't really think or care about how much money this is costing me. And when you're building for consumer entertainment and you're paying for it. And it's not for doing a task. It's actually for enjoying yourself. You actually do think a lot about it. And so, a, it makes our jobs harder. We actually have to deliver more. But B, you want to make sure that you provide the right experience where you're not constantly thinking about, oh, gosh, how many credits do I have left
- [21:45]
B
before I hit the paywall? If Ben Thompson were here, he'd probably ask you about advertising. I feel like consumers want to be entertained. They don't necessarily want to be productive. It's very hard to get consumers to pay at scale. Obviously, the AI era is like a big narrative violation of that because there's tens of millions of consumers that are paying for AI tools across the entire industry. But there does seem to be a limit to how many consumers will pay even $10 a month. Are you thinking about trading credits while you're waiting for a song to generate ads? How are you thinking about that?
- [22:23]
F
We're not thinking at that scale yet. I actually kind of disagree with Ben here. I mean, he's awesome, he's brilliant. But I just think about, like, 800 million people about stream music. A little less than half of them pay for it. That's actually a lot of people. Like 300 million people pay for streaming music. That's a lot.
- [22:42]
B
I just think that, like, the history here.
- [22:44]
A
So there's billions and billions of people that don't listen to music online whatsoever. Is it really only 800 million accounts, or is that accounts on Spotify? That doesn't maybe count YouTube.
- [22:56]
F
That's to my knowledge. That's like Spotify, YouTube, Apple, and maybe like five others.
- [23:01]
B
There's like a billion vinyl record daus now too.
- [23:05]
F
Vinyl's gonna make a big comeback. And I mean, I think as we change the place of music and culture, as the product gets better, we change the place that music has in culture. We see these creative moments happening. And so you see them digitally now, where maybe you saw the Puerto Rico song, or you saw an amazing creative trend where people are making, like, their cringy text messages into songs and they go viral on TikTok. And I think you'll start to see that penetrate actually the IRL world as well. And so I'm long vinyl.
- [23:30]
B
Yeah.
- [23:32]
A
I was talking to someone who wanted a job at Suno, and they were asking me what I thought about the business, and obviously I've been bullish for a long time. Like this new entertainment form factor. People love the product. You guys have real revenue. You're growing really quickly. My unsophisticated view was that I told the person, I can easily imagine this as like a $10 billion company. It's hard for me to see the world in which this is a hundred billion dollar company. Why am I wrong? I know you're insanely ambitious. You're not going to 2x from here and then be like, job's done, everybody.
- [24:12]
F
No, the job's not done. I think this is one of those things, actually. I think you've just been hanging around investors too much, which is the thing that you have to be right for a long time while a lot of people are wrong. And the thing is, every revenue scale we hit, every usage scale we hit, the question is, but how many people really want to create? And that was we got those questions at $100,000 in revenue. We get those questions at $100 million in revenue. And the lived experience we have playing with the product, with people, but also the metrics and looking at our growth curves and seeing that every time we make the product better, yes, things inflect up, but we appeal to a broader and broader audience. And so I feel very.
- [24:49]
A
Yeah, I mean, the same thing would have been said with early creative tools on the Internet.
- [24:54]
F
Absolutely.
- [24:55]
A
It was how big with Figma, it was always, how big could Figma be? You know, there's only so many designer, professional designers then of course, like the rest of the organization. Same thing with Canva. I'm sure a lot of people passed early on because they just thought like, yeah, there's just not that many people relatively that do design. And then of course you make the tools more accessible.
- [25:18]
F
So anyways, yeah, so there's a couple things here. One is I think actually we'll have a very large fraction of those 300 million subscribers to streaming. Also subscribe to Suno in the long run. But the other thing that's amazing is that for the first time in a long time in music, there's innovation on the end user experience. And so you can have something where there are multiple products out there. Like right now, the end user experiences for music all look the same. They all look like a streaming provider. Right. It's just like I have all of the world's music living on my phone and I can listen to it and that's it. In the future, when you can do more things and play with it, you probably subscribe to multiple video games, you probably subscribe to multiple video streaming platforms. Why can't you subscribe to multiple music things? And I think the answer is in 10 years you will?
- [26:01]
B
Yeah.
- [26:01]
A
Do you think Apple would ever make a play in this space?
- [26:05]
F
I think lots of people are going to make plays in this space. It's almost validating to see that people finally think of music as a growth area again and so. But again, I think that there's so much room for innovation here. There's so much Greenfield to build upon that I'm actually really excited.
- [26:20]
A
Yeah.
- [26:21]
B
Can you take me through the story of the Puerto Rico song? What happened there? Like, what were the key inflection points?
- [26:27]
F
Actually, I don't know is the answer. You know, this is like a song from a creator, and it hits the right cultural moment and the right vibe. And I don't want to say it's a good song or it's a bad song, but it is a platform where so many people can be creative on top of it, and that is really fulfilling to see. And that stuff that it's like new use cases for music that haven't been around for a minute.
- [26:48]
A
Is AI music becoming less controversial, or are there still people discovering it and getting angry? Because I could imagine there was a wave over the last two years where people were like, wait, they're making music with AI? This is meant to be a human creative pursuit, or whatever criticisms they've thrown at it. But I imagine over time, people start to realize that these are fun entertainment products and tools. And before you form an opinion, you should just try using it and form an opinion based on your experience, not Internet comments or things like that. But my expectation is that it will actually become less controversial year over year, as it sort of normalizes, it is
- [27:31]
F
definitely becoming less controversial. There are still plenty of people who have preformed opinions, and then actually playing with the product is the best way to. To change their minds. The thing that has become apparent in the music industry is that so many people are using it without talking about it. And as people realize that everybody knows that everybody knows, it certainly normalizes there. And then as a consumer, either when you play with the product and you realize how fun it is, or when you realize actually there's some AI music in the music that I love, all of a sudden you're not so anti AI. I think this is a lot like people being anti AI. And then they realize that ChatGPT can help them with their homework, and they
- [28:06]
B
kind of soften a little bit as a useful tool. I mean, Martin Scorsese came out yesterday in a Black Forest Labs video talking about the value of AI in the filmmaking process. But the video demos we'll Watch it later on the show. But it demos sort of like a pre visualization step, almost like a storyboarding step. He's describing this alpine town and it's generating visuals for him. He's not taking it end to end. Of course, that's happening more and more. But I think having a filmmaker like Scorsese kind of step up and say like, yeah, this is useful, I can see this working.
- [28:44]
A
Absolutely.
- [28:44]
F
There's an analog in music which is that you'll have songwriters make beautiful songs and instead of making demos for an artist, it can make low fidelity songs like low fidelity productions, which are demos with Suno so much easier. And it will get ripped out and produced without AI or with minimal AI when it's produced for the last time.
- [29:04]
A
Do you think there's still an opportunity for major artists to be a first mover in the space? We were talking, I think a mutual friend of the three of us, and I was telling him he loves Suno, he uses it all the time. And I was telling him, like, why don't you just be the first one to take a proper leap and do this? And the first person that goes and says, hey, this is okay. It's not so scary. I used it in the creative process. We still use a bunch of, you know, traditional producers, songwriters, etc. I think that person will get a lot of like, credit and potentially, you know, a lot of attention will obviously be polarizing. But then the fourth, fifth, sixth mover won't really get any potential benefit. It'll just be normalized by that point.
- [29:53]
F
I think that's totally possible. Like we've, you know, there are people, there are a few people who are outspoken about it, but I don't know if we've had like the moment yet.
- [30:01]
A
I want like the project. Yeah, somebody has to make like a hit record.
- [30:05]
F
There have been, I think what, what
- [30:07]
A
you're gonna, they have to, they have to go out and say. They have to be just very upfront. And I, what I want is them to document the whole process. Like I want a video camera.
- [30:16]
F
How much is enough Suno in there?
- [30:18]
A
Actually, that's, that's the whole point. It's like, like a guitar might have only been like one little riff in a song and it could have been some originally. Like somebody recorded it on their phone and then they brought it into the studio and then like maybe it got transformed and made with a different machine. Yeah, exactly. And so to me, to me there just needs to be. It's always going to be a spectrum right there's going to be songs that are heavily Suno, but I would expect for like, major artists, even if, like the original kind of idea or some element of the song was made on Suno, like you said, it'll just get transformed and adapted and evolved. And so I think it's just positioning it as not like a replacement for anything, but another kind of individual artist in the room, another instrument in the room that's contributing to the project.
- [31:11]
F
Here's how I think of it. We haven't hit that moment, but at some point, somebody will really document and celebrate the use of these tools in making things. Like, we're already in a lot of very popular music, but at some point, somebody will really do that actually proactively. And that will be a huge cultural moment. And then the next amazing cultural moment will be an artist releases an album that is meant to be sunoed by their fans. And this is the new format. This is the thing that will pull people in, that deepens the artist fan connection, and that will be a very, very important moment.
- [31:43]
B
How spiky is Suno's intelligence? I want to know about the training
- [31:48]
A
process and we know you have to leave. So final question.
- [31:52]
B
Maybe that's the last question. But there's this famous story about, I think Andrej Karpathy told it about GPT 441 some iteration where at a certain checkpoint, it was terrible at chess. And then all of the AI researchers just were really interested in chess, and they were like, there's no reason it should be bad at chess. And so they added a bunch of chess training data. And then all of a sudden the models got pretty good at chess. And it seems like we're in this spiky intelligence race where coding is really valuable. There's a lot of money there. So the models are getting better at coding. But if you look at the models, it doesn't really feel like they're getting that much better at copywriting right now because that's not where the boom is. They're getting really good at coding. And I'm wondering if you're seeing, like, okay, there's an opportunity for us to nail jazz in the next iteration or us to nail Edmund in this iteration. What fingerprints is the team leaving on the model versus just we're scaling up and it's getting better at everything and there's surprises and it's very indeterminate.
- [32:54]
F
We actually try not to leave our fingerprints on it. And we have like, the approach is we should not be the arbiters of good and bad music. You know like a file corrupt or low quality sound recording. But like you can go and you can listen to a song that has 10,000 YouTube plays or 10 million YouTube plays and you won't actually be able to tell the difference. And so we should not be the arbiters of that. The thing that's different here also is that in the chess and the coding examples there are objective right answers. And that's why the models are so good at getting spiky there. Yeah, Music has no right answers.
- [33:26]
B
There's no verified rewards, not at all verifiable rewards.
- [33:28]
F
And so all of those techniques actually don't work nearly speed.
- [33:32]
B
Yeah, you can't do rlvr.
- [33:33]
F
So we're actually in the regime where it's much less spiky. And actually when we find a hole, an anti spike, then we actually try to fill that in. It's not like we're amazing at jazz. You suck at jazz or you suck at country. And now we have to go and figure that one out.
- [33:43]
B
Okay, last question. We'll let you get out of here. What about adding on to the viral fuel? When I hear about turn your text messages into a Suno song, I think why can't I just screenshot my text messages and then you edit the video for me and I get a viral format that feels like very tool, you see, very like each video format is going to be a whole process for you. It's gonna be hard to like one shot the full end to end. But what do you think about it?
- [34:12]
A
I want personally, I want the slack integration. I want a sunosong at the end of each day.
- [34:16]
B
Yeah. Why weren't you on the stage at Microsoft build.
- [34:19]
A
I want the workday at the office summarized.
- [34:22]
B
You want a workday integration?
- [34:23]
A
No, I want the slack summarized. And I want to be able to listen to an hour long song on my drive home about what's happening at work.
- [34:31]
F
We'll try to make that happen. We're not doing a workday integration. But I will say the thing that's amazing is actually this is a platform where people are creative on top of it, of course. And so we're going to come up with the right cringy text message video. Somebody is going to do that.
- [34:44]
B
Well, that's a challenge for you at home. Go make the cringy text message. Go make the workday integration. Prove Mikey wrong. He says he won't build it. You can, you can go and build, you can go and build the workday integration yourself. Well, I want to react to some Timer.
- [35:00]
A
Imagine keeping up at work through music.
- [35:04]
B
John, be fantastic.
- [35:05]
A
What's more powerful than that?
- [35:06]
B
You know what's more powerful than that? Console.com console builds AI agents that automate 70% of it. HR and finance support, giving employees instant resolution for access requests and password resets. So we can talk about. There's a bunch of funny posts in the timeline we gotta go through before our next guest, Samir from Colin Samir joins us to talk about Hollywood. You can tell when a coworker got nothing going on outside of work. Why is that in here? 220,000 people like this?
- [35:39]
A
Is this. Did you put this in to kind of subtweet me? Tyler, I didn't send this in. The team always asks Monday morning, what
- [35:48]
B
would you do this weekend? 6,000,600.
- [35:49]
A
I always have the same answer.
- [35:50]
B
Million views. Anyway, there's some news on the timeline about GOP Rep and Michigan Senate candidate Mike Rogers. There's an actual photo from the campaign trail and then a photo put out by the staffer. Let's compare these. It's one of those. Can you tell? Can you spot the differences? Video. They made him a little bit bulkier. They just added just a couple pounds of muscle.
- [36:15]
A
Just a couple.
- [36:17]
B
He looks huge in this photo.
- [36:20]
A
I mean, he looks like he's ready for the Olympia.
- [36:23]
B
Yeah, I don't think they went far enough. So I took it upon myself to generate an even more accurate photo we can pull up because I think this is what he should look like. If you're gonna AI your photos, just go full.
- [36:36]
A
Just go stringer.
- [36:37]
B
Yeah, get yourself in the stringer and make sure you're looking like you walked off of the Olympia stage and still claim natural, absolute mass monster. Yeah, you can.
- [36:47]
A
A typical politician thing to do would be to come.
- [36:50]
B
I'm not on performance editing drugs. I just run every single photo through Images too. And ChatGPT people are having fun. It was funny. I thought I was so original for doing this. But if you scroll down the comments, everyone had the same idea and made him way, way more jacked. If you scroll down, there's 25 others
- [37:06]
A
and there are some says enhanced politics. Jack says the enhanced Senate.
- [37:12]
B
Yeah, there are some that are absolutely crazy here. The Enhanced Senate. Okay. We got a debate on our hands. We got a debate on our hands. So Philip Johnston, friend of the show, founder of Star Cloud, says founders, before you go out and raise, ask all the best AI models this play the role of a venture capitalist and give each startup in the XYZ category of your startup space. A ranking from 1 to 10 on how much they would want to invest in them with 1 don't want to invest at all and 10 desperately want to invest. Do deep research and give reasons. All the VCs are doing some version of this. He's claiming that the VCs are just AI psychosis, letting AI do their job. Maybe it's true, we don't know. Maybe screw it up your funds. He says all the VCs are doing some version of this and if you're not at the top of the list, you're going to have problems. So it might make sense to delay the raise until you get closer to the top. So if, if the frontier models don't approve of your startup pitch, take, go back to the drawing board, build a little bit more before you go out on that fundraising journey. Seth Bannon at 50 years says founders do not listen to this. You should in no way time your raise based on how an LLM ranks you. The best VCs are not sitting around asking AI, who should I invest in? And then doing that. Well, that's a different question. Are you trying to raise or are you trying to raise from the best VCs? These are two different things.
- [38:29]
A
Yeah, I think the LLM would do a pretty good job of figuring out like the relative interest and excitement. Right? Yeah, you can look at, you could if an agent was running. A pretty interesting metric is like, you know, headcount growth on LinkedIn. Right. Are people leaving, you know how many people are leaving the company? Yeah, all these things. So I actually think it has kept
- [38:58]
B
pretty good at funding. Has this fund's competitor done another deal and they're excluded from, and that competitor is excluded from year round, but they're
- [39:07]
A
looking for some category. Yeah, Also, also, you know, Pedigree, right?
- [39:12]
B
Yeah.
- [39:12]
A
This is whether you like it or not, a big factor in especially like king making right now. So if you're in a category and you're, you're a smart individual, talented individual, but you're at the seed stage and then there's another seed stage.
- [39:29]
B
Pet food company Pedigree Pet foods. No, that's what I know of a
- [39:33]
A
pet food, dog food and treats. And you have another company, your same stage where, you know, all three founders just came out of SpaceX after being there for 10 years and you're both competing, you know, in some adjacent category. Right. Like I actually think the model is going to do a pretty good job and say like, hey, VCs very likely might be three times more excited to invest in this team than you. And that's just reality.
- [40:05]
B
Okay, so Philip Johnson says, I'm glad I provoked a reaction. This will age badly over the coming few years. He's saying Seth is wrong. He says also, I'm not talking about timing arrays based on AI opinion. I'm talking about actually improving your startup. Get more customers, for example, which will at some point be picked up in the AI rankings, which is a lot of what you were getting at. So I guess the question is, should you delay a raise based on AI opinion and would you actually listen to a clanker tell you how your business is doing or are you going to say I know better? What do you think? I think you're arguing it's a good source of feedback, it's worthwhile. But at the same time I can't see you taking, doing this. I can't see you actually going to a model and being like yeah, they're right. I am washed. No way. No way. Zero shot.
- [41:02]
A
Like I think when I was, I think I was originally, I think that earlier in my, a decade ago this actually would have been super helpful.
- [41:11]
B
Yeah.
- [41:11]
A
Because I wasn't fully aware of all the different factors that made, that went into, that went into making a company.
- [41:18]
B
This is a good point.
- [41:18]
A
Yeah.
- [41:20]
B
Early on like there was like a checklist from zero to one Peter Thiel and you could go through it and say like okay, do I have defensibility? Am I building something that will have staying power? How many of these boxes do I check? Because when I go out to raise a lot of the VCs are going to be thinking the same thing. It's worthwhile and an AI can sort of help you do that checklist. So that's valuable.
- [41:40]
A
Another thing I was talking, I don't think you're taking no an LLM. A smart enough model would be able to do this if it was trained on an llc. VC blog posts maybe. Which certainly I think some are quite well trained on PG's blog and others. But I was talking to a founder yesterday. Two year old company has gone through a pivot. Figured out a much more interesting opportunity than they originally started with. Have a bunch of pilots, a lot of positive momentum. Revenue is growing. But I had to tell the founder like look, you are the VCs that you're going to talk to if you go out to raise right now. Are, are, are, are talking to companies that are adding what the revenue you will add in the next year, they're going to add it in the next 30 days. So even though you Have a super exciting opportunity. It's going to be really hard to invest right now just because there's, they have a limited amount of capital they want to be making bets against, you know, oftentimes momentum and that's just, that's just the game.
- [42:44]
B
Yeah. And a really, really good model, a really good model could say, look, your startup is very mid, don't even bother pitching founders fund top tier firms. But I got some guys who, they got crazy as it goes this right now. They'd love to invest. So you should definitely DM this person because they've been posting crazy slop on the timeline non stop and they're just ripping checks at anything and so you fit that category. So go get the, go get the bag from them. This is possible. This is possible with super intelligence. Anyway, our next guest is in the waiting room. But first let me tell you about the New York Stock Exchange. Want to change the world, raise capital at the New York Stock Exchange. And we are very lucky to be joined by Samir from Kyle and Samir in the TVPN Eltra. How are you doing?
- [43:34]
A
Cabin in the woods.
- [43:36]
D
That's right.
- [43:38]
B
I missed you. Can you see the view? Yes, I want to see the video. This looks amazing.
- [43:43]
A
Take me home. Take me home.
- [43:46]
B
How off the record is this conference? Can you describe what you're actually doing?
- [43:50]
D
Never. Whenever you come to Montana, you can't tell anyone.
- [43:53]
B
Yeah, that makes sense.
- [43:53]
D
I mean I did, I did just tell you I'm in Montana, but no, I'm actually.
- [43:56]
B
Geogasser knows exactly where you are. Let's get Geo Rainbow on next and we'll dox you fully.
- [44:03]
A
What happens in the woods stays in the woods.
- [44:06]
B
I imagine, I imagine you're, you're there for good reason. Hopefully there's been someone there that has provided more context on a story that we've been hunting around.
- [44:16]
A
First, thank you again for hosting us at your event last week.
- [44:19]
B
Press Publish LA was amazing.
- [44:22]
A
Incredible.
- [44:22]
B
It was amazing.
- [44:24]
D
Thank you guys for coming. Also, extremely relevant weekend, you know, to have that event. I mean we knew backrooms was coming out the next day, but obviously we'll get into that. What a time for Hollywood and creators.
- [44:37]
B
Yeah, yeah. And yeah, funny because I was telling you, yeah, the markiplier story, so cool. And you're like, I just talked to him on the stage and so obviously you've been tracking this for basically your entire career. You've seen this coming. But has it felt overdue to you? Ben Thompson was taking a victory lap, saying he sort of predicted this in 2017. It's been a decade, and it feels like this intersection of YouTube and Hollywood. It's been very combative. But my. My takeaway is that this feels like the first moment when we're seeing glimmers of. Of real cooperation between the two separate communities and a way that traditional Hollywood can work with YouTubers in a way that's mutually beneficial within the current system. But how are you processing this? How overdue is this? Like, how have you been processing the current moment?
- [45:25]
D
I don't know if it's overdue or if it's just happening as it's unfolding, like, in real time, where, you know, Cain, when I started on YouTube, Kane was five years old. So, like, you know, these are Internet native overnight filmmakers.
- [45:39]
B
Yeah.
- [45:40]
D
And I think there's three reasons why it's happening right now. One really important thing when we think about backrooms is backrooms. You guys talked about this Backroom started as a 4chan image that then became like a collaborative story. I think the reality is Gen Z IP is not legacy IP like the Mandalorian. It's IP that is collaboratively created on the Internet. So backrooms feels like something that everybody on the Internet had a hand in creating, or millions and millions of people on the Internet had a hand in creating. So when that comes out in theaters, I want a thing that I had a hand in. I think the other thing is that cutting your teeth as an Internet filmmaker, it teaches you a lot about how to earn attention. Curry Barker, the director of Obsession, talked about this in an interview. They said, what did YouTube teach you about directing movies? He said, the reality is when you make something for the Internet, the audience is bad, begging, and you have to convince them to stay. And I think historically in Hollywood, there's been an entitlement of, hey, we have George Clooney in the movie. We spent a lot on marketing. You're gonna buy a ticket and you're just gonna sit and enjoy it. But I think Internet filmmakers know when you're making something for YouTube that, like, the audience can click away at any moment to something more interesting. So I better tell you a great story that's unfolding frame by frame.
- [46:59]
B
Yeah. I think the third thing is.
- [47:02]
D
Third thing, because I know we all love lists. And the third thing will blow your mind, by the way.
- [47:08]
A
It'll shock you.
- [47:10]
D
The third thing is that Hollywood talent is, I actually think, like, the linchpin in this moment in time, that there's a lot of talent that's available that are like craftspeople in Hollywood who have focused on specific crafts, whether those are DPs or screenwriters, or set design or set designers. Like, when that is paired with the acumen of Internet storytelling, it's proven to be very powerful. Powerful. So I, I think there's been this, like, narrative. Of course Hollywood is dying, but I think it just looks very different now. Like, we didn't know theaters were available. Like Markiplier put his movie himself. He picked up the phone and put his movie in 4,600 theaters. Theaters are available and they're, they're great. You know, like Hollywood talent is available. They don't have other jobs and they're great. So I think these worlds finally coming together. There's been a lot that's happened, but it's super exciting to see.
- [48:03]
A
Yeah. I look at it as like the last 15 years has been a story. There was collaboration happening, but it was overwhelmingly a disruption story. You look at the market cap, the value creation of the platforms, the YouTubes, the Instagrams, et cetera, how much the explosion of enterprise value in these new platforms, the overall reduction in EV in some of these sort of like more legacy entertainment. But now we're meeting at a point where there can. Both sides can say, like, hey, now is the opportunity to figure out how to both win. Because clearly there's a place for both. Right. We still want to see the. I mean, I don't, but John would like to see that. The blockbuster.
- [48:51]
B
Jordy's going to see movies you're going to see. This is going to be the tipping point because you were never a movie guy but one. Once the new creators, once there's a Wombo film and a whole cinematic universe around Professor Sendy, you're going to be.
- [49:04]
A
I'll be there. But yeah. So this is the moment where I think the. I expect digital platforms to consume more and more and more and more attention. That just seems to be. They're frictionless. Right. They're free, they're available everywhere. But now is a moment of, hey, both sides should be actively trying to do more and more stuff together.
- [49:29]
B
Yeah.
- [49:29]
D
The reality is also, it's important to note that this has all happened in horror.
- [49:32]
E
Right.
- [49:32]
D
And this has happened once before with the Rocka Rocka brothers signing with a 24 and making talk to me for about 4 million bucks that ended up making just about 100 million. These are outsized returns, like Curry Barker making something for a million dollars. Obsession was made for about a million dollars. Focus purchased it for around 15 million and then generating $148 million, not even a month in is absolutely insane. And so horror has historically been cheaper to produce.
- [50:02]
B
Right.
- [50:02]
D
If you even remember in the 90s, the Blair Witch Project was like this cheap indie horror movie that ended up exploding. And I think there's something to explore about the absurdity of horror and the absurdity of these movies and the absurdity of our world right now. And maybe the comfort in going to see something that is more absurd than the world around you. That's just a hypothesis I have around horror. But horror has historically been cheap to make. I think we're about to see every agent in Hollywood ping their creator talent and say, do you have a horror script? Because it is that moment that's gonna happen. There will be likely an overreaction to this in Hollywood. But I do think what was proven over this past month is that Internet native storytellers can tell a story that reaches millions of people, that fills theaters and potentially better than and the storytellers that cost you in the hundreds of millions of dollars. The Mandalorian cost $265 million all in and it's made 247 million right now. So it's like it's super hard to make a big blockbuster movie like that. And even Markiplier making something like he made independently for under $5 million is pretty incredible.
- [51:18]
B
Okay, talk about the state of horror on YouTube or fiction on YouTube. It's somewhat interesting to me that I would put backrooms like the actual content as it's horror and the movie is a direct adaptation of that. The original backroom series on YouTube is creepy, but Obsession, Curry, Barker, that was comedy and Markiplier was horror. But video gaming, he wasn't creating the fiction himself. And so I'm wondering about how if we will see a shift. Basically, does this give YouTubers permission to do things that are more fiction based versus explainer videos, podcast, interviews, reactions, commentary, which is like the bread and butter of most YouTube comedy as well. Or will we see more people pulling, okay, this person's talented, but horror is ROI positive, so. So we should just give them a crack at this even though they're a commentary channel as opposed to someone coming through. Okay, they've been telling love stories in a fictional context. They have a rom com series on YouTube, let's let them make a rom com.
- [52:29]
D
I think animation will be the other space outside of horror and it's already happening. The amazing digital circus is a great example from an animation studio called Glitch. Their first few episodes crossed half a billion views of Amazing digital circus. They didn't know that was going to happen, but it did. And a community gathered around them. I think horror and animation and anything video game adjacent works really well on YouTube. I think some of the other stuff is harder. I think that stuff may emerge more on Instagram and TikTok and whether it can be adapted into feature length stuff I think will be up for debate. But I do think both things will happen. I think YouTube will be looked at now as a real incubator of IP and a real incubator of talent. And I think that's always been like a concept, but it was never like, like I think when you meet with Hollywood execs they're like, yeah, but we can't make any money off of it, right? Like yeah, they can make money through brand deals and a couple million bucks here and there, but like we can't make real money. These movies have proved you can make real money. And so I think that's where both things will be true. I think creators who upload to the platform and you know, incubate IP backrooms was found by a 27 year old basically staffer who was just poking around and seeing what was going on on YouTube and Reddit. I think that will be like, everybody will have a young person just perusing Reddit and YouTube to go, what's catching? And I think if I was giving advice to those people, I would say really take a hard look at Reddit and see where things are being built in real time by big communities. Another great example of this is like in the Minecraft community years and years ago, the dream SMP was this like collaborative Minecraft story that had so much lore to it that a bunch of
- [54:15]
C
people were involved in.
- [54:16]
D
So I think the reason why video games is such a ripe place for IP to build is because it's interactive storytelling and interactive world building that a lot of people are involved in and that then more than just a like large audience that guarantees like a mobile audience because they're invested in it.
- [54:34]
B
And then there's Red versus Blue with Halo and, and gtarp. Like there are rich sources. And it's interesting, there's been so many video game adaptation flops. Basically Detective Pikachu is the only one I think that has a Metacritic score above 80. Last of Us did well, so it feels like we're hitting a tipping point there. But you go back to the original Doom adaptation. A lot of the Resident Evil films, some of them made money, but none of them were critically acclaimed or really became like Cultural moments in the way the video games did. But if you have a creator who actually really deeply understands the community, the content, the lore, and then they build their audience on top of the ip potentially some of these video game companies that are maybe not as Internet, not as linear storytelling native, if they have someone in the mix who's already sort of pre vetted this can work as a two hour film, then they can unleash them, turn them loose. But of course there's an IP negotiation.
- [55:34]
A
Is there any missing part of the stack right now to meet this moment? You have the, you know, the CAA is the talent, you know, the talent agencies that can, you know, wear a lot of different hats and help make plays. You have production companies, but is there any, is there any part, is there like financing component missing? Maybe it's too risky to create some type of vehicle just to back creator led films. But at the same time maybe having domain expertise would be actually, I think
- [56:11]
D
the missing piece and this has been historic in our industry is development. I think there is a lot of assumption that creators with big audiences know what works. And part of that is true. But you know, Kane got signed at 16 years old and a 24 kind of mentored him from the age of 17 to the age of 20. And I think 824 has a history of making great movies that are culturally relevant and knowing how to make something iconic, knowing how to make us all feel something. And so that I think should not be overlooked is the development step. Like surround me. If I'm a filmmaker, surround me with people who have done this before. Help me write the script, which Cain did have a lot of help in screenwriting. He's looking for a collaborator for part two, but give me the development support to make it long lasting IP or big global ip. And I think that happened with Kane. I think Curry is like a really great talent and what he's done in comedy historically has led him here. But I think development is going to be the missing step because what's going to happen is it's going to be a very quick reaction. Let's get creators with scripts, let's put money behind them, let's move fast. So we can also make $150 million. But you know, you have to remember this is years and years and years of these storytellers cutting their teeth and having good people around them to help them learn how to tell a good story. So I think the financing will not be a problem because you can finance creators know how to make stuff for
- [57:34]
B
cheap
- [57:36]
A
curry Last question.
- [57:38]
D
20 days that curry made that movie, which is crazy.
- [57:43]
A
Somewhere out there someone is sitting in a dark room for like 18 hours a day trying to prompt their way to a hit movie. I don't know if, I don't know if the models are good enough yet. I have a feeling they might be. And it really is about figuring out the storytelling component and piecing it all together to be cohesive. But do you have any sort of personal timeline to that kind of moment where some one individual takes you know, an idea through a, you know, even a 30 minute film that actually gets a, you know, positive reaction from critics or you know, something like 20 million views, something like that?
- [58:29]
D
I mean I, you know, obviously we all live in la. I feel like, like the first thing that comes to mind for me is the Spencer Pratt Mayoral ads. Because those are like a lot
- [58:39]
H
ads
- [58:40]
A
are so much easier to make.
- [58:41]
B
It's a glimmer.
- [58:43]
D
They're much easier to make. But I'm saying that's where we are in the timeline.
- [58:47]
G
Right.
- [58:48]
D
Those ads are clearly AI, but they got shared a lot and they did do millions of views and people just accepted them for the story that he was telling, I think. And so I think if you look at that, maybe it'll be in a year or two and I think it'll start in animation with short films because we'll accept that more. But I don't know all things. I think another element of this whole story is I do think maybe you guys feel this, maybe you don't. I think there's a bit of Internet fatigue that is allowing theaters to come back. I think maybe I'm just overstepping there, but I do feel it.
- [59:24]
B
I personally feel Gen Z screen time finally sort of dropping off a little
- [59:29]
A
bit after spiking like running over my phone with my car on purpose just to have a little bit of forced time offline.
- [59:39]
D
So like the models are really good. Like we even playing around with Gemini Omni, obviously we had them at Press Publish la, but like they're really good models. They're crazy good. You know, whether you can. That'll probably happen in the next two years.
- [59:52]
B
We'll probably watch something like that One somewhat related cut on this that I want your reaction to is I found it very interesting that backrooms started as a purely CGI VFX effort in Blender and After Effects. And what's interesting is there's actually a whole discord for modeling different 3D models that are shared so the community can build around it. And the assumption if you are worried about the relentless march of technology is like, well, of course they had everything modeled. They already had the Blender assets. Like, just hand that off to the VFX team and you got yourself a CGI movie. Put Chiwetel EJA 4 on a green screen, and it's done. But they didn't do that. They built 30,000 square feet of physical set design. And it's interesting that I feel like the CGI wound up being like, the audition for actually building and underwriting the build out of that set. And I imagine that that might happen in AI where you get someone that. That has some interesting idea that they were able to prompt, and then they take that into a CGI that's a little more polished, and then they can go build the actual sets. And it's all just this extra underwriting process. But I was wondering if there was anything else that stuck out to you about the fact that they sort of went light on cgi. When you think the fans of backrooms, they're already Blender. They love Blender. Like, they're not upset about CGI, but
- [61:20]
D
if it was CGI, then I would be like, just make it for YouTube. Like, the reason why it's an event is because of all that. And I think 24. And, you know, I don't know who. Yeah, I don't know who. Whose call it was, but a 24 is very good at making things iconic. All the interviews of the cast are done in the back rooms. There was a DJ set from the back rooms. There was people who visited the back rooms. Like, the fact that it was tactile.
- [61:45]
B
I'm not even 360 experience. I don't like that out.
- [61:47]
A
I've seen five movies, but I can tell when. I can tell just based on. With hardly paying any attention. I can tell when A movie's an A24 movie from, like, the first video that I see. The first 15 seconds of a trailer, there's just something about it.
- [62:05]
D
Yeah, it's the same with you guys. Like, you can, like, there's, you know, countless shows that try and be tpbn, but you can see tp. I could, like, see something for a split second even without you guys on screen and know if it's TPBN. And I think A24 has that, and that is really hard to do. It takes a lot of focus and it takes a lot of discipline. But, you know, I think A24 has done it again. Like, this is. This is incredible. And. And pulling someone like Kane into this and Beating Marty supreme, which had, like, a crazy marketing push.
- [62:39]
B
Huge marketing.
- [62:39]
D
Shows you the power of the Internet.
- [62:41]
B
Yeah, yeah, yeah.
- [62:42]
A
They overpowered the blimp.
- [62:44]
D
Powerful, man. The Internet's powerful and the communities are powerful.
- [62:49]
B
Okay, last question.
- [62:50]
D
I'm assuming we're about to leave, and I just wanted to say I'm thrilled to see ads back. I couldn't tell you how great this looks to me. I'll give it up for that. I'm thrilled.
- [62:59]
A
It really is part of the back ads.
- [63:01]
D
It's part of the brand.
- [63:02]
B
It is part of the brand. Do you have time for one more question?
- [63:06]
D
I got time, man. I'm just here in Montana.
- [63:08]
B
You guys have to go.
- [63:09]
A
Probably doing mysterious things.
- [63:11]
B
Yes. Next question. Where are you? And don't dodge the question. I want advice for YouTubers in the context of passion projects. There's a lot of YouTubers that have grown off of a single format. They know what works. The title, thumbnail hook, the format, the structure of the video. They know they can put up a million views for every video that they do in that series. But they get sort of bored and they want to do their passion project, their series, and then they do it. They take the time off, and then it just doesn't quite perform. Maybe the 200,000 true fans show up, but it's not getting the viral hooks. I noticed this with Johnny Harris, did this fantastic series on Cyprus, and he went around these islands, but it just. It wasn't financially the same as all the other videos that he does. And I'm wondering if there's a new format where you should sort of give YouTubers the permission to keep that special project in their back pocket and take it to a different place, take it to Hollywood, where that type of project can live and flourish and go through development as opposed to trying to stuff it down the same funnel that is very optimized for a particular type of video.
- [64:24]
D
I think monetizing your passion projects is almost like an oxymoron, right? Like, if it's a passion project, you care about it. You should not get other people involved. Like, we are in the business of media. You can sell media. You can come up with your media product. You can sell your media product, you can sell your ad integrations into your YouTube videos, but you have something you deeply care about, that you just know how to make it. You know what it should be.
- [64:48]
A
I call rank.
- [64:49]
D
Not a bunch of yeah, or you have a financier that you trust. Yeah, you got a financier that you trust. That's not Going to hinder it. And again, Curry, I think the danger with these creators now is actually having too much money. Money, you know, like Curry making that for 750 is what makes it special. It's what actually gets less strings attached.
- [65:10]
B
Yeah.
- [65:11]
D
And why it's good Markiplier making something no strings attached. Like that's just his money. He gets to make something really special. So I think markiplier is a great example. Like when you have a passion project, you, you ideally have the cash to do it or can get it from someone you trust who's not going to intervene with it. But I think that is a danger. I think, I think too much money can kill creativity very quickly and it just becomes a product. Of course. And so I would caution.
- [65:39]
B
Yeah, I have one last hot take that I want your reaction to. We need to bring the host read mid roll ad to the cinema. Movies should have host read mid roll because if you look at the, if you look at the budgets of some of these films. 750. A million dollars. If you're watching Obsession and then you just at the one hour mark and
- [66:01]
A
then you get an ad for Chase Sapphire Reserve.
- [66:04]
B
Chase Sapphire Reserve.
- [66:05]
A
Not just product placement. I want the, I want the characters to sit down.
- [66:09]
B
No, no, no, no, no, no. The whole point is that it's not product placement because everyone goes to product placement movies. This is, this is separate. This is a host red ad.
- [66:19]
A
I want the characters to sit down and not just use like a Chase Sapphire car to pay. I want them to look at the camera and deliver.
- [66:29]
B
I'm actually sick of seeing James Bond use the Samsung Galaxy S26. I would rather just James Bond tell me about it separately and then I can go back to enjoying the movie with whatever phone he would use. Whatever phone the character would normally use.
- [66:42]
D
I love you guys and I love how you speak about advertising. I disagree wholeheartedly. I don't want that at all. I'm happy to have the theatrical run sponsored by someone like Chase. If you have a Chase Sapphire card, you get in for free. And like you get perks for something if you have a Chase Sapphire card. The one thing my hot take is I want movies to be like 90 minutes and under. Again, I don't, I, I can't do the two hour movie. Like, I just cannot do it. I would prefer like 90 minutes and under. I got time for. Yeah, two hours feels like. And I think all of our attention spans are being trained to like be way shorter than that. Even an hour and a half is like Working out like it's hard.
- [67:20]
B
Maybe go shorter. 90 second movie. Movie vertical.
- [67:24]
A
Well, John and I are gonna. John and I are starting 20 bucks.
- [67:27]
B
20 bucks. But I, I, you know, I checked the box. I saw a movie. Yeah, I went to the theater.
- [67:31]
G
That would be good.
- [67:32]
B
90 seconds in about.
- [67:32]
A
That'd be good for me.
- [67:33]
B
That'd be great for you. You'd be able to catch up. You'd be like, yeah, I've seen hundreds of movies.
- [67:36]
A
Yeah, well, we can agree. We can agree to disagree on ads. John and I will start a media buying company called Fourth Wall. And it's just focused.
- [67:44]
B
Fourth Wall.
- [67:45]
A
Focused on breaking down the fourth wall and delivering ad impressions during your most sacred moments of entertainment.
- [67:53]
B
Yes, yes, yes. The next Godfather will be sponsored fully. Anyway, enjoy Montana. Thank you so much for taking the time on a busy day to come chat with us.
- [68:01]
A
Great to see you, dude.
- [68:02]
B
Great to see you again.
- [68:03]
D
Not that busy, but appreciate it.
- [68:05]
A
Okay, have a good one. Serious activities.
- [68:08]
B
Goodbye. Let me tell you about Cisco. Critical infrastructure for the AI era unlock seamless real time experience is a new value with Cisco.
- [68:17]
A
Hey, team. Is this Ben? Is this real?
- [68:20]
B
What is real?
- [68:21]
A
Did you take a screenshot of what happened?
- [68:24]
B
What is this?
- [68:25]
A
Look in the chat.
- [68:26]
G
No.
- [68:26]
B
No way.
- [68:27]
A
He found the target. Just off that one screenshot. Our intern Ben was able to find Samir's exact location.
- [68:36]
B
Okay. We will not dox it. We do know exactly where it is. Potentially. I don't.
- [68:40]
E
I don't know.
- [68:40]
B
This might be a losing, but it looks pretty. It looks pretty accurate. Anyway, we've been keeping our next guest waiting. We have Tom Fairley from Bullish in the waiting room. Let's bring him in to the TVP and Ultra Dome. How are you doing?
- [68:53]
A
What's going on?
- [68:54]
B
Good to meet you. Welcome to the show. Can you hear us?
- [68:59]
A
Cannot hear us.
- [69:01]
B
Okay, we're turning on the audio. Hopefully we'll hear him. In the meantime, let's go to this debate. Okay, back. Becky Quick at CNBC is getting in trouble because she claimed the GLP1s are responsible for school budget shortfalls. When I saw this, I was like, what could possibly be the connection? And Bio Investment 5 said she's so misinformed. She fact checked. It turns out it's true. School budgets are paying for GLP1s. They're very expensive and so that's driving up healthcare costs in a lot of. In a lot of school.
- [69:34]
A
Welcome to the show.
- [69:35]
B
Anyway, welcome to the show. Thank you so much for taking the time. How are you doing?
- [69:38]
G
Sorry about. Sorry about the audio issue. And I found the little bit about Becky interesting. She's not just speak about somebody else, somebody on a different network, but she's a pro. She's a serious journalist.
- [69:53]
B
Everyone in cnbc, we love the Squawk Box crew. They pioneered this whole format, invented television over there. Anyway, anyway, let's go to you. Tell us about Bullish. Give us the update, take us through what's recent in the business, and let's get started.
- [70:06]
G
Yeah, sure. Just before I do, just to follow on. On that, you said you're a fantasy NBC. I too am a fantasy NBC, but also we have our own media business here.
- [70:15]
B
No way.
- [70:16]
G
And I've watched with some, as both a practitioner and a fan of somebody who's done a lot of it, I've watched with some astonishment the success that you guys have met with with by just creating a fresher, more interesting format. You guys do an awesome job. Thanks for having me on.
- [70:30]
A
Thanks for saying that.
- [70:32]
B
It's been a really wild ride. It's really fun. And best part is getting to talk to people like you all day long and again and again and talking to six people a day. A lot of people start podcasts because they want to talk to interesting people, but then they talk to one a week. We get to talk to seven a day. It's great.
- [70:49]
G
Yeah, yeah, super Cool.
- [70:51]
B
Cool.
- [70:51]
G
Anyway, I'm happy for you and happy for me.
- [70:53]
E
Yeah.
- [70:53]
G
What can we talk about?
- [70:55]
B
Well, why don't you take us through the shape of the business currently and then some of the new deals that you've been working on.
- [71:01]
A
Wait, can we get some backstory first?
- [71:03]
B
Oh, yeah, yeah. Let's start there.
- [71:05]
A
You're at nyse, right?
- [71:07]
G
Yeah. I was born in Bowie, Maryland, just outside of D.C. when I was gonna
- [71:13]
B
say I was born on the stock
- [71:14]
A
exchange, I was born on the floor.
- [71:16]
B
Home, birthday, trading floor.
- [71:17]
A
On the trading floor.
- [71:18]
B
That would be amazing. Lore.
- [71:20]
G
I actually didn't know, like, I wasn't born with any, like, pre. Any. Any destiny. I didn't. I didn't know. I always. I was always interested in business and money and no, it wasn't. Nothing was preordained, but so
- [71:37]
H
grew up in.
- [71:39]
A
A lot of people are afraid to say that. I was never afraid to say that.
- [71:43]
G
Yeah, no, I could tell. My mother would shake her head sometimes. My mom was a nun for five years and worked in. I grew up in a place called PG county, and she worked at Elizabeth Seton High School, which by and large had kids from a different socioeconomic strata than me. And then she would come home and I would talk about making money, and she looked at me like, oh, he's the black sheep. But met my wife at Georgetown. I needed a dollar to buy a Gatorade, and she lent it to me. And now we have three kids in New York City. And I just fell in love with business. And first couple years, I did like banking and private equity. Honestly, I love that too. But when I really kind of got my mojo is in the markets industry because it's the intersection of tech and business. And that, to me, is my happy place. And I've been there ever since I started when I was in my 20s. I'm like, granddad now in the digital asset space. I'm 50, and I've been right there at that kind of overlap of the Venn diagram of technology and markets. I have these friends, the Romeros, and I love them. I love them all. Stay in touch with all of them. There's four brothers, and they're all super smart, super successful, and three of them are just like, meatheads. Always had their shirts off and wrestling and playing hoops. And then Danny was always, like, reading Java Coding for Dummies when he was, like, 10 years old. And I would always keep an eye on him. And so in 2013, I'm like, Danny, are you going back to Duke? And he said, no, I'm going out west to start a blockchain company. And I'm like, all right, I don't
- [73:20]
H
know what the hell that is, but
- [73:22]
G
if you're doing it, I want in.
- [73:25]
A
There you go.
- [73:26]
G
A few days later, New York Stock Exchange, where I was at the time, we agreed to put $10 million into a little bitty, maybe pre revenue company called Coinbase. And so I got this bird's eye view of the digital asset space.
- [73:41]
A
Dan Romero.
- [73:42]
B
Yep.
- [73:42]
A
Whoa.
- [73:43]
B
Yeah, we know him well. He's been on the show.
- [73:45]
A
That's a crazy. That's a deep cut.
- [73:47]
B
That's so good.
- [73:48]
G
It was great. It was great. Literally, I'm like, danny, going back to Duke? He's like, no. He's like, no, Tom, I actually graduated this past May. You're a year off. I'm starting this blockchain company. Meanwhile, I know I'm going in to run the New York Stock Exchange change. He does not.
- [74:01]
B
Yeah.
- [74:01]
G
And I'm like, hey, dude, you got. You got a couple minutes? Come sit on the porch. And he comes up and he starts telling me about blockchain. And I'm like, I'd heard of Bitcoin, but this is early, right? It's like smart contracts.
- [74:13]
B
And. Yeah.
- [74:13]
G
And I'm like, but, but, but what use is there for these smart contracts? He's like, oh, it's going to be amazing. It's going to disintermediate. It's, it's this layer of programmable finance going to disintermediate all the kind of big institutions like the New York Stock Exchange. And I'm like, what, what, what are you talking about?
- [74:29]
B
It can't be.
- [74:31]
G
And so I said, I got to put some money. And he said, oh, my roommate Fred is going to be in New York tomorrow. Why don't you meet with him? And so I met Fred Ursam. I ended up meeting with, with, with Fred at Union Square. And we put that dough in and it was like the best money we ever invested. We got lucky on the return. Just dumb, you know, Is this like
- [74:50]
A
this Series A stage? Like, I don't know, I can't remember
- [74:54]
G
last Tuesday, let alone 2013.
- [74:56]
A
I could go early, but early, I
- [74:58]
B
mean, Coinbase was in my YC batch in 2012 and summer 2012, so I think seed or Series A was right around.
- [75:06]
G
We should punch, we should punch it in, you know?
- [75:08]
B
Yeah.
- [75:08]
G
Anyway, it was a long time ago. Punch in New York Stock Exchange, Coinbase, it'll come up.
- [75:12]
B
That's crazy.
- [75:14]
A
But that must have been what, what kind of reaction did you get on Wall street from that investment? At the time? I don't know how public it was, but I imagine you still had some, some of your peers that were saying, you know, what, what the heck is I'm up to?
- [75:27]
G
You know, I think if you ask Brian or Fred or Fred at Union Square, I think they would all say, with humility has nothing to do with me. It has to do with the NYC brand, that it, that it was hugely valuable for Coinbase. It was not only public, it was massively public public. So in other words, if you were to look it up, you're going to find story after story after story after story, which is, hey, traditional bellwether. A traditional bellwether firm. New York Stock Exchange validates crypto firm Coinbase. So on the, on the west coast, it was, it was like a lot of celebration. And, you know, hey, this is finally the coming together, these two worlds. On the east coast, it was like, what the hell is this? You know, what a weird investment from these guys. And it wasn't as, it wasn't as celebrated. I will say just my thesis was this was a programmable layer of finance for institutions.
- [76:31]
A
Yeah.
- [76:31]
G
Like that was the thesis. It was dead wrong, if I'm honest. You know, there were no institutions. There weren't until.
- [76:38]
A
Yeah, was that just a. You had the timing wrong. You didn't.
- [76:42]
G
Yeah, but pretty darn wrong, right?
- [76:43]
A
Yeah, pretty darn wrong.
- [76:47]
B
Yeah.
- [76:47]
H
Yeah.
- [76:47]
A
But you're only off by. Only off by what, a decade?
- [76:51]
G
A decade plus.
- [76:53]
A
But miss that retail would be the early adopters.
- [76:58]
G
Yeah.
- [76:59]
A
And even some of the late adopters.
- [77:01]
B
Yeah, I guess. How do you synthesize the fact that crypto did not, not disrupt the traditional finance system like entirely? Like, you know, J.P. morgan, Morgan Stanley. These firms still exist, they're chugging along, New York Stock Exchange is still chugging along, NASDAQ and many others. And yet crypto firms have been very successful. It was sort of a like heads I win, tails you lose situation. Like both, both sides did. Well, there was not total destruction of the traditional financial infrastructure and institutions, but crypto was, you know, wound up flourishing. Like what does that say about everyone's preconceptions? I guess because there were two sides a decade ago. Either crypto is a zero or it's going to take over everything and we wound up with the middle road.
- [77:56]
G
Yeah, I don't. Well, I agree, I agree. And I'm going to challenge it a little bit. I'm not sure it was, it has been up to this point such a big win for crypto. There's actually relatively. For all the money and attention and time and, and you know, ink spilled and for all that has gone into digital assets, there's actually relatively few successful companies and certainly the disruption to traditional finance has been relatively small and very small actually. I, and look, it hasn't always been great. Just I'm not a cheerleader. I call balls and strikes. Like not only were there winters, but there were frauds and scams and you know, like I'm not in it for the fart coins and monkey drawings. I'm just not. And I respect it and I understand why they're there and just to pivot to that. The nice thing is they've brought a lot of attention into the industry, a lot of money into the industry and I think it's all been a warm up act. And remember, blockchain is going down in some cases once a day. None of that happens anymore and they've been battle hardened and regulators have now come around and you've seen broad acceptance of blockchain rail for serious use cases in Asia, in Hong Kong and Singapore, in Europe, in, you know, in Brussels, Europe. But then in the individual countries and now in the US not just with Genius, which is the state that, you know, it ratified. Stablecoins kind of gives them papal blessing, but hopefully the, the Clarity act, we can talk about how to handicap that, but if that gets passed, that, that'll be perhaps the biggest regulatory or legislative step ever. And so I am very, very optimistic that I come on this show a decade from now and the global securities market is, it's on blockchain rails and that's a $270 trillion market. So there's no mission accomplished banner behind me in terms of crypto up to this point. But it's coming and it's coming in real time. That's kind of how I, I view it.
- [80:09]
A
Yeah, crypto, you know, the, the, the cycles and the hype cycles are so vicious and yet it keeps creating these opportunities where the true believers, if you just maintain that conviction, like this window right now where there's some, at least you know, we, we primarily cover private, private markets and so there's so many, there's some interest at the intersection of AI and crypto. There's been a decent amount of investment there, but I don't see hundreds of new companies focusing at the intersection of what you guys are doing, which is that true institutional adoption across global equity markets and other asset classes. I can, and you know, thinking on that 10 year time horizon, I very much can see how the best days are yet to come and huge opportunity to just like, you know, be in the foxhole and focus on the opportunity.
- [81:14]
G
Yeah. And I think agentic commerce will intersect with blockchain technology. And I'm not one of those who Sundays oh, and 100% of you know, agent commerce is going to happen on block chain rails and you know, know, Fiat, Fiat Rails will, will see their share. But that's, that's going to be a big driver as well. But you're absolutely right, Jordan. Thanks for doing the prep work and learning up on Bullish because you're absolutely right. We are focused on institutions really institutions exclusively in our exchange business. Over in our coindesk and consensus business. Of course we're a media business so we service all sorts of people and customers but we focus on, on institutions exclusively.
- [81:57]
B
Yeah, I'm reflecting on the question of like the, should there be a mission accomplished banner behind you? And I take your point at the same time. Yeah, there has been a lot of success in crypto. Trillions of dollars in market cap, a lot of that captured by the coins themselves. But there's a number of Very successful companies, yours is among them. But I'm interested in where you think we are in 2026. A lot of the prices are. We're not in some insane falling knife scenario, but stablecoin volumes are sort of stable, even though they were growing very quickly. The bitcoin price is fairly stable, which is what?
- [82:40]
A
Fairly stable microstrategy selling.
- [82:43]
B
But, but it just doesn't, it doesn't feel like, oh, there's, this is the FTX moment all over again. Like crypto is about to fall into zero and people are panicking. It's more just like this is a mature, stable industry. There's some growth, there's some companies that are beating earnings, there's all sorts of different stuff happening. It feels very mature. And so I'm wondering, yeah, well, it's
- [83:06]
A
mature, but still feels like there's an opportunity for pretty parabolic growth as you can get real institutional adoption, not just, hey, we're going to hold some bitcoin on our treasury, etc.
- [83:22]
G
So yeah, I agree with everything you both just said. I think what John said is kind of feels like a more ordinary industry. And by ordinary, I don't mean that in a pejorative sense, but it's no longer based on hype, it's based on real use cases like stablecoin coins and stablecoins are being used for trading and stablecoins are being used for payment and stablecoins are being used for, for land borrow and they've gone from zero to 300 billion. And you know, the whole industry is now 2.7 trillion. And yes, bitcoin is more stable. Now. We happen to be sitting here at a moment in time when bitcoin's down, you know, 15% over the last two weeks or something of that nature.
- [84:04]
B
But still times when these, when like there was like the whole crypto industry is down 50% and people are actually penning the take, everything's going to zero and it's being debated seriously. And that's not happening right now. No, it is a different time.
- [84:22]
G
And what Jordy said, that I agree with. So you mentioned, John, that crypto is a couple trillion the last time I looked. Coincidentally, I'm not that good at math. In my head, the numbers made it real easy. 2.7 trillion trillion. And the global securities market is 270 trillion. And that is coming. Right. Money market funds have already moved on to blockchain rails. US equities have already moved on to blockchain rails. You look at all the industry bigwigs from My old haunt, the New York Stock Exchange or DTC the Clearer or many of the electronic brokers, Kraken or Binance put out an announcement yesterday. Yesterday they're all tokenizing stocks. You're seeing more examples of tokenizing fixed income. And so you're looking at the parabolic growth that Jordi referred to I believe is going to come from the so called tokenization of the global securities industry.
- [85:18]
B
What's more important there, cost or speed? Because and when I mean speed, I guess I mean like 247 trading. That seems like it's crazy. Crazy. We don't have that yet everywhere. It seems like a purely technical issue. And yet I've talked to folks at hedge funds that are like, yeah, like we actually can't get traders who want to work at 2am Eastern on a Sunday. It's hard to recruit the best to actually make that market and provide liquidity at that point in time. But then on the other side you have you know, fees and all of the different things that happen with crypto potentially dropping costs. Like what are the different pieces of adoption?
- [86:01]
G
Yeah, the 247 what, what people miss sometimes is actually the investing demand especially for the most liquid US based securities like think Nvidia stock.
- [86:13]
B
Yeah.
- [86:13]
G
Comes the incremental buyer comes from abroad and you know, Singapore where we have an office is 12 hours away and they don't want to trade some crappy derivative or they don't want to trade some like iou, they want to trade the actual stock. And so by going 247 you can bring on new liquidity which drives down the cost of equity of the, of those companies. And so there's, there's real incremental benefit. I think to answer your question, like what is the benefit that's most attractive? You have to look at the particular constituent the issuers, they hate little things like wow, you know, I say little, they're not little to them. But like this issue of naked short selling which is illegal but is possible in a world where trading is obfuscated in a Russian series of Russian dolls of, of brokers or other intermediaries with blockchain rails. There is no naked short selling. You just look are the, are the shares in a wallet or are they not in a wallet like it's, it's done. But also issuers want the ability to pay stable dividends and stablecoins. Issuers want the ability to be able to, to reward long term holders. Imagine your company say hey we're going to pay a dividend of a dollar but we're going to give an extra penny for every quarter you've held our stock. There was a whole exchange on the west coast that started. That was really great.
- [87:30]
B
Eric. Long term Stock Exchange.
- [87:32]
G
Right, exactly. Marc Andreessen introduced me to eric Ries maybe 10 years ago, 15 years ago. And I remember the first meeting I had with him. I thought, thought this guy's onto something. Like, yeah, we should reward that. We should reward long term holders. Well, it turns out you can't do it if you don't know who owns the damn thing. And it's buried within five Russian dolls or blockchain technology. Again, you just look at it.
- [87:56]
B
Oh, they own it.
- [87:57]
G
It's in that wallet. It hasn't moved. So for the issuer, it's actually incremental functionality as opposed to cost or speed. And then for the investors, it's just the ability to do things at a lower cost. So I guess that's more, that's more cost. But then on the speed perspective, being able to settle instantly is interesting. So it kind of depends.
- [88:20]
A
Depends which constituent you're looking at. Short on time, couple questions. People on the west coast who know nothing about finance have been excited about compute futures markets, or treating compute like, like trading it like a commodity. Is that something, how do you think about that potential market? Is that something that you think could start on blockchain rails because it is a new type of commodity in some way?
- [88:55]
B
Yeah.
- [88:56]
G
Two questions embedded in that. One is, will compute futures actually be a successful product? I think it will be. The issue is standardizing it and commoditizing it successful. The real successful futures contracts with lots and lots of open interest all come down to the question of are you able to standardize it and commoditize it and come up with the cheapest deliver that has value in the world and everybody agrees upon. So that's not going to be dead easy. It's still probably going to take a couple of years, but I suspect that that will happen.
- [89:25]
A
The second question is that because like an onion is an onion, but a GPU is not necessarily a GPU. Like an A100 is not A, is not a Blackwell. So then do you need.
- [89:35]
G
Exactly.
- [89:36]
A
And the providers, you know, basically sub markets.
- [89:40]
G
Yeah. And the providers have a vested interest in not being commoditized. So sometimes it takes time for the market to commoditize, but it's, it's, it's exactly what you just described.
- [89:51]
A
Got it?
- [89:51]
C
Yeah.
- [89:53]
G
Go ahead.
- [89:53]
A
One more random question. Friday IPOs. We're looking at a SpaceX IPO next Friday. My sense is, does that not introduce more chaos? Because you have one real trading day, there's going to be so much retail activity, then you go into a weekend. Of course people will be trading it over the weekend in different ways, but I know there was Alibaba and Uber that both went out on the NYSE on Fridays. How did those go? Does that make it more complicated or chaotic or are the exchanges set up to handle that?
- [90:43]
G
The Alibaba IPO in particular was one of the best days of my career, maybe even my life. But it was also coming from the
- [90:50]
A
guy who as a young boy knew he loved money.
- [90:56]
G
I also was so friggin nervous that honestly, when you just said that, I like was tingling remembering it. We were in our beautiful ballroom that morning and Jet Li was there, Masa and Jack and Jimmy Lee, who's departed was a iconic and legendary JP Morgan banker. He kind of sidled up to me. He's like, hey, Tom, did you guys test this thing at all? Is this actually going to work? Because you're about to get a gazillion retail orders. Good God. Next Friday. It's going to make Alibaba look like child's play. So Adena at the NASDAQ is fantastic. And the NASDAQ is fantastic. Oh, man, it still makes me nervous. But to answer your question, I think a Friday's fine. There will be awesome price discovery on Friday and you're going to see extreme volatility for the first hour. It's. That's nobody's fault. That's. That's called markets. Could it be better? Could there be better mechanisms that could be put in place to smooth it out? Yes. And we should all strive to get there. I just went through an ipo. I have a thousand things that need to change. I think blockchain technology can be very, very helpful in the future. And I think you're going to see more capital formation on blockchain rail rails because it can do just that. But high volatility to be expected. It's going to calm down dramatically through the day and it won't be a problem getting through the weekend because there's not going to be any substantive, substantive new news.
- [92:24]
A
Right.
- [92:25]
B
Thank you so much for coming on the show.
- [92:27]
G
Yeah, it was fun.
- [92:28]
A
Maybe you should come back on next Friday.
- [92:31]
B
Yeah, we got to do it.
- [92:32]
A
This was super fun.
- [92:33]
B
This is a fantastic. We'll talk to you soon.
- [92:35]
A
Yeah, great to meet you.
- [92:37]
B
Have a good one. Let me tell you about railway. Railway is the all in one intelligent Cloud provider. Use your favorite agents to deploy your web apps, servers, databases, more. While Railway automatically takes care of scaling, monitoring and security. Our next guest is Nikesh Arora from Palo Alto Networks. He's the chair, CEO, friend of the show. Welcome back, Nikesh. Great to see you. How are you doing?
- [92:58]
H
I'm doing great, gentlemen, how are you?
- [93:00]
B
We're doing well as.
- [93:02]
E
We're doing great.
- [93:02]
B
It's been too long.
- [93:03]
H
Let me see those watches.
- [93:05]
B
Ooh, I have a Rolex Submariner in TVPN green on. We got them for the whole team.
- [93:12]
A
Yeah, not gonna be happy to see.
- [93:13]
H
What about your favorite guests?
- [93:15]
B
Our favorite guests. We do have some of those, but.
- [93:18]
A
We do. We do. I actually, I can't wear watches when I'm using a computer. Really, it's too annoying to me to have it clacking against. It's just. I can't do it. Total skill issue.
- [93:29]
B
Total skill issue.
- [93:30]
A
Total skill issue.
- [93:31]
B
Well, no skill issues over at Palo Alto Networks, thankfully. Incredible progress. Take us through it. What's the latest in your world?
- [93:40]
H
Well, I think the good news is we can officially declare the Sassocalypse dead for cyber security.
- [93:51]
B
Why? Why is it dead?
- [93:53]
A
Are you dead? Were you surprised who killed it that it took so long? It obviously has only been, what, a couple months, but still.
- [94:02]
H
Oh, good news. It gave every one of our happy shareholders a buying opportunity. Is good.
- [94:07]
B
Yeah. So there you go.
- [94:08]
H
It took a few months because I think, look, the general tendency in the market was AI is going to eat every piece of software and it's going to be amazing. It's going to take care of it. Now, AI is good, but even now the false positive rate on unaided models is 25%.
- [94:25]
B
Right? Yeah.
- [94:25]
H
So if I don't explain the harness, if I don't explain the context, if I don't explain what the code is trying to do, the model still gets one out of four wrong. In cybersecurity, our business will get everything right, start getting one out of four wrong. I don't think I'm going to get replaced anytime soon. Yeah, it's like having your car driven by OpenAI. Because I don't need a waymo. I don't need to have all the machine learning code that tells you everything. So let's just stick a model in a new car and go drive off. It doesn't work like that. So I think what the market is beginning to realize is cybersecurity is going to get enabled with AI. And if you want to get it enabled, you have to go with the people who have been doing this for a long time who sit at various enforcement points because we sit at north of 125 million enforcement points. You need AI to enable the enforcement points with new techniques to go suss things out. That allows us to do a better job. But it's not going to come over and take my enforcement right away and replaced me. So are we going to use.
- [95:19]
A
Yes, there's an alternate reality where the world wakes up to AI being good at, you know, hacking computer systems and all of the cyber stocks just like rip like off of that news. But it was interesting. Interesting.
- [95:34]
H
Well, if you saw, you know, one of the core skills of midsoles is ability to daisy chain vulnerabilities and develop an attack. So yes, it is not, not an alternate reality. It's not hypothetical. It is true. It is. The capability exists. And that's why they were being so careful, rightfully so, in making sure they don't release the model at large because they know the model unconstrained has the ability of determining attacks based on vulnerabilities. So I think is the right thing for them to do is to make sure they do that in a measured fashion. But what you're beginning to see more and more capabilities getting embedded in the core models. OpenAI has one now and Anthropic has another one. I'm sure Google has one too. So just play this movie forward six to nine months from now. It'll be impossible to put the genie back in the bottle or constrain them. We have six to nine months to make sure that if somebody's able to use AI to attack somebody, the infrastructure on the customer end is equally fast to be able to repel it or block it at the same speed. You don't get that without modernizing your stack, without of course buying a lot of Palo Alto and having us help you get to a place where we can defend in real time. Just the way AI can attack in real time.
- [96:43]
B
So the cancellation of the SAS apocalypse is clearly it was a referendum for cybersecurity. I'm not sure. But the claim was AI will make Palo Alto Networks business worse. And I think that's been disproven. We've discussed that. The bigger question I have is how does AI make your business better? Is it just the demand side because there are more threats than ever, so people have to buy more Palo Alto or are there other things happening internal to the organization that are making you more efficient, higher growth? Like what are the other levers that are being Pulled in the AI era.
- [97:23]
H
Well, like AI going back to multiple places, Right. The first place is the amount of build out that's happening across the world on the AI require means that more bits are going to be traveling across networks.
- [97:34]
F
Right.
- [97:35]
H
Everybody's trying to deploy, collect more data. Remember, cybersecurity is the inspection business. We inspect every bit. We're the TSA of the Internet. Right. Every bit has to go through a firewall. Every bit has to go through a product that looks at the bit. Say good bit, bad bit, bad bit, stay away, good bit can go. If you're going to explode the amount of traffic in technology, in the world, you're going to need more cybersecurity capabilities. That's why one that drives more, more traffic, drives more demand. I think the other part is capability. AI models are good at certain things. The question is how can we take them, take their false positive rate from 25% down to zero and leverage them. And what I do, what we do. And that's happening right now. We're all looking at how to take certain key categories of cybersecurity, embed AI into them to make sure we build a better product for the customer from a protection perspective. And of course, like you said, said, there's a third aspect where we can all use AI internally in our business to make it much more efficient. And I always maintained, as I said, perhaps last time I was with you, that AI will democratize intelligence.
- [98:33]
B
Yeah. Which means you said the average. Yeah, yeah, you said, you said efficient. Efficient. And it feels like you're a leader in this efficiency moment where so many companies are token maxing. We've saw $500 million bills going out. When I saw the Palo Alto network's headline of using AI, using Mythos to to find critical bugs, the bill was token min. I don't know what happened, but you only spent a million bucks. What's going on? Is this a cultural thing? Are you already looking at roi? Are you head of the ball there or are you being. Is it more like a scalpel? You use it just where you need. Like what is the actual token maxing? Token leaderboard, Token ROI maxing. What are you doing?
- [99:15]
H
Well, remember, we have parsed our efforts into what is extremely critical in crown jewels.
- [99:23]
B
Sure.
- [99:23]
H
And we can't let open source models, open models touch them because we don't want our code to ever train anything in the public domain. It doesn't matter if whatever they tell us, I don't want my code training public domain. So for that we built our own harness. People are all using our own open source sequestered model which can't talk to the to the external world so we know it's safe. So how our code is going through that, we're not paying anyone for it, we're using it ourselves and built it ourselves. Is it going to be 100% of what's out in the public domain? Probably not, but it's probably 85 to 90% there and that's okay, I'll go deal with that.
- [99:55]
B
And on that. Is inference cost an issue yet? Are you burning GPUs to the max and dealing with a half billion dollar bill?
- [100:01]
H
We're just using trained models to run it through because we're using for task, we're not using them to go generation generate new stuff. Right, got it. So that's one part of it. The other part is when it's stuff that is, I don't need to spend expensive tokens for an IQ180 to solve a customer support problem. I can use a smaller language model, I can use an open source model to solve that problem because it's task specific. So we're trying to be judicious about the use of AI, but at the same time I think we're not using it as well as we should be using it. I don't think anybody in the world, except unless you're a frontier AI model company, company has got it perfectly right in terms of how we need to use it. So we've got to get up the learning curve and we'll get there. I think the part which I was trying to say to you earlier, I think it's a very important point. What AI is doing for me internally is, as I said, is democratizing intelligence. The biggest risk in businesses is you hire 100 marketing people, not all of them are equally good. So you'll gravitate towards the ten good ones. And you got to find a way to deal with the 10 really at the other end of the spectrum and 80 in the middle. Good news is if I train my model right with all my marketing data, I can increase the average intelligence of all 100 of them to 95%. Then I just need the judgment of the 5% from them, which means it's a rising tide lifting all boats. From an average intelligence perspective, if I can get the outcomes of 3,000 customer support people to be amazing and there's no disparity, no standard deviation, I become highly effective, highly efficient and pretty good at what I do.
- [101:30]
B
Yeah. Jordy, do you have something or Can I keep going?
- [101:36]
A
What's your point of view on AI psychosis in the enterprise? John and I have been talking. There's a handful of our friends that we believe do have AI psychosis. They're heavy users of the tools. And you ask them what they they've actually made and they can't give you a good answer. Now, certainly there's an entire other category that is being very productive, making useful things. But there's sort of a like Tyler on our team.
- [102:08]
B
Tyler is extremely productive, very low AI psychosis. You ask him, he just builds it immediately and it's there, but it's usable,
- [102:14]
A
tangible, but it feels like.
- [102:16]
B
But there's a lot of make work.
- [102:17]
A
Yeah, there's at least one model that has created a sort of psychosis nexus in the enterprise over the last six months. And tell me which one. We don't need to name the specific model, but there's like, there's a, there's the same maybe sycophancy that you saw with 4o with a consumer, I think is touch the enterprise in an interesting way.
- [102:43]
B
Well, there's just this risk of having the AI model. An employee goes to an AI model, says, help me with this. And then the AI model responds in a sycophantic way. Oh, yeah, you're, you're, you're, you're making a major breakthrough. You're making a huge impact at this company. And it's just shuffling, you know, spreadsheets around, really recontextualizing data, building analyst dashboards. It's not actually moving the needle. Again, it gets back to the ROI question. And I feel like as a leader of a large organization that's very AI forward. This is something that you'll have to contend with, you know, our, my employees using AI tools effectively. It's another question on just the effectiveness embedding.
- [103:20]
H
That's a great question. I think we're past that stage at Palo Alto. So what we did was we let people use AI models and AI tools to experiment, to understand the art of the possible, train themselves. Thankfully, we did it without token maxing. So people have a good sense. They use all kinds of tools to understand where they are. But when it comes to enterprise efforts, we're positive. Say that we've got one great Tyler. I'm sure Tyler's amazing. I have 21,000 people. You know, one great Tyler doesn't change my move the needle for 21,000 people in the company.
- [103:48]
B
So I think he's good. I'm going to take the other side of that, but I, I take your point.
- [103:53]
H
You don't want me to take dollar away.
- [103:54]
B
So that's why don't tell. Yeah, no, last time take my boy. Keep your hands off.
- [104:00]
A
Right.
- [104:00]
H
So I'm sure we have our own dollars. Don't worry about that, we'll be fine. But the more important part I'm trying to make is that, that now we sit and understand what are the 20 things we could do and how much scale impact would they have If I say let's refactor customer support, I have 3,000 people, let's train model, let's train intelligence, let's build harnesses, let's build diagnostic agents, let's understand the problem using AI instead of human beings trying to diagnose it. Now that, that's not abusable model. That's almost like a business transformation project. What data do I need to do? How should my product behave? What's the first agent that you want when you get the data, what data does the customer have to give? So it's reimagining how we do. The whole activity is a business transformation project. I've got 30 people working on it. They're not using AI tools, they're embedding the LLM in a certain place. They're sort of complementing it machine learning, they're complementing with new knowledge bases, they're complimenting with new data that comes from my product. So I'm trying to reduce my false positive down to low single digit. Right. I can't run an enterprise business AI with high false positive. So to get it to low single digits a lot of programming goes around it. The psychosis works or happens if you're just randomly arbitrarily using models to do certain things. We're not doing, we're building hardness.
- [105:18]
B
Most importantly, the psychosis sets in when you're letting the AI evaluate itself or take tell you if it did a good job instead of you knowing the ground truth. If you know that you're looking at a false positive, you're going to remain in reality. If you ask it is it a false positive, it says no, it's not a false. This is the best, dude.
- [105:36]
H
Yesterday I was doing my earnings, okay. I had Gemini up through my script. I don't like the way it reads. Can you just give me alternate ways to say this? So it has this feature called refine. So I had to look at it. I said refine and I'm reading both of these them. My team had written that we simulated a cyber attack and it was possible in 24 minutes using new for trigger I models. When I refined it, he said we attacked in 24 hours.
- [106:03]
B
Like shit.
- [106:03]
H
That's not what I'm going to say. Right. So I can't let AI lose and not pay attention to what it's doing. I still need the human supervision. I need the guardrails.
- [106:12]
B
Yeah, yeah. You have to know the ground truth
- [106:14]
H
that it can do wrong.
- [106:15]
B
Yeah, exactly. Talk to me about not, not the crazy AI doom Terminator scenario, but just I think that there is some anxiety around AI and cybersecurity where you see these powerful models, they're able to find vulnerabilities effectively and there's a fear that there might just be, you know, a rash of exploits that result in like your, some emails leaking or some social network data leaking and just, just a leakier Internet that causes people to be like upset. But my white pill on this, and I don't know if this is correct, is that the gap between private closed source frontier models and open source models appears to be widening. And that feels like there's going to be an increase. Maybe it's six months right now that you have access to the frontier models before commoditized. If that grows, that gives me more confidence because I know you're going to go and clean up the Internet and police it and be the TSA for longer. If that gap was closing and all of a sudden it was, you know, open source, closed source model, you know, gets a new capability and the hackers have it next week, I'm a little bit worried. I think you could do it in a week, but it's much better to have you have six months. But is that, am I in interpreting that information correctly? How afraid should normal people be about the risk to a leakier Internet, a less safe, less cyber secure Internet in the future?
- [107:46]
H
So let's step back. All of us test our software on a constant basis before we launch it. We don't want to have vulnerability in our software. Now guess what? Humans write bad code sometimes. Surprise, surprise. So what AI is doing is discovering that bad code code. And we've been testing our products and so has everybody else in the industry. And every company does that. And on average we used to find four or five issues every month and we'd go build a patch and we'll fix 108 products, we'll fix the patch every, every month and put a patch out there. What we found in the last six weeks was what would have taken us five to seven years to find. Right. So a huge deluge of vulnerabilities. Now of course we have people, we redirected them, we patched them, we had to put out a bigger patch and go patch all of our customer. So and the, the point of that is that there'll be a lot of testing happening in the next three to six months. Already is with Opus 4.8 or 5, you know, OpenAI 5.5 and other models out there. There's a lot of testing going on in the industry across the board. Everybody's testing their code. All that code is going to get patched in the next few weeks, months, that's what have you. Now there'll be huge influx of patches to every IT department because you got to patch everything right away because everybody showed up with all these patches. So it's going to be bumpy. But I think what we're going to be doing the next three to six months is paying off a lot of technical debt we've accumulated over the last many years is bumpy. But it's a better starting point six months from now than it is today. So I'm an optimist. I think if you're a better starting point six months from now, the next model is going to find a lot fewer things than the last one found because we had to deal with all the stuff that we found. Yeah, well the next one finds something, I'm sure it will.
- [109:19]
G
Right.
- [109:19]
H
But it may go back to finding five. And we seem to have survived in a world where there were five vulnerabilities found every month by software vendors. Now the second problem, like you said, is AI might be able to concatenate them or daisy chain them and find a way to attack them faster. For that it's a different problem. The vulnerabilities already always exist. When somebody attacks you, the question becomes how quickly can you find it and shut it down. Now that is a slightly longer term problem because that requires to have a much more modern stack, requires you to have way more, more data to analyze anomalous behavior and be able to defend against it.
- [109:53]
B
But we'll get there, pay down all the technical debts, start building technical equity. I like to hear it.
- [110:00]
A
Well, it's, I mean like this is why AI is so exciting. You can do five to seven years of work in a month.
- [110:08]
H
Yeah, it's amazing. I think it's amazing. The only thing we have is we have to find a way of getting 21,000 people transformed without a psychosis to amazing leveragers of AI. So they're using it in their day to day life and Trying to treat it as their assistant as opposed to be afraid of it.
- [110:24]
B
Yeah, yeah, I love it. Last question. I don't want to get too geopolitical, but I want to know about America's compute advantage or the AI race in the context of cybersecurity. Because there has been this debate over our Are data centers the nuclear bombs or are the weights the nuclear bombs? If the weights were stolen by a North Korean hacker firm, they could wreak havoc. But I don't know if that's real because I think you have to actually go and inference these massive models millions of times and test every little system to find the actual exploits to actually do the hacking. And so another maybe white pill for America and the American Internet remaining secure is that as we build more infrastructure, more data centers, more GPU racks, we are able to do more of good guy inference, more cybersecurity, more testing and our systems become more secure relative to the near peer competitors that might want to hack us but have not invested enough in energy and compute capacity. Capacity.
- [111:34]
H
Well, let's back up like look, we're all internal technology optimists, right? We all Silicon Valley, we hang out here, 99% of use cases we talk about AI are good use cases. Is doing amazing things, going to do drug discovery, it's going to do a whole bunch of positive stuff, is going to get rid of mundane things that we don't have to do. So generally being an AI optimist is a good thing. Does America have an advantage? Yes, we are building on the last lead in the technological revolution. Contributions on last round go around, right? Where all the hyperscalers live, where do all the large consumer tech companies live? Where do all the CyberSecurity companies live? 90% live in the United States, right? So we have the core of building the next wave of technology on the back of the last amount of success we've had. And I think all the efforts you're seeing, whether you're seeing it from the recent executive order cybersecurity, whether you're seeing it around investment in chips available, seeing it around investment in electricity, everything else we are trying to build the underpinnings of building an amazing future with AI with us. Now, does it mean there will be an edge case of single digit percentage of where people will try and use for bad things? Every technology, every invention has had people using it for bad things. Can we protect cases? Yes, we will have protections. Will there be some notable mishaps? Most likely, you know, it's happened. We have 2000 ransomware attacks a year, we move on, we go solved it. Them life moves on, we fix and we move through it. So I think I've taken an optimistic outlook now. Can models be distilled?
- [112:58]
G
Yes.
- [112:58]
H
Can weights get reappropriated? It doesn't look like you need to because every three months somebody seems to build the same capability open source. They're figuring it out. And do they need the three month advantage or not? I think they'll get the advantage. I think the key is going to be is we have to keep sort of marching along and making sure that our infrastructure is up to speed so we can actually manage match sort of defense with the speed of offense can happen. And it's always been an asymmetric battle. Cybersecurity always had to be right 100% of the time. And the bad guys are right. Once they don't need a lot of compute, they can go rent compute and come after an enterprise. They so choose to do so. And this game of cat and mouse will continue.
- [113:36]
B
Last question, how much of your time is spent on quantum?
- [113:41]
H
A lot. So what's interesting is we launched a capability where we can assess people's quantum cryptography readiness.
- [113:50]
B
Yeah.
- [113:51]
H
And we build capability where we can wrap your traffic with quantum secure keys. So even if you haven't upgraded infrastructure to quantum yet, there's a fear that quantum is coming. And there are nation states who might be collecting data which they're going to harvest later when content becomes reality, hoping they can hang on to classified data and break the keys afterwards. So we're already wrapping the traffic now. Yes.
- [114:15]
B
That's really cool if you know there's
- [114:17]
H
classified data and you can start sucking the data. So I don't know what's in it, but it's classified. But I'm going to break it as soon as quantum is ready. And I'm pretty sure nation states will have quantum capability much before it's commercially available.
- [114:28]
B
Wow. That's amazing. I love that you're already working on that. That's.
- [114:32]
H
So we have a quantum wrapper. You can buy a wrapper from us, you can go look at, do full cryptography traffic inventory. We can encrypt the traffic with quantum secure today across your sort of traffic nodes and make sure that your data can be intercepted.
- [114:44]
B
I love it. Thank you, thank you for doing that. And then also thank you for coming on the show. It's fantastic to catch up with you.
- [114:51]
A
Great update.
- [114:51]
B
Congratulations on all the progress and thank you for canceling the SaaS apocalypse. At least in cybersecurity we're going to cancel the SaaS apocalypse everywhere. We're going to cancel all apocalypse.
- [115:03]
H
Sounds like a plan.
- [115:04]
B
Have a great rest of your day. Thank you so much. We'll talk to you soon. Goodbye. Speaking of a company that canceled the SaaS apocalypse, we got Figma agents meet the canvas. Your AI agents can now create and modify your Figma files with design system context. And we got a timeline post here somewhat related to Figma design. Teddy said he couldn't decide on art to hang on his apartment walls, so he built a drawing machine to decide for him. This is a very cool little hack project. I guess that it will move the weights so that it can draw on an x and Y axis, it can generate an image and then it can draw it with a pencil or a pen that you reload and it can pick an image and draw a different thing every day.
- [115:49]
A
Calling it a generative pen train transformer.
- [115:52]
B
Yeah. Theodore Dun. What a good domain name for just an individual like solo creator. It's a generative pre trained transformer. Pen trained transformer. Of course we'll go check it out. But we have our next guest in the waiting room. We have Henry Stern from Privy back on the show. Welcome back. How you doing?
- [116:15]
A
Are you at the Rainforest Cafe?
- [116:16]
B
Are you at the Rainforest?
- [116:17]
C
I'm at the Rainforest Cafe. I'm in Europe because in fact TVPN is the TVPN of Europe, but I'm calling in from Amsterdam.
- [116:26]
B
Yeah, fantastic. What brings you to Amsterdam? Work Money 2020. Okay. And this is fintech conference. Yes. And I imagine you're speaking networking.
- [116:39]
C
Yes, all of the above. We launched a few products. We had a great launch with, with Deal today to enable contractors to basically receive stablecoins globally and do a number of things with them. And then speaking, meeting customers, all that.
- [116:55]
B
Yeah, yeah, that makes a lot of sense. Deal's been really ahead of the curve on that.
- [116:59]
A
So Deal's been dealing with stablecoins for quite a long time.
- [117:04]
B
Oh yeah. Did they have their own like in house solution? Like what? Yeah. What is the nature of the partnership? How does it change? Like what was the, what was the, the before era like?
- [117:13]
C
So I think the before era they had done a few products to enable people to. I believe they'd experimented a lot in the space. But this is basically the coming together, I think of the full package where they're doing three things really. One is enabling any business to pay out their contractors in stable. So the contractor so chooses they can receive basically stable assets as payment no matter where they are. And they're starting to rollout in Argentina, but moving a lot from there. Two is they're basically putting out a wallet app. Doesn't look like that at all. It just looks like an account where you can receive your payment, but that allows you to basically hold the balance, earn on the balance, so really a savings account in many ways and then basically spend from the balance. And so this is sort of the whole package coming together in an exciting way. And then the third thing is they are actually issuing their own stablecoins so they can have a lot more control over how the underlying treasury is managed and so on and so forth. And so from the previous standpoint, I've been very focused on the wallets and how do you connect all of these pieces together. But then obviously across stripe we've been working with them on a number of products. So it's been kind of fun to watch all these products coming together really well.
- [118:26]
B
Yeah.
- [118:26]
A
So I imagine the announcement from their side is really, they have their in consumer fintech now.
- [118:34]
C
Exactly.
- [118:35]
A
Very cool.
- [118:37]
B
I imagine internationally there aren't like spending the stablecoins. I understand receiving for contractor payments makes a ton of sense. Earning on the money that's in the account, that makes a ton of sense. But I imagine actually living your life in any country, even in America, it is still hard to go and spend stablecoins. Or is it not? Because there's crazy credit card infrastructure that can be tied in very quickly. Like what does it actually take to run your life? If you're a contractor in Brazil or Egypt or Thailand and you make money from deal they pay you in stablecoins, you winds up in a wallet, you're very happy with your interest that you're getting. But then you want to go live your life, you want to pay your rent, you also want to, you effectively want to be able to send wires and peer to peer payments and also buy food and get money from an ATM at some point. How does that all work?
- [119:33]
C
Yeah, it's a great question. I think this is the part that's very exciting is kind of how the whole space is getting stitched together so at a high level. At this point there's just a huge explosion in the usage of stablecoin cards that enable you basically to settle directly on traditional rails like Visa and MasterCard from your stablecoin balance direct directly. So for the merchant accepting it, they don't have to think about this at all. It's the same experience as if they were getting a purchase from anyone else. And then obviously in Reverse. We're doing a ton of work at Stripe to make sure that merchants can also accept stablecoins directly if the user has the balance and so on. So we're kind of working on the platform from both angles to make sure that it's completely multimodal.
- [120:15]
B
How high is the Great Wall of Great Firewall in China? Like, is there any stablecoin penetration in China? Are people doing VPN stuff or is it because there were some tech companies? Apple obviously wound up working with China effectively. Airbnb was over there for a while, as if you wanted to rent a place because, you know, it's not like the countries are actively fighting. It's, you know, it's a tense rivalry. But there are companies that have been successful there at the same time. Facebook not successful. Google not successful. Uber sort of was unable to get a foothold there. Tesla doing very great. Model Y sell over there. So what is stablecoin adoption? Does China have their own ecosystem or is there some cross pollination?
- [121:04]
C
No. So today I think the situation has changed over time, but basically today, very much remote in this, in a sense, and not an active part of what's happening. So we don't, for this particular speaking for privy, we don't have customers that we serve there in terms of stablecoins because very specifically, it isn't something that's running. So the approach is very much the technology itself. In a sense. This is a file format for money. You're kind of sleeving your fiat into the drawset. And the way I like to think about it is like different stablecoins are like different formats. So we accept JPEGs and PNGs just the same. But then ultimately getting currency on the ground in every country depends on the local regulation. And in this case, China is very much not open to stablecoins today.
- [121:55]
B
Interesting.
- [121:56]
A
Yeah. More and more and more capital control.
- [121:59]
B
Yeah, I mean, yeah, makes a ton of sense because you load a bunch of stable coins bitcoins and then you transfer all your money out of the country. Are you tracking any? Like, is there like a mirror industry, like an indigenous Chinese, like, digital yuan effort that's getting traction? Because, like, there's been a whole open source AI movement. I'm wondering if there's, you know, Chinese versions of crypto projects that are actually maybe, you know, six months behind.
- [122:26]
C
It's a great question. I mean, to be honest, I don't know. No, I think the honest take is twofold, which is one. We've seen a ton of adoption in the US since Genius and with a lot of what's happening around clarity and so that's kept us quite busy. Likewise in Europe and then the adoption in Latam and other parts of Southeast Asia is huge. There's a really vibrant set of communities in Singapore, in Korea. So there is a lot happening in the region, region generally. But I just haven't seen that much by way of companies working in mainland China doing this sort of thing. So I couldn't tell you to be honest.
- [123:03]
B
So in terms of maybe other countries that are the fastest adopters, it feels like America is on the frontier here. Clear regulation, some more coming down the pipe. A lot of the companies are based here. Who's in second? What country is leaning in the most to stablecoins broadly?
- [123:24]
C
It's a great question and I think part of it is that today by nature at least, it's a very cross border technology. So it ends up being pairs basically as you go along. And so there's very good corridors. For example US to Mexico, we serve a company called Felix Pago is doing like 5% of the US to Mexico remains corridor. I think that's like $60 billion a year. So it's a really meaningful. The entire quarter, $60 billion a year, but it's a really meaningful pair. Likewise I talked to a company not too long ago that was trying to help performances from the UAE to India and so you've basically got these pairs as you go. The most leaned in though Mexico is doing a lot and there's a lot of clarity there and then we're seeing just generally a lot of activity in Latam so that's where a lot of the volumes are happening.
- [124:11]
B
Very cool.
- [124:12]
A
What does NuStar up formation look like in the stablecoin market broadly look like at this moment in time? I'm sure you get pitched all the time as an angel but for a lot of these it feels like in a lot of these categories there is now a category leader and maybe less opportunity even though we're getting so much adoption.
- [124:40]
C
Yeah, it's a great question. I think basically there's like two types, right? There's very pure software plays and I think Privy was such a company in that we run cryptographic infra to enable people to do things and then there's a lot of people who want to take care of the data piping on the ground in various countries. So I think of these as money pipes in a sense. And you've got the software layer and then you've got the connections into local banking systems. And so we are Seeing a lot of things happening, trying to build better APIs, better abstractions on top of the aggregation. But I think there's actually a huge gap in building localized partnerships because this is really hard obviously in any of the individual countries that you can do. So that's one. And then the other, frankly is we're seeing a lot of stuff around agentic commerce with people basically trying to build better control systems for agents to actually take these assets and pay with them. So this isn't actually stablecoin specific. There's a lot happening in Fiat as well, and Stripe's doing a ton in the space. But it's broadly around service discovery. How can you find what you can pay for and actually connect into it? It's around the actual protocols for delivering payment really easily via sort of agentic interfaces, LLMs and so on. And it's about being able to actually sort of meter spend so you don't get wrecked if your agent hallucinates. So those are probably like two of the super active areas that we're seeing. And then we're seeing just a lot more adoption is something that struck me at this Money 2020 in Amsterdam, which is a lot more institutional adoption, including from larger banks and players who historically have been kind of waiting to come in.
- [126:18]
B
Amazing. Well, enjoy the conference. Thank you so much for calling in. I'm sure it's late there.
- [126:22]
A
Enjoy the Rainforest Cafe, enjoy the, the
- [126:24]
B
Rainforest Cabinet and we'll talk to you soon.
- [126:26]
A
Great to see you, Henry.
- [126:27]
B
Have a great rest of your day. Goodbye. Let me tell you about Shopify. Shopify is the commerce platform that grows with your business and lets you sell in seconds in, online, in store, on mobile, on social, on marketplaces, and now with AI agents. You know, I think the AI agentic commerce could be big because if you get hit with like a $500 million bill for your agentic AI, you might not. It might be over the wireless, so you might need to use, you might need to transfer like tens of thousands of bitcoin. Low fees. This is valuable. That's the future of agentic commerce, potentially.
- [126:59]
A
Who knows, for all the people getting hit with half a billion dollar bills,
- [127:02]
B
it happens to the best of us sometimes. Well, our next guest is Alexander Good from Post Fiat. Welcome to the show. How are you doing?
- [127:09]
A
What's happening?
- [127:10]
E
Hey, what's up?
- [127:11]
B
Not too much. It was great to have you on on a Wednesday.
- [127:13]
E
Oh man, I can't hear you guys.
- [127:15]
B
Okay, we'll try and fix that. I will tell everyone about MongoDB. In the meantime, what's the only thing faster than the AI market? Your business on MongoDB? Don't just build AI, own the data platform that powers it. And let's see if Alexander can hear you. Can hear us. We can hear you. We can hear you. He can't hear that. We need to put up text on the screen that says that we can hear him and then he will know we need update the chyron. We can hear you.
- [127:45]
A
Joe Weisenthal is tracking the popularity of various running shoe brands on r running shoegeeks. 1 out of 18 posts mentioned a major Chinese running shoe brand. Just last quarter, it was 1 out of 40. Li Ning is the big one. Li Ning, of course, signed a shoe
- [128:04]
B
endorsement contract with Golden State warriors star Steph Curry. Very, very interesting. I mean, haven't the basketball shoes been made in China for a long time? That was like the whole Nike thing, but I guess it's like the brand now is big.
- [128:18]
A
No, that's a big. Chinese companies have acquired a bunch of brands like Arc', Teryx, things like that. And the next step is to actually get meaningful.
- [128:30]
B
What is the best penetration of their own brands? Consumer brand in America, the one that Americans like the most that comes from
- [128:37]
F
China,
- [128:39]
B
it's gotta be dji, right? Yep. Drone videographer.
- [128:43]
A
Yeah. Bernie was using like Bernie was using.
- [128:46]
B
I know it's a controversial company because people see it as industrial capacity, but if you're just like, I. My only. I don't think about geopolitics, but I like that my wedding video had a drone in it and it was a DJI drone. You think about that very positively. Has sort of a GoPro aspirational brand. Is there anything else that pops up as Chinese brands that are loved in America? I can't really think of anything in particular. Obviously tons of stuff gets made there, but anything at the top of your
- [129:15]
A
mind, every product on Amazon that has a name that.
- [129:18]
B
Yeah, but that's not. Yeah, I was gonna say hoverboards, but they didn't have a brand and a lot of the cars aren't here. I believe that they would be popular if you could get a hypercar for $20,000 like they make over there.
- [129:30]
E
Temu or Shein are those like really
- [129:33]
B
loved brands in the same way Marketplace DJI feels a lot closer to like a Nike brand or an Apple brand than Shein or Temu. Temu seems more like a Walmart brand. I agree with you. It's popular. Like, it's certainly brand recognition is there, but in terms of brand Admiration. I don't know. Can we play this video from Instagram explaining the new Call of Duty map format? They're getting into generative level design. It's not genai, it's not fully transformer based, but they're creating a whole bunch of different pieces of the pie and they remix them.
- [130:08]
E
Let's play it other like slabs.
- [130:11]
B
Look at this.
- [130:12]
A
And we're able to randomize that at runtime.
- [130:14]
B
When you're playing three slabs on the map, they randomly randomize them every time you load into the map. That is 100. We have the content to do upwards of 900.
- [130:25]
A
When you play a multiplayer game for
- [130:27]
E
a while, you have this sense of
- [130:28]
A
discovery each time you play a new map. But that kind of quickly fades.
- [130:33]
E
Right.
- [130:33]
A
You've played that map maybe five, ten times, you're like, I get it.
- [130:36]
E
I know all the places that never fades.
- [130:38]
B
With Kill Block. I see this as an existential risk. I think this might be the end the of. Of tvpn. I think that once this goes live and we're playing this in the Ultra Dome, we're not going to remember to go live because we're just going to be gaming too much. It's entirely possible.
- [130:53]
A
I don't know. I keep coming back to Rust.
- [130:55]
B
Yeah. If it ain't broke, don't fix it.
- [130:58]
A
I'm so happy. Like, Rust for me is like a vacation destination that you have so many good memories.
- [131:04]
B
Yeah. You don't want it remixed and you're
- [131:05]
A
just like, no, I'm good. Going to this beach, this place. And I'm happy to keep going back.
- [131:11]
B
He's a Luddite. He's a Luddite, folks. Anyway, we have Alexander Good. In the waiting room. Let's bring him in and see if he can hear us. How are you doing? Can you hear us now?
- [131:21]
G
Can you hear me?
- [131:22]
B
Yeah. Fantastic. How are you doing?
- [131:25]
A
Fantastic. Green screen background or zoom? Background.
- [131:29]
E
Yeah. It's the new AI. It's very powerful.
- [131:34]
A
Great to finally have you on the show. I've enjoyed your posts for a long time and some of your rants. You have some of the best rants on the Internet.
- [131:44]
E
Appreciate that.
- [131:45]
A
Been looking forward to this.
- [131:47]
B
Yeah. How would you introduce yourself these days? How much of just, I mean, I guess set the table for people? Give us a little bit of backstory and your thesis right now.
- [132:00]
D
Cool.
- [132:01]
E
Yeah. I used to be an army. I worked at Citigroup Palantir Ballet. As me started a company, sold it and then I got into crypto and I changed a lot of Obviously what I was working on and what many people were working on, and I was really just trying to figure out how to, you know, I got very, very pilled early on on the idea that I would just destroy people's work product. And a lot of people are getting on that train now. But I guess I was on it in 2023. So a lot of the rants were coming from processing that and going through my own AI psychosis. And now I'm like, post psychosis. Now I'm just in a Codex terminal working on building a protocol that hopefully can move the needle for people who own it, and also just our investors.
- [132:55]
B
A lot of attention, money, every. Everything flowed from crypto to AI. There was even that meme, like, if you're in crypto, pivot to AI. Has. Has AI strengthened your belief in crypto? Like, what do you think the intersection is? Because a lot of people jump to like, well, the agents will need. They couldn't possibly figure out stripe, so they'll have to use a token. And I've never really believed that fully. I think that there's some opportunity there. But it feels like the AI trends, you're a believer in the that, but also your faith in crypto broadly has not been shaken and you haven't like, pivoted or like gone to like the hot thing, which is not. It's just less common these days, I think.
- [133:36]
F
Yeah.
- [133:37]
E
I mean, let's be honest, the intersection of crypto and AI has not been a particularly good intersection so far. There's been a lot of, you know, hacking events. Every human piece of code that's ever been written is being evaluated by agents right now, which has obviously created, like an acceleration in hacking. X402 has kind of like gotten exploited. Right. So there's been, you know, there's this kind of, you know, you can prompt engineer agents to send you money, and that's like, not a good thing if you're running a protocol. So it's like, do you really want a protocol that can just like, send people money? I think the thing that's kind of kept me in the space or kept me excited about the space is I was always kind of fundamentally in this because of something I call the doom thesis, which is that that AI might not cause productivity to accelerate, it might actually accelerate the onset of the surveillance state, or basically a type of world where capital is going to flee the system. And I think that thesis is very much intact. I think that's why you're seeing a lot of crypto like zcash run very, very aggressively and. And I think that's going to continue to be a story. I think that's why Thiel moved to Argentina. I think it's sort of like it's in motion.
- [134:57]
B
Where are you on bitcoin versus other coins? You may obviously mentioned zcash. How strong do you think the bitcoin thesis is these days and how much maxiness needs to be considered in that thesis?
- [135:12]
E
The bitcoin thesis is getting weaker by the day. You have privacy coins accelerating in terms of their adoption. I think it's notable also the European Union and other countries tried to kill Monero and we're not successful at that. The Tornado cash rulings in the United States have given clarity to coins like Railgun, which are smaller but are good tech and Vitalik uses it and their big privacy summits. And bitcoin is fundamentally not private technology. And it also has the unfortunate leverage concentration for Michael Saylo dollar who has issued a lot of preferred and complex convertible financial instruments on top of it, which have kind of concentrated the ownership and the liquidity profile and now is kind of driving a market downturn. Right. So today's STRC drop as well as MicroStrategy's equity situation are driving down moves in Bitcoin and unfortunately, like, like, it's hard to see that structurally reversing.
- [136:16]
B
Do you know if there's been a material impact on crypto from not just the capital markets moving around and founders and employees pivoting to AI, but the actual mining operations pivoting to AI? Because a lot of the neo clouds that we know, they got their start in crypto mining and they're sort of out of that. That market. Is that actually affecting crypto in any way? I haven't really followed it.
- [136:43]
D
Yeah.
- [136:44]
E
I mean, it's interesting. Cor weave, you guys might know Vance at Framework, he had a fantastic return for his fund, partly because Core Weave literally pivoted out of crypto into AI and they're the OG pivot story. So they were like pre Jason Calacanis.
- [137:01]
C
They did it.
- [137:04]
A
Yeah. You know, the JCAL post, if you're in crypto, pivot to AI in like 2023.
- [137:10]
B
It was post ChatGPT, so it was a little bit late, but it turned out to be correct for a lot of people. A lot of companies launched very well.
- [137:21]
E
So, yeah, I mean, like, there is an interesting thing going on right now. You guys are probably going to hear about something called Perl relatively soon. Soon it's aggressively backed. There's rumors that Thrive is in it. And the TLDR is that it's a proof of work AI coin. And basically there's a paper that came out in late 2025 by SG Lang which talks about determinism in basically GPUs, which allows you to use AI to mine cryptocurrency. And so this is something my, my brother started working on with Ambient, which got backed by a 16Z. And then Perl is an Israeli team which is doing something kind of similar. They're using Matmul mining and they have a partnership with Together AI that got tweeted out. Essentially the argument that a lot of these crypto AI protocols increasingly are trying to make is something called useful proof of work, which is the idea that you have a miner that can also deploy a service. So my brother is delivering verified inference. And then the Perl argument is that you can basically buy GLM, not GLM, it's Gemma on together AI for a 25% discount because the miners are subsidizing it. Now the funky thing is that you can also buy it for 50% less than their subsidized version on open router. So you're kind of like market cap depends. You need this thing to really, really move in order for it to provide a meaningful subsidy. But there's an argument that, I think the compelling zoomed out argument is that TSMC, back in 2017 or 2018, 10% of their product was Bitcoin Asics, and now it's way less than 1%. And so if you're like, okay, you want to secure money with proof of work, very, very high level, it's like, why wouldn't you secure it with GPUs? There are arguments why you wouldn't, right? Because you're like. It's because you're literally burning GPUs, which are productive assets, to secure this thing, which is even worse than burning energy for bitcoin mining. But that's kind of a separate argument, right?
- [139:31]
B
How separate is this from the render token render network from Octane, where individual computers or folks with computer would render CGI frames and then earn a token. It was not nearly as decentralized. And then I know Prime Intellect was working on this as well, trying to sort of wrap spare compute and put it on a network that was sort of transacted with a coin intermediated by that coin. It sounds like this is at a much lower level. Is that correct?
- [140:05]
E
It's at a. Yeah, it's an architecture level, one render and a of lot of These coins, including Bittensor, actually are stake coins, right? So they're not actually secured by the operation of a gpu. So this is kind of like this pearl thing is a regime shift and it's trading at like a $2 billion FDV on private. There's token sales. So it's starting to move. It's a relatively new story in crypto AI. You know, my protocol is like a completely different bet. You know, my essential. I think the. There are some interesting intersections of crypto and AI that people are probably less focused on, which is that, you know, all of crypto is open source. So actually when you use Codex or you use these AI agents, they are like phenomenally good at, you know, crypto work because it's in the train, because it's in the training data and it's all open source. So if you screw something up, you can just be like, hey, can you just double check, check this with Railgun's implementation to make sure that we're doing the same thing. And then five minutes later you have a working orchard, shielded transaction on a
- [141:12]
B
tuning network, some Fortran code that's locked in some old financial institution and it's never seen the light of day. Not going to be in the training set whatsoever.
- [141:21]
E
Exactly. And so there's an OPEX efficiency argument in crypto because historically crypto has been a very wasteful industry. There's a lot of money spent on conferences and just the Ethereum foundation is a good example of how not to run a company. And it's sort of like there's a lot of waste in crypto and AI is. Crypto is the ultimate software industry where you're like, okay, I could credibly reduce engineering by 60% here.
- [141:50]
B
How are you thinking about AI psychosis? You said you went through it, you got to the other side like, what's the state of AI psychosis? Where's it showing up?
- [141:58]
A
Is it bearish if someone hasn't fallen in and then climbed their way out?
- [142:02]
B
True. A true Luddite versus the white pilled got through the trough of disillusionment is in the.
- [142:09]
D
No, not.
- [142:10]
A
It's the plateau of a trough of sloth, illumination, something.
- [142:15]
B
I don't know. But how are you thinking about it? Where is it cropping up? Where is it bad? What does it take to get out of it if you in it?
- [142:21]
E
Well, you kind of converge on this conclusion, Right? We start with a psychosis. And the threat of the psychosis is that my work is not meaningful and that everything that I've Done is basically meaningless because an AI will be able to do it better.
- [142:34]
B
Sure.
- [142:34]
E
And then you actually start encountering the tools and you're like, okay, in the areas I don't understand, AI is really good. But in the areas I deeply understand, AI is pretty bad. Right? There's not like a TVP where there's like two AI avatars that are doing a better job than you guys. I mean, Maybe the new A16Z1 is that and we just don't realize it.
- [142:55]
F
Right?
- [142:55]
E
It could just be that those guys are running a really, really advanced AI system. There's some evidence, I think there's some evidence of that. I'm not going to discount the possibility. But essentially you're like, all right, this is kind of directionally going to obsolete my work. And so I better be the first person to obsolete it. Right? Like, that's sort of where the race. So the way that you get out of the AI psychosis is you say, like, the only way out is through, right? So you're like, all right, how do I literally automate everything that I did before? And then how do I also just like, start to think about network effects that were impossible previously? So I'll give you an example. Like, you know, I'm working with. There is a pretty robust community with my crypto project, and we have what we call the data lake. And the data lake is basically like hundreds of contributors who are doing tasks all day. And then their aggregate intelligence gets kind of like summed up by an AI and it comes up with actionable trade ideas, right? And it's sort of like you're like, okay, you can make all kinds of alt data now. I think Palantir kind of of inspired me originally when they started talking about how well they're doing with corporates when they're like, hey, like, we have the biggest strat. You know, we have the best demand story in history to take unstructured data and turn it into structured data. And like, okay, that's cool in a corporate and government context, but like, nobody's ever done that with our Wall street bets, right? Or like, nobody's ever done that with trading information. And like, okay, like, I'm just going to do that. I'm going to take the Palantir ontology method and apply it to a group of individuals who create market alpha. Then that's how you get out of AI psychosis. You can identify what new capability does AI enable that allows you to generate the type of edge that you would have historically, just with a new data Primitive. It's like, okay, you need to have a story for how you use AI to get out. That's the short answer.
- [144:59]
B
Yeah, it's interesting. We're sort of at this moment where, yeah, the AI is replacing a lot of things. But I mean, at least as an entrepreneur, I don't know that the moats have actually changed because people are saying like, oh, like SaaS Apocalypse, your big pile of code is no longer a moat. But if you go back and look at the literature, whether it's 0 to 1 or Hamilton Helmer 7 powers, big pile of code was never a moat, it was always cornered. Resource brand, network effect, complex coordination. These other things that were not just, oh yeah, you were the first person to build a lot of momentum around a piece of SaaS and you've accumulated so much technology and software that that is your moat. It was always something else. But, but I don't know, how are you thinking about entrepreneurship outside of trading, outside of just like the startup etc
- [145:52]
A
before, I wanted to ask how you think market cycles are evolving. We were joking in Q4 of last year that we were like, great, like the bears win, like the bubble. The bubble did pop, right? It did feel like it's a big sell off. Yeah, deflated. You had a big sell off and then you got a new paradigm, that bear trap that came, you know, it came at the right moment and kind of reignited a lot of, you know, excitement and, you know, revenue and all the good things that you want to see during a technology cycle. But it feels like, you know, everything's been massively sped up. But I'm wondering how your kind of mental model is for market cycles today.
- [146:42]
E
Well, the interplay of AI and the market itself is kind of accelerated. So the way that people interact with markets has changed. So they're now very commonly pasting a transcript or pasting a company's earnings release into an AI system. And those AI systems have biases that are quite separate from the biases that humans have. So I'll give you an example. Like the company ASML is extremely complex. Like, as a human listener to the earnings call, you would be asleep halfway through the earnings call if you paste that Into Cloud Opus 4.8, it loves it. It's like, wow, this is like the most interesting, cutting edge. Like, I don't even believe this is true. Let me web search that. You know, Claude is highly engaged with ASML in a way that a human kind of wouldn't be. And the really esoteric bottleneck.
- [147:33]
A
Well, there are, there are Humans that would be. There's just like a very small number in the world and now there's effectively infinite, right?
- [147:40]
E
There are. And so some of these stocks that are, that are pumping like hp, you know, like I used to trade this thing back when I was at a hedge fund. It is the most boring company in the world. And this guy, David Gills, who lives down the street from me, you know, it's like we were trading meme stocks and now he's trading hpe. Right? And so it shows you, you know, this is like a bottle back. 70 year old guys. Nothing against wearing suits or anything, but it wasn't an exciting story. Now it's like, okay, now it has triple digit data center growth and the things that are going up are kind of historically stocks that retail would have never thought of. And a lot of them like sandisk and Micron. I know one woman who's like, I inherited an account account from my grandparents that had a ton of intel, micron and sandisk in it. I'm like, yeah, it's literally 75 year old. People were owning these stocks because they thought they were reliable bellwethers. And these are the things that are up 300%. I think just the market composition is really different. If you look at Coinbase, it's got substantial sequential revenue deceleration. So you know, people are not investing in Solana, they're, they're investing in HP. And so, you know, the sort of B2B focus of the markets has completely changed. And I think that's going to continue to be the case. Like I think, I think people are going to continue to engage with extremely complex equity stories in like a way that they just never would have before. And I think that benefits like a lot of, of for example, in my industry I think it is going to benefit coins with more complex technology stories. I think zcash is a good example where it's like, okay, there are encryption technologies that are moving forward, that are esoteric, that are hard to understand, that wouldn't have generated engagement before. But now the UX of the user interacting with is like I just copy pasted this into Claude and it says it's the best.
- [149:43]
B
Sure, sure, sure. How does Lore play into those stories? I feel like I heard, I don't know if it was zcash, but there was one about like they brought eight people together and they all had different pieces of the key and they were all like they brought a journalist with and there were different computers and they broke the computers and it was like building all this lore, obviously the Satoshi story is like extremely lore filled. Even the Vitalik stories story is like build something you want. And it feels like starting the perpetual motion machine of some of these projects is important. How does that change?
- [150:21]
E
I want to say lore becomes less important just because the stories that I just shared with you like HPE or some of the intel. Nobody knows who Intel CEO is. It's like, and it's like nobody is like invested in this story.
- [150:39]
B
Are you saying they still think Pat Gelsinger's around in the place or what?
- [150:43]
E
I think people just don't know.
- [150:45]
A
Right.
- [150:45]
E
I think that's the sort of real story. Is that Micron? Like I don't think people know what Micron does. It's like a trillion dollar memory bottleneck company. So I think lore becomes less important in one way. The more contrarian view that I have is that IP universes become a lot more valuable. And so for example, you know one of the. There's a lot of AI stuff about Harry Potter distills, right. So you can like extract 98% of Harry Potter out of Deep Seek or quad if you're, if you're motivated.
- [151:18]
B
Yeah.
- [151:18]
E
And so it's like it wasn't supposed to train on it. There was the big lib gen settlement with Anthropic, like they did train on it and there was a big multi billion dollar payout. But it's like that kind of multibillion dollar payout creates a strong incentive to kind of have IP treatment in a proper way going forward. And so Games Workshop for example has the cinematic universe which is World of Warcraft that becomes infinitely generative when you combine it with an AI system. Actually what you get from amd, why this is I think interesting is that you've seen like a big structural crowding out of entertainment because of the enterprise focus of AI. So anthropic, for example, like if you listen to Dario, he's like yeah, I'm not going to ever generate videos or images because like that's just a waste of time. Right. Like he's not. Or you know, OpenAI shut down Sora because they're.
- [152:11]
A
Yeah, we had, we had Mikey, we had Mikey from SUNO on today announcing a big round. But he has had the blessing of a lot of his would be the most intense competitors he could have have had to just deprioritize entertainment because it might be a $20 billion market instead of a $2 trillion market.
- [152:36]
E
Yeah, that's the interesting Jeevan's paradox argument in My opinion is that right now, for example, Nintendo is down 40% over the last year, or Take 2 is. Even though they've got a new GTA game, it's getting smoked. And there was a narrative that for a while that video games would be a big beneficiary of AI because intuitively we see all these, wow, what if all npcs were LLMs? Or what if you had in game generative content? And the reality is given the cost structures right now and the bottlenecks, you're like, okay, that's just not affordable. You can't generate real time generative characters and expect to break even on that because that would be $6 per game session and that's not a good investment. And as AI gets cheaper, that story is actually going to change. Right. So I think that's an interesting kind of where we're going is I actually think we're going to get a big acceleration in entertainment stuff because the cost of AI stuff is going to go down.
- [153:38]
B
Yeah, you'd have to imagine that the most mildly bullish take on Take Two would just be like, yeah, the next GTA game is going to take seven years instead of 10. Or our DLC is going to be pulled forward by three months and that's going to accelerate earnings. That seems very believable. But yeah, people have been caught up in all sorts of, oh, maybe you'll just generate it at all, maybe it'll all go to some foundation model. But that does seem further away, even when we see amazing demos from Google. But again, it's like six seconds here and there. Very expensive, not really persistent.
- [154:19]
A
How has Roblox done over the last year? Because that was always my thesis there. It's just, it's going to be a lot easier for somebody to figure out how to make, make compelling games, but
- [154:31]
B
down 50% in the last six months. Not good.
- [154:34]
E
Yeah, these things are very, very bleak.
- [154:37]
B
Very bleak. Shouldn't be. Anyway, hopefully they build back. I'm optimistic. I like Roblox. Well, thank you so much for coming on the show.
- [154:46]
A
Yeah, it was great to finally have you on. Let's do it again.
- [154:49]
B
Let's do it again soon. Have a good one. Cheers.
- [154:51]
G
Goodbye.
- [154:53]
B
There's one more announcement we gotta share. Anduril and NASCAR are teaming up to sell a $14,000 VR racing rig simulator. Palmer Luckey's back in the VR business. I mean, he already was with the Eagle Eye headset, but he's back in the consumer entertainment VR industry with this. Of course, he's not making the Actual headset.
- [155:12]
A
I believe it is a really three days after I just put down a deposit on my own sim.
- [155:18]
B
Really?
- [155:18]
A
Yeah.
- [155:19]
B
Did you actually.
- [155:20]
A
Yeah.
- [155:20]
B
No way.
- [155:21]
A
But I don't think that's going to be. That's going to be very specific, I imagine for I imagine NASCAR simulation.
- [155:31]
F
Right.
- [155:31]
A
Because that's where they're focused. But very, very cool. Matthew Prince over at Cloudflare says, well that happened faster than I predicted thought it would be. End of 2027 then early 2027. But it just traffic growing so fast that bots have now passed human traffic online for the first time in the Internet's history.
- [155:53]
B
That's crazy. Lots of agents running around on the Internet.
- [155:59]
A
Yeah. Bots are now, according to Cloudflare, 57.5%.
- [156:04]
B
Every time you fire something off, it's hundreds of pages. You see it in the reasoning traces and the tool calls. Let's go watch this demo from Eric Lyman, friend of the show, sponsor of the show. Introducing Stack, the AI operating system that lets accounting firms take on more clients without hiring. Learns your firm's process runs the close posts, the journals fully auditable. We're living through the biggest shift in accounting since the spreadsheet. That's a good way to frame it. And they did a very cool video for this. All practically produced. I believe this is all shot with an actual camera. They got these actual pieces. They filmed this. Not cgi.
- [156:44]
C
Not and accuracy are not negotiable.
- [156:48]
B
These one off experiments.
- [156:49]
A
Such a calming voice.
- [156:51]
B
I love Eric doing the vo. It's great.
- [156:54]
C
One secure place to orchestrate AI co workers for accounting from day one.
- [157:00]
B
This has been the vision of ramp since they launched and it. And it feels like, oh, this is the moment. Of course they have to do an AI thing but they've been pitching this exact thing since what, 2020, 2019 and working on it longer. Honestly, you go back to parabens and it learns over time. Always giving you the final say before anything gets posted.
- [157:24]
C
And every action it takes is fully recorded and auditable. The more you build, the lighter the work gets and the more clients you
- [157:33]
B
can take on Stack Ram Stack. They're saying it's God mode. It's God mode for the counter. It's literally God mode. Well, go check it out. We should also watch this one last video from Martin Scorsese Black Forest Labs. I alluded to it earlier. I saw this on Instagram and I really enjoyed it. 30:30 seconds of Martin Scorsese, storied filmmaker. Jordy, name your favorite Martin Scorsese movie. Doesn't feel modern. A town, not a village, not a city.
- [158:08]
H
Mafia.
- [158:08]
B
The movie Mafia. Yeah.
- [158:10]
A
Isn't there.
- [158:10]
B
Is it the Godfather? There is a movie. You're thinking the Irishman, maybe that is like a mafia movie. Gangs of New York. Or maybe Casino. Or Goodfellas.
- [158:26]
A
Goodfellas.
- [158:26]
B
Goodfellas. Anyway, play this. Sorry, I was not paying attention.
- [158:30]
A
Let's try it and we'll go from
- [158:32]
B
what you think you need a place that doesn't feel modern. A town, not a village, not a city.
- [158:37]
D
Almost medieval.
- [158:38]
B
Even the streets are narrower, cobblestoned.
- [158:42]
D
The main road through the town is twisting and turning.
- [158:45]
B
Put the camera higher, looking down.
- [158:48]
D
DeMille would have his production designers do oil paintings. This is that, in a sense, conveys a cinematic. A cinematic intelligence.
- [158:57]
B
Cinematic intelligence is a good tagline for Black Forest Labs. I thought that was really, really good. And yeah, what a way to like, you know, Hammer. Obviously, filmmaking's deeply controversial, but you get Martin Scorsese talking about it and he's at least going to perk up people's ears and they're going to listen to what he has to say and think about it. Is it a tool? Could it be useful in a workflow? Could it speed something up? Is it going to make the next Martin Scorsese movie? Probably not this year. But will he potentially be using it when he's thinking about what to work on next? Sure. So, fun, fun project and very cool video from Black Forest Labs. Anyway, tomorrow we have a special show, but we will still see you at 11am Pacific Sharp. Sharp. And leave us five stars on Apple podcasts and Spotify. Sign up for our newsletter, tvpn.com and we will see you tomorrow.
- [159:54]
A
We love you. Have a wonderful evening.
- [159:56]
B
Flashback.