← This Week in Startups
Loading summary
Transcript108 lines
- [00:00]
A
OpenAI is partnering with Disney. Sora was a potential IP catastrophe for OpenAI. Litigation concerns, worries about payments. Well, they've ironed it out. So Disney is going to invest $1 billion into OpenAI. It's going to have warrants for future purchases. And I think this is a big damn deal.
- [00:16]
B
If you're not part of one of these deals. As Disney sent a cease and desist letter to Google the day before the OpenAI deal was announced, you cannot use these deals, these IP. So it's really, I think, the first attempt to figure out a sort of formalize this whole process. Like here's what these collaborations will look like and if you don't have one of these collaborations, you don't get to use these characters.
- [00:39]
C
So the most important thing here is now that there's a commercial relationship, every other lawsuit that's going on will be able to point to this. The New York Times versus OpenAI, Disney versus Google. They're all going to be able to point to this commercial arrangement and say that there is a cost to people stealing their content because this is a new business opportunity. It was essential that somebody get a deal like this done. Now that we have a major deal like this, every single LLM is going to have to do this.
- [01:08]
A
This week in startups is brought to you by Crusoe Cloud. Crusoe is the AI factory company. Reliable infrastructure and expert support. Visit Crusoe AI build to reserve your capacity for for the latest GPUs today. Devstats. Devstats integrates your dev work and your business goals into a shared language that everyone can understand. Get 20% off plus access to their dedicated Slack channel. Just go to devstats.com twist and LinkedIn ads. Start converting your B2B audience into high quality leads today. Launch your first campaign and get $250 free. When you spend at least 250. Go to LinkedIn.com thisweekinstartups to claim your credit.
- [01:56]
C
Hey everybody, welcome back to this week in Startups. There's a lot in the news. I've got Alex Wilhelm and Lon Harris with me. What's the first story? Let's get to it within 15 seconds.
- [02:07]
A
OpenAI is partnering with Disney. We talked a lot on the show that Sora was a potential IP catastrophe for OpenAI. Litigation concerns, worries about payments.
- [02:17]
B
We.
- [02:17]
A
Well, they've ironed it out. So Disney's going to invest $1 billion into OpenAI. It's going to have warrants for future purchases. It's going to be a three year deal to use certain Disney IP inside of Sora. And I think this is a big damn deal. Lon, what did I miss?
- [02:31]
B
Oh, there's a lot. There's a lot going on. I mean, I think we're seeing the first effort to sort of codify this relationship. Here's how it's going to work, here's how the deals will work. And if you're not part of one of these deals, as Disney sent a cease and desist letter to Google the day before the OpenAI deal was announced, you cannot use these IP. So it's really, I think, the first attempt to figure out a sort of formalize this whole process. Like, here's what these collaborations will look like and if you don't have one of these collaborations, you don't get to use these characters.
- [03:03]
C
Who could lon have ever seen this coming? Well, who could have seen this coming?
- [03:07]
B
Notably over the summer. It was back in July that Jason made a very similar pitch to this. He made it here on this Week at Startups. We talked about it on all or he talked about it on all in as well. I believe we do have Marcus this clip, if we could show it, of Jason making his Disney prediction.
- [03:26]
C
Here I am on the all in podcast talking about it. I was speaking at a KPMG conference where I do my Nostra Kannis, my predictions for the next year. And one of those predictions was somebody would create licensing. And when I always talk about licensing and respecting ip, especially on the all in podcast where it's three other technologists, they always give me a hard time because they're like, we don't need to respect ip, essentially paraphrasing. And I said, well, here's what it could look like if OpenAI licensed the Disney characters and you authenticated your OpenAI app with Disney 40 seconds or so from the All In Pod, the Disney Channel. Imagine you could make yourself into a Jedi Knight and you could then upload your photo. You know, kids might really get into this. You upload your photo, you can make. You talked about this friborg a couple of times of the future of narrative storytelling up your photo and then it makes you into a Jedi Knight. There's. There's Darth Calacanis.
- [04:20]
A
So that looks to me like you're.
- [04:22]
C
Infringing on their trademark. What's that?
- [04:25]
B
Are you infringing on their copyright?
- [04:27]
C
This is fair use. This is fair use. This is a perfect example of fair use for editorial. You're also infringing on some Ozempic. That's absolutely.
- [04:36]
A
I am definitely bridging on.
- [04:37]
C
What you're witnessing here is anytime I make a great point on the all in podcast, they personally attack me and they make it ad hominem, which is when you know you've won because all they can do is make fun of your weight or make this false claim that you're infringing on it when you're doing an example.
- [04:53]
A
Can you explain that, though, Jason? I think it's an important point. Why was that Fair Use and what does. What does now OpenAI? What are they able to do that is more interesting with this deal?
- [05:03]
C
If I were to show on a podcast, here's an example for education or commentary purposes, that's totally allowed under Fair use. For example, I could take a clip from any TV show right now, play the clip for 30 seconds, 60 seconds, just a portion of the show, and say, here's what I think of the lighting or the dialogue from that show. It's not infringing on their ability to make money. And it's commentary. So commentary is protected under Fair Use. That's why people can, I don't know, do a song reaction video. It doesn't interfere with the ability for that person to monetize parodies.
- [05:37]
B
Like what we do on honest trailers is allowed because we are creating something new from the trailer. It's not just.
- [05:44]
C
And that would be parody. There's another section of Fair Use that's for parody. Anyway, you know, David, is. The AIs are. The AI companies don't want to pay for IP. They want to get a free pass. So it is not surprising that anybody who is in the AI space doesn't want to see these things happen. But I did predict that this would be an opportunity and that one of the savvy companies would jump the fence and say, and I think I said this many times on this part as well. If you got a 500 billion, a trillion dollars company, what's a billion or $2 billion here to license some content. In fact, I said 30% of consumer revenue should go, 40% of consumer revenue in these systems to content creators, which will then make the content sustainable. What's great about this deal is now Sam Altman, because he has a commercial arrangement with Disney. If anybody else uses those Disney characters globally where there's IP protections, he can go to Disney and say, disney, remember, we're paying you for it to license these. And I didn't see in the deal how much OpenAI is paying for 3 years of this. I did see that Disney has the opportunity to invest a billion. They didn't specify it, but there is a payment.
- [06:57]
B
Yes. Iger said specifically that, you know, they're not using any human faces, they're not simulating any actor voices. So you could put, you know, you could put Kylo Ren with his helmet down in there, but you couldn't put Adam Driver as Kylo Ren.
- [07:13]
C
Right. Or Adam Driver's voice as Kylo Ren.
- [07:16]
B
No, Adam Driver's voice as Kylo Ren, either. You can use the image.
- [07:19]
C
That's a separate issue which we'll get into, which is protecting the artists. Yeah.
- [07:23]
B
So he's also saying, as part of that, he was also saying, and if you are a creator who say you designed Elsa from Frozen, you will also get paid via this licensing fee that they're going to charge. But we don't know exactly what that licensing fee is.
- [07:39]
C
So the most important thing here is now that there's a commercial relationship, every other lawsuit that's going on will be able to point to this. The New York Times vs OpenAI, Disney vs Google, they're all gonna be able to point to this commercial arrangement and say that there is a cost to people stealing their content. Because this is a new business opportunity. It was essential that somebody get a deal like this done. Now that we have a major deal like this, every single LLM is gonna have to do this. Now the counter argument by the people in Washington D.C. the law AI lobby and the other AI companies is they're going to cry, we have to beat China. Under no circumstances do you get to say, I get to steal IP in order to beat China. The American way is to respect ip. Once we lose that, we lose everything. If you lose IP protection in this country, you will lose everything. It's in the best interest of our country not to try to beat China or use China's stealing of IP as some weapon against ourselves. We want to have a united front with China and other countries. This is where long term thinking is critically important for David Sachs, as the AIs are for Nvidia, for Grok, for Elon, for Claude, for everybody to think long term. Do we want our IP protected in other countries? If we do? You gotta protect Disney's if you want to protect OpenAI's. So Sam doing this is a gift to America and Bob Iger doing this is a gift to American exceptionalism and protecting our IP long term. That's fantastic about this deal. It was obvious this would happen and I got lucky by just picking the best IP in the world, Disney's or the Most recognized as an example that within 100 days it actually happened. I got lucky on this prediction.
- [09:31]
B
Well, the ownering of Pixar, Marvel, Star wars it makes so they have the most AI text to image friendly catalog of any company on earth. Yeah.
- [09:43]
C
If you're selling a business to business product, you're looking to reach professional customers who are doing business. And the question you have to ask yourself is where do those business leaders spend their time? And the answer is obvious. LinkedIn. And these are the real decision makers. So when you advertise on LinkedIn, you're not wasting your money. You're going to target your ads based on job title, based on industry, based on skills. So maybe you want a CFO in the technology space who has been in their job for five years. Right. You want to zero in on those demographics and behaviors. LinkedIn has over a billion members and 10 million of them are C suite executives. They're going to help you reach exactly the right person with the authority to pay for your product. Right. Go right to the top. Use LinkedIn and we've got a special offer just for twist listeners. Launch your first campaign and get $250 free. When you spend at least $250, go to LinkedIn.com thisweekinstartups to claim your credit. And now they could take Pixar and what the Disney, what Disney should do next is they should say Pixar is available for your LLM to the highest bidder and they should go to Sergey and Google and say Sergey, would you like to license for Nano Banana all of the IP or these 200 characters from Pixar? These 20 Pixar films are now available. Then they should take the Marvel characters and they should go to Claude and Anthropic and say we have the Marvel characters available. Would you like the Marvel ones as you go around?
- [11:18]
B
OpenAI's deal is over 200 Disney, Marvel, Pixar and Star wars characters. So they're not opening the floodgates. They've got a custom set of characters that they've contracted. So presumably there would be a ton left that aren't. And Disney has way more than 200 characters.
- [11:35]
C
So Lon, maybe you could explain Hollywood's concern. So we just went through the chessboard here in Silicon Valley and how everybody doesn't want to pay for IP and now we have dollars to donuts. 20 more of these will happen. I'm going to predict it right Now. We'll have 20 IP deals in 2026. Put that on slash bets. I'LL put the over under on 20 major IP deals by the 6 or 7 LLMs against the IP.
- [12:03]
B
I think there's two really outstanding big questions that remain after reading this that are going to be a concern for everybody in the creative sort of field. The first is what do these license fees look like and how are creatives being compensated? There's a long, long history in Hollywood and media of promising that creators are going to get a little piece of the pie and then they don't actually get cheated. We've seen this a ton with Marvel movies where all the guys who created all these characters, you know, like Steve Ditko's estate isn't making millions every time a new Spider man movie comes out.
- [12:35]
C
Because they were full time employees who signed work for hire agreements, which by the way, was an economic decision they made to have full time employment to not be independent creators.
- [12:45]
B
Yes, I mean, it was very hard to make a living in comics unless you were getting a contract with one of the big players. But, but you're correct that their deal was, look, Marvel is going to own the Hulk. I don't own the Hulk anymore. And so that's why like Jack Kirby's estate doesn't make a killing every time there's a Fantastic Four movie that hits theaters or what have you. So that's one concern is like, are these license fees going to be, you know, like pennies every time, every year for the rights to these characters or is it going to be a meaningful payout to creatives? So that, that's definitely like a huge big picture deal. I think, you know, you know, there's, there's a lot of sort of vagueness. I think the guardrails would be the other one that OpenAI and Disney are both saying. We are going to have guardrails. We are going to have protection so that you can't do explicit things, inappropriate things, things that would violate the ip. But I think that's a. How are they going to do that? Is that going to be effective? And especially as we start integrating more and more work by creatives, I mean, for now it's not that huge of a deal because, yeah, you can't get Daisy Ridley as Rey in Sora. So you couldn't violate Daisy Ridley's, you know, prize.
- [13:53]
C
Correct me if I'm wrong, Lon, the person who voiced Darth Vader, his estate.
- [14:00]
B
James Earl Jones sold.
- [14:01]
C
James Earl Jones sold in perpetuity his voice as Darth Vader. And they can clone it now. So that deal has also occurred in the world.
- [14:12]
B
And we've seen a lot of those kinds of deals. Like Sir Michael Caine's recent deal with 11 labs is the same way. Like, they're going to be able to use him and his likeness, his voice likeness for whatever projects they like moving.
- [14:23]
C
Okay, so that's between him and 11 labs, but not the characters he played. Disney did one specifically. They don't get James Earl Jones in all uses. They get him just for Darth Vader. So what we're going to see now is the second wave of these deals, which will be different. Artists might have in their contracts already that their voice can be used. They might not, but a voice performance is critically important. And we saw this first in Back to the Future 2, where Crispin Glover was used without the permission somebody who was a lookalike. Not.
- [14:57]
B
Yeah, they found a lookalike and they had an impersonator do a voice that made it sound like it was Crispin Glover, but it was not crispinverver.
- [15:03]
A
Sketchy, sketchy, sketchy, sketchy.
- [15:05]
C
And he won a lawsuit for that. Okay, Alex, what else do we need to know about this deal and then we can move on?
- [15:10]
A
Yeah, a couple of things. So Disney will invest a billion dollars, as we said. It will also get warrants, Jason, into OpenAI. So if it wants to purchase more shares down the road at whatever price they've decided that's going to be in play. It's a three year deal, one year of exclusivity, which means that I don't think another company can land a similar contract with Disney in the near term. And I will say that I think the timing of this announcement and the suit against Google is rather interesting. I wonder what Google's going to do to get out of the penalty box here.
- [15:38]
C
Did we describe the suit with Google yet?
- [15:41]
A
Lon brought it up. So essentially, I'll just give everyone a quick summary here. There is a cease and desist letter sent by Disney to Google before this OpenAI deal was announced. The actual letter was reported a little bit afterwards, but it's all pretty contemporaneous. And what they say is that Google is doing copyright infringement on a, quote, massive scale by copying a large corpus of Disney's copyrighted works without authorization to train and develop generative AI intelligence. They're exploiting our information. And they said, quote, google operates a virtual vending machine capable of reproducing, rendering, and distributing copies of Disney's valuable library of copyrighted characters. Google, I think, in an attempt to lower the temperature, said, we have a long relationship With Disney, we're going to sort it out with them. But reporting also indicates that Disney's been beating down their doors and, hey, you're not doing this right. And Google's trying to dodge. I just think that maybe OpenAI and Disney came to such a helpful, healthy and positive agreement here that down the road, no company is going to be able to say, oh, we can't afford to do this, or it's too punitive. Jason.
- [16:42]
C
Correct. Which is what I said earlier. Now you have the precedent, and Google is a good actor. And this is different than indexing the web, taking a small snippet and using it. The output is now protected ip. It's indisputable. If you make a character and you make a profit building that character, you're now infringing on Disney's ability to do that. So every single IP holder needs to do a deal like this. They need to plant a flag so that if you're the New York Times and you have Wirecutter, Wirecutter needs to immediately do a deal with Grok or Claude. Consumer Reports needs to do this deal with somebody. And Consumer Reports, as a nonprofit, makes probably low tens of millions of dollars, I believe. I was thinking about buying it at one point. Then I was informed, it's a nonprofit. They'll never sell. Although we've seen some nonprofits go for profit. If Consumer Reports did a deal with Grok or Claude or Perplexity or somebody, they could then say to anybody else who outputs Consumer Reports data, no matter where you got it from, that they are breaking their ability to do commerce. It is the IP's holder. The IP holder has the right to monetize new forms and new media types so you don't have to anticipate what those are. If VR becomes one, if AI becomes one, if neuralink becomes one, and just that will be the next one. Incepting a Jedi story in your brain using one of these, and you ask Neuralink or one of the other people doing this, like, is that infringing? Well, yes, because Neuralink will become the new Disney World. And so in five years, Disney will do a deal with Neuralink that you will be able to experience any of the Star wars movies. I know I'm sounding like a crazy person here, but doing this with AI would be considered crazy. Like 20 years ago, if you said, oh, you can put yourself and make yourself a Jedi, you're going to be able to play Luke Skywalker in your mind. And that will be incredible. IP to be licensed for it. And in fact that is the premise of the incredible film. By Kathryn Bigelow Strange Days Keeping your business and engineering teams in sync can prove to be a considerable challenge. Trust me, I know. And you can't build a truly world class product that your customers will fall in love with unless these two teams are communicating perfectly. That's why I'm always recommending founders check out Dev stats. They translate your dev work and your business goals into a clear shared language that everyone on your team can understand and make plans around. That's going to keep your team focused on the kind of high impact work that truly matters for your startup. We talked to a company that used dev stats and discovered that nearly half of their development time was going to bug fixes and other inessential internal tasks, adding zero value for their users. That level of insight is invaluable for a founder. Check out Dev Stats today and get 20% off plus access their dedicated Slack channel. Go to devstats.com twist for your 20% discount. That'S-E-V-S-T-A-T s.com twist.
- [19:59]
A
This is one reason why I'm so excited about BCIS or Brain Computer interfaces, Jason because it does unlock an entirely new form of interactive entertainment that I don't think we can really do right now because we're not using our brains to generate scenes, we're just kind of using.
- [20:11]
B
Strange Day is kind of a dystopian take on it. I feel like there's like Ready Player One sort of the more the more fun.
- [20:17]
C
Absolutely, yeah.
- [20:18]
B
Before I drop off, there's one more story from the world of AI and streaming that I thought was very interesting and I wanted to bring up before I leave you folks. So in two weeks Amazon Prime Video has introduced AI features and then almost immediately pulled them due to backlash and them not working correctly. Last week they were automatically dubbing anime films and shows into various languages using AI, but it wasn't working very well. The AI performances were very poor. Sometimes the dubs would go beyond the parameters of the scene that they were in, so it didn't really make sense. Audiences did not like it at all. So they pulled it all back. And then this week, just the other day, they uploaded AI recaps for a lot of their most popular shows. So Fallout, that video game adaptation is coming back next week for season two. So they uploaded this AI generated season one recap. The problem is it was full of errors. People noticed a lot was wrong. It described one scene as taking place in the 1950s. It was taking place in the year 2077. It actually misdescribed the final scene of the season, which is a super important thing to get right if you're a recap getting people ready for season two. So after all these complaints, they've now pulled back the second wave of these AI recaps. So they're all gone. They had one for Jack Ryan, they had one for the rig. They've pulled them all back. And it leads me to wonder, why is this happening? Why is Amazon pushing these AI apps forward that don't work right?
- [21:53]
C
I think we've been given permission and society and consumers have accepted that AI is imperfect. And it gets 80% right, 90% right, which then leads people to say, all right, let's go for it. And people are over trusting AI. Here's a clip that I just shared in X. And you can see an example of this.
- [22:12]
B
Yes, this is from the anime Banana Fish. This was the most famous example.
- [22:16]
A
I need space to move.
- [22:18]
C
I said I need space, you overgrown son of a fish. This way.
- [22:25]
B
Don't let him.
- [22:27]
C
After him. I read all your moves. I saw right through you. No, don't do it. Seemed okay to me. I wouldn't have known.
- [22:38]
B
Yeah, I mean, the. The performances are a little flat. They're not. They're not synced correctly, so it's a little hard to tell who's talking.
- [22:46]
A
Don't die on me, Skip.
- [22:47]
B
If you're used to a professional editor and voice actors doing this, it's. It's lower quality than anime fans have come to expect.
- [22:55]
C
So I can give a judgment on this. There are times to do this and times not to do it in high art. Which anime would be for people who consume it. It's important that it be right for an article in the New Yorker. I just heard the editor of the New Yorker talking on the Semaphore podcast, I believe, and they were talking to him about a debate they had at the New Yorker. And the New Yorker had a similar debate. We want people to be able to listen to articles. A lot of people like to listen. People have gotten used to this with Audible and podcasts. So to get a voice actor and to do one of those New Yorker 7000 word articles, you know, 4000 word articles, it takes weeks, and then you have back and forth. Did you pronounce this word right? You pronounce. Even the voice actors pronounce words wrong. They get context wrong grammatical stuff, pauses, et cetera. And they said, well, the AI is going to get some of those things wrong. Too. But if we put this as AI generated and we have it on the page when we publish it, consumers will be okay with that because they're doing it themselves. In fact, I love Speechify, and one of my hacks is when there's a long story. You know, in the mornings, if I'm going rocking on the. On the ranch, I tell my team, hey, any of the long stories, just throw it into my Speechify. I open my app up and I'll listen to, for example, the executive order. And if it mispronounces something, I don't care. I want the convenience of it. So it's going to be up to people to make a judgment call and then inform consumers. I feel like disclosure is number one and judgment is number two. Consumer. And it's going to get better. So we're going to be talking about this for, I don't know, three months, three years, but we won't be talking about it five years from now.
- [24:44]
A
Thank you. Thank you, Lon. All right, see you, folks. Next up on the docket. Jason, let's talk for a minute about this executive order. I think it's very important that people understand what's in it, what's not, and what impacts it may or may not have on the national AI conversation. So really briefly, everybody, after a lot of signaling, the President did sign an AI focused executive order this week. And what matters here is that it's going to do a couple of things. One, it's going to shape future congressional legislation. So the EO demands that we put together a framework for what Congress may or may not pass down the road, essentially kind of guiding Congress towards what it thinks is the correct way to regulate AI. Light touch approach, etc. It puts pressure on states that are currently or have already regulated AI. This takes a couple of different forms, one of which is the formation of an AI litigation task force that will challenge laws that don't comport with the eo.
- [25:34]
B
And.
- [25:34]
A
And we're also going to have the Commerce Secretary go in there and figure out which of those laws might fall afoul. Then we're going to have a lot of authority for the special advisor for AI and crypto, which is David Sacks. He's going to have a way to weigh in across these procedures, plans, whatever you want to call them. And what that's going to do is put an industry ally, a former industry member, kind of in the driver's seat for dictating which states need to be argued against and so forth. So it's a really Big win, I would say for folks who think that the technology industry has the right of this. And if you don't agree with that, you're probably a little bit bummed out. But Jason, the temperature that I've seen in and around social media is pretty excited from tech folks.
- [26:15]
C
Tech folks will be excited about this because, yeah, it's a pain in the neck. To have gambling laws across 50 states has been a bummer for online poker, for prize picks, for, you know, other wagering sites. It is nice to have a national law. That being said, we are the United States of America, not America. And so we do have to recognize that states have specific rights and can interpret laws and deploy them in the best interest of the people living in those states. So Sachs did point out four areas where this would not apply. Child safety. I guess they're saying, hey, child safety is a state issue. Communities, you know, maybe they don't want to put data centers and have their energy spike potentially, or pollution from them. Okay, great creators. Copyright law is already federal. I'm reading from Sachs tweet, so there is no need for preemption here. It's already. So that's great. And it's already playing out in the court as we discussed in the first two stories. Censorship. I, I'm torn because AI is everything. AI means everything. What I think is more interesting than my opinion or yours on any specific ones here is what's going to happen in 2026 at a meta level. And we are shaping up to see the fight of our lifetime between AI companies and the technology industry and rank and file Americans. We actually had a big discussion with Tucker Carlson sitting in for Friedberg. We had a big debate on this in the MAGA world. And MAGA was essentially created by Steve Bannon and Trump. They're the architects of it. When the MAGA originalists look at what when they see Nvidia and Google and Tim Cook and everybody in Washington D.C. you know, cozying up with the administration, they say, well, what about us? Where's the, where's the working man? Where's the working woman? Where's. Where's our jobs? Where's generation tool belt, et cetera? And you're going to hear this play out in 2026, into the midterms and certainly into the 2028 national election. And I pulled a couple of clips here. I think it's worth us getting ahead of the curve here like we did on the IP issue on twist. Let's get ahead of the curve here and you could maybe tee up these two clips and I'm interested in your take as well.
- [28:48]
A
So first up, we're going to hear from Joe Allen. He is an author of a book that argues against transhumanism. And this is a book that Steve Bannon wrote the intro for Jason. So clearly there's simpatico here. This is him discussing the issue on I believe War Room, which is Bannon's podcast. I'm not a listener, but I believe I have that right. Let's take a listen.
- [29:05]
D
What is the objective of the companies that people like David Sachs and Ted Cruz are running cover for? What is this supreme intelligence that the President is talking about? These companies are driving towards artificial general intelligence. And whether you believe that's possible or not, what it shows is they are willing to create something that would replace every worker on Earth. Whatever percentage of that objective they achieve, it will be a net negative for any working American or any worker across the world. So what are we fighting against? We're fighting against people who not only don't care about your well being other than to use you for training data or to use you as a pet after their singularity. These are people who actively want to replace you. They want to mine your data, replicate you in silico and replace you. These are the people we're fighting against and we'll fight against them the rest of our lives.
- [30:13]
C
Launching a new company is all about finding your first customers and then just learning how to solve their problems. And that is going to put you on a relentless pace. And that means you're going to be releasing new products and features, hopefully at a brisk pace or better. But my lord, the complex complexities of working with AI, it's going to slow your developers down. They're going to have to install new GPU drivers. They're going to be provisioning clusters. All of this detail oriented, labor intensive work can distract you from working on the thing that you're building your product. But now there's Crusoe. They are the AI factory that worries about your back end so you can focus on your code. And their hardware has been designed specifically for AI development. So they're going to make sure sure that your models run with unparalleled performance every time. Head to Crusoe AI build to apply and receive $100,000 in credits for virtualized Nvidia GB200 NVL72 on Crusoe Cloud pending availability. Okay, that's C R U S O E AI Build to apply and get those hundred thousand dollars in credits.
- [31:23]
A
So the argument here, Jason, that people are being forced to essentially submit their information, their training and then become replicated by computer systems and then replaced by them. That does not sound like conspiracy theory to me. That sounds like it's dead on. On the Wednesday show I joked that AI fixes the problem of wages because it replaces people. So my question to you is, when, when we have these conversations, you and I are pretty aligned on this. Why? Why is this a controversial perspective amongst the people that are, you know, currently praising the aieo?
- [31:53]
C
So there is a strong group of accelerasts and there is attacks on deceleras. People have called me a decel at times, but my whole life in technology, investing in companies, you know, objectively I'm an accelerist obviously based on what I do every day for a living. And I take pitches from founders and for 20 years, founders, when they pitch their companies, you know, I'd say 10% of them 20 years ago would talk about eliminating jobs or they would talk about 90% of the time, they would be talking about productivity gains. 90, 10, these days it's 50, 50, 50% of folks say, hey, we've got a technology that's going to make accountants, lawyers, salespeople five or ten times more efficient. And we're going to do that with AI. We're going to make designers more efficient, we're going to make writers more efficient. You know, Grammarly makes writers more efficient. Superhuman. Makes cleaning your email box with AI faster and reclaims time for executives and business people. But a full 50% when I get pitched, say we're going to replace the customer support department. We're going to build one amazing pilot that replaces all pilots. We're going to make one driver that's so good, the Neuro driver, the FSD driver, the Waymo driver, that singular driver will be so good trucking driver that it replaces all of them. And what's happening now is the 47th administration is avoiding this topic at all costs. And the MAGA movement is so upset about this that they are vocalizing it. And so you have a civil war happening within conservative circles where Steve Bannon and J.D. vance, let's say, who might be very pro individual worker and very concerned about these things, are now being pitted against, you know, the technologists who are in the White House who are pushing for a national regulation so that, let's say self driving or you know, teachers could be replaced with this technology. The fact is that will happen independent of all this hand wringing.
- [34:12]
A
Go back to last year. You and I are Talking about Trump, we're talking about immigration, we're talking about what technology wants. And I was saying at the time that I think, you know, I think Bannon and MAGA are pretty much on the other side of a lot of these issues. And it was going to be interesting to see who was going to win out at the start of Trump too. You know, we have Doge going on, we have, you know, Sachs getting the gig. And it really seemed like the technologists were running the show. Now that's temperatures come down a little bit and we're seeing, I think, more dissent in the GOP about these issues. Governors like the governor of Florida, governor of Arkansas, saying, hey, whoa, whoa, whoa. Don't you take away our ability to, to set things up the way we see them fit.
- [34:48]
C
You're talking about for states having rights. Yes, the states will reclaim their rights. Independent, like the Republican states are going to say, hey, no, no, wait, wait, wait, wait, wait. We want to decide if cannabis is legal here. We want to decide abortion. We want to decide AI and self driving. So if states get to decide right. And the Trump administration said, we're going to overturn Roe v. Wade, we're going to make it a state issue. So you have this, okay, Roe v. Wade will become a state issue, but we want to consolidate power federally for AI. Make they make a strong argument for it. Here's what will happen in 26 to 2028.
- [35:30]
A
Okay. Because this is what I'm really curious about. Yeah.
- [35:32]
C
You're going to have this debate and we're going to need to see concessions. When I run for president, it's going to be based on three things. AI is going to be fantastic at solving these three problems. Education, health care and housing. Education, Healthcare. Housing. Education, healthcare and housing. Eh. Okay, how does it help education? That's the easiest one. Custom tutor. You go online, it teaches you a language, it teaches you guitar, it teaches you how to fix your dishwasher. Now you're a dishwasher repair person. Literally, I could take any person with a ged, give them chatgpt on a phone, teach them how to use it, and they will solve 90% of dishwasher problems and have a really great job. The 10%, they can't, they got to call in somebody else. But you'll be able to send, you know, a first line dishwasher repair person, refrigerator repair person. And because AI is going to very quickly find the manual, find the model number, find the fuse box, the reset codes, whatever it is, you could do that. So education is Going to help and we're going to lower the cost of education and we're going to come up with ways to make it free or close to free. And the market's just going to do that naturally. Healthcare. We need to figure out ways to use this technology to make health care cheaper and cheaper and cheaper. How do we do that? Well, full body scans, which Daniel Ek has a company I forgot the name of, they're doing three or $400 body scans. Pronovo is like two or 3,000. Does a little bit of. Does a much better one.
- [37:05]
A
Daniel Ecospect Necco Body Scan.
- [37:08]
C
Okay. Necco Body scan is like the more accessible $300 version. Prinovo, the one I use, a little more expensive, a little more, much more comprehensive. But between those two, lowering the price of those and making sure every American can get a GLP1 and can get a body scan for free and can get their blood work done for free and can put that into a large language model and get advice. The free market is doing that. Can we accelerate that? Can we say every American gets to have their blood work done every year, a body scan done every year and a GLP one every year paid for by the state if you're under $100,000 in household income, whatever it is. So we start down that path. And then for housing, can we use this technology, 3D printing of homes? It's a little harder to find the AI angle here, but it could also be training construction workers back to education and can we use new materials, et cetera. That's a little bit more of a Hail Mary. If we solve those three problems, the cost of living will go down so dramatically and people will live longer that we can start to say AI will do something for the every man, the every woman in America, every child. So that's my hope, is that somebody takes that position. There's another vector that's coming in here, which is immigration and the anti immigration movement that's going globally. There's multiple reasons for the anti immigration movement. Primarily it's jobs, sometimes it's culture. Both of these are valid arguments to have. And by closing the border and decreasing the number of people in the United States, then businesses will have to pay $30 an hour to have a dishwasher, not 13. They'll have to pay $40 an hour to have somebody work as a nanny, not 20.
- [38:58]
A
Which will robots effectively much cheaper in comparison.
- [39:01]
C
Yeah, well, they'll still be robots. That'll be cheaper, sure. But we can stem the tide of job loss and keep it at 4% and increase salaries, this may become less of an issue. So my prediction is people like J.D. vance, when they run for president, are going to lean into this. Closing the border and deporting people who are here illegally. Increase the salaries of Americans. And those jobs go to Americans. They're not wrong. That is what happens. You can disagree about immigration. So this is all coming together. And at the end of the day, it comes down to hope. And a man without hope is a man without fear. That's the key that people have to remember. And people in this country are starting to feel helpless. And when two or three thousand trucking jobs and cabin jobs go away, when you know, two or three thousand housekeeping jobs go to robots, whatever, the quickest jobs to go away from the entry level one, this is going to become the debate of our lifetime. The debate of our lifetime will be jobs, automation, immigration. It's all around the same thing. The hope that you could own a home, the hope that you could have a family. And we are as the technology industry, inserted right into the machine. Love it.
- [40:15]
A
If you want to know a little bit more about AI and how it will impact house building, we did have Monumental, a twist 500 company on the show. Episode 2217 All About Robot Bricklane, super cool company. Just throwing that out there.
- [40:27]
C
So what's your take on the sort of field I've played out here and independent of political party, what are we going to need to do as leaders in technology, leaders in government to assuage people's fears and deal with the reality of automation?
- [40:48]
A
So I do think that we are going to see AI and automation have major impacts on education and housing and healthcare. I also think that we are going to see more job loss than people like David Sachs are allowing. Today, for example, he said that today we're seeing more job creation from AI than job destruction. Maybe. But I think you and I are on the same page of where we're going down the road. I think the question becomes who is the next champion of the GOP and what's their take on this? And are they tarred with this particular feather if what we just put into place in this executive order becomes as unpopular as you intimate it may be? So for example, when I was watching the coverage of this and I saw actually one half of the all in podcast standing next to the president. Yeah, I don't think that I saw J.D. there. I don't think he was in that lineup. And double check me somebody. But like my thought there was, JD is making A little distance between himself and this particular policy.
- [41:37]
C
So that's the political landscape. But what do you think should happen on a practical basis? Should we do more deportation so Americans get more jobs? Should we tax these robots and use the tax from that? You know, that's one thing people have floated tax robo taxis and use the money from that for education or retraining. I use the word re education on the podcast, which Tucker, we don't mean re education. We mean education opportunities for, you know, trans getting a new skill. What would you do practically, putting aside all the shenanigans of politics and Trump's despot and the liberals are woke, put all that aside if you can. All that cruft, all of the characters, just practically. I laid out practically what I think should happen. Right. And I'm trying to see if I can move the dialogue from everybody's hand wringing about these personalities and labels just to some common sense. So let me give you the one common sense when that people are floating. Should we put a tax on each robot and each robo taxi of, I don't know, 10 or 20%? So you're a robo taxi company. Every time you do a $20 ride, you give four bucks, 20% of the ride to a retraining pool of capital for cab drivers. Every trucker, every automated trucker is a dollar a mile added to it. And the billion miles that get done, a billion dollars gets put into a fund to train people who lose their jobs as truck drivers. What do you think of that as a practical solution?
- [43:13]
A
I think it's too cute by half. So I think that if we start to apply a per robot or per job, taking away cost, we're going to end up just dickering about that forever. So what I would do instead is I would say, look, we know where technology is going to. We know historically, as technology has impacted the labor market, we've always gotten back to full employment. So what I would do is pretty simple. I would tighten up the corporate tax code. I would raise the rates a little bit, and I would just ensure that we have enough money in the central pot using normal taxation methods to help retrain people and support them in the transition point. I don't think we need to get too complicated here. I think if we try to come up with a specific solution, then we're going to see people like, I don't know, Chamath say on the all in podcast, hey, we're taxing innovation. Why are we trying to slow down the improvement of our economy? Well, let's not do that then. Let's just ensure that there's enough money to take care of ourselves as we reinvent ourselves.
- [44:01]
C
Yeah. What do you think of deporting more illegal aliens here in the country in order to raise the job prospects and the salaries of Americans for or against?
- [44:14]
A
I'm going to answer that in a more nuanced fashion. I have worked on farms, I have worked with migrant labor. I have worked night shifts with migrants labor. And let me tell you about people who work on farms. It's not, it's not Americans. So when we talk about specifically illegal immigration, we're not really talking about people coming in taking white collar jobs, for example. We're talking about people working at the absolute bottom of society.
- [44:39]
C
But if we need those jobs, if, let's just make it even easier for you to, to give an answer. If we got to 15 or 20% unemployment in the country, then, and if Americans wanted those jobs and Those jobs were 20 or 30 or 40 bucks an hour, would you be in favor of deporting more illegal aliens working on farms or as dishwashers in order to reclaim those jobs for the 15% unemployment.
- [45:06]
A
Rate if we got to that particular situation? Jason? Sure, But I think if we get to 15% unemployment, there's going to be pitchforks and burning down buildings. So we're other problems.
- [45:15]
C
We have had it happen. We have had it happen and it.
- [45:17]
A
Led to political change. I mean, think about the New Deal, the timing of it and the Great Depression. So, so I, I just, I, I think that there's a middle ground here that no one is seeming to hit in which we don't have to be full bannon opposed to changes in how we execute labor across different strata of the economy. And we don't need to be so willfully blind to it that we just say it's not going to happen. I just think that if we could lower the temperature a little bit here and come to some reasonable middle ground here, because I do think that an economy where we can use AI to reduce human drudgery and free us up to do more intellectual work, that's amazing and a really great future. So I can see rays of sunshine here instead of just two differing kind of doomish environments that people are offering.
- [45:59]
C
It's really simple. You know, I always like to put people back to like metrics and statistics. If we look at the per capita income in the country and we look at the quality of life and we look at the unemployment rate and you just look at those numbers and Then you say, let's have immigration in some way tied to those, and then have really targeted immigration where you say, hey, we don't have enough doctors. Let's this amount in in order to fill the gap. But if we hit 8% unemployment, 12% unemployment. Okay. Then things need to change. And tying those two things together, nobody knows exactly how many jobs will be lost and how many new jobs will be created. And even if you're like, hey, we get through this every time. Okay, great, we do get through it. We have traditionally gotten through it every time. But there could be pain and suffering, so we obviously want to alleviate that. Earth.
- [46:54]
A
Yeah. On the data point though, you make a really solid point. But a couple of things we've talked about on the show in the past, something that you've said a number of times, is we are at the lowest unemployment of our lives. And you and I both agree we're at 3, 4%. And that's fantastic.
- [47:06]
C
4.5. Yeah.
- [47:07]
A
Well, now, but at the time, this was, this was kind of more of a last year conversation. But yeah, good point. At the same time, we're coming off of a simple enormous wave of illegal immigration during the Biden years.
- [47:16]
C
Yeah.
- [47:17]
A
And the unemployment rate was still low, lowest of our time. So that's, that's kind of where I'm sitting. I just, I'm a little worried about immigrants getting dragged into this. And I think we should more focus on just what we're doing at a state and federal level. Can I ask a question though, before we move on, Jason? Sure. So, Steel manning the contra position here. So I wanted to make the argument that the Democrats and part of the GOP are making about the state level AI regulation when we talk about self driving cars as a success now in the United States, spreading across the country, growing well. And I think we're leading the world there. And that is something that does have to deal with a Patrick of state level laws, even local laws, and it's going just fine. So what makes AI sufficiently different from applied AI self driving that it needs a special carte blanche that other technologies have successfully matured and won under?
- [48:10]
C
Yeah. So there would be a moral imperative to save more lives with self driving. And if there were states that were decels and they wanted to stop self driving to protect jobs, but then you had the counter argument of saving lives. This becomes a talking through each other kind of issue. And then on top of it you have, well, who's getting rich from eliminating 5 million drivers or 10 million drivers? So you wind up having like three discussions going on concurrently. We've had this before when we wanted to put airbags in cars. You know, there was like the cost of it. When we wanted to do CAFE standards and get rid of smog, there was a debate about it. And then speed limits. You know, in Texas, you might want to go 75, 85 in Utah, in other coastal cities where it's more crowded, you may want to keep it to 55 or 65 because you want to lower the debts. But if you're driving between Houston and Austin or Dallas and San Antonio, you know, it's pretty okay to go 75 or 85 in a modern flat and.
- [49:21]
A
Straight, like, come on now.
- [49:23]
C
Exactly. So, you know, we've had those discussions before. This one is going to, I predict the driver, the perfect driver concept is the one that will become the first one up the hill. Why? Because it's ready and because there's a lot of jobs and because those are entry level jobs or gig work that are supporting many families. And in China, there's been rumblings and protests and the government has indicated or intimated that they are going to give licenses to robo taxis and they're going to do it in small numbers at first in order to not have very quick job displacement. So I bet you in a year or two, Steve Bannon and you know, this, you know, Bulldog, he's got there going out, you know, talking about this cult of AI and AGI happening. You know, this is an AGI, it's verticalized AI to your point. And it's going to eliminate all driving jobs. Who wants to drive? It's tedious. I mean, maybe around a track and going fast, but most people don't want to drive, most young people don't want to drive, and most people want their Uber to go from, you know, $3amile down to a dollar. Right. It would be going 10 miles for $30 to the airport versus $10. Everybody can afford 10, you know, half the people can afford 40, you know.
- [50:53]
A
Yeah, the thing that I, the reason why I asked this is I don't actually have a well defined view on state level AI regulations versus one federal standard because I'm not quite sure about the Commerce Clause versus the 10th Amendment. And we're going to sort it out in courts, but I can really see both sides of this and I'm a little bit uncertain. So I'm just trying to think it through as we go through this because not everything is super, super cut and dry. But yeah, it's going to be A fun, fun, fun. Next couple of years politically and technology is going to be all over it. Because, Jason, I know you love to hear from people who share your first name. And. And so why not turn to our dear friend Jason Lemkin from Saster, asking about what you should do when a large company reaches out to your startup. He writes down on Twitter, one of the weirdest calls you get as a founder is when a big co out of the blue wants to talk about acquiring you. Literally out of nowhere. He points out that sometimes these are real, sometimes these are not. But he thinks that you should take the meeting at least to listen, even though, yes, sometimes this will be a phishing expedition. Expedition. But we're talking a lot on the show about how there's going to be more M and A next year. So I presume more founders are going to get this call. So I want to know the Jason Calacanis advice to founders for dealing with interest from larger companies.
- [52:09]
C
Yeah, so there are departments, corporate development departments, and they have been asked to stand down under the Lina Kahn, you know, regime and just not even pursue M and A, because what's the point? Now you see Netflix wanting to acquire Warner Brothers, and you see DoorDash acquiring three or four international companies. Of course, you don't just have to get through the M and A filter here in the United States, as we saw with Adobe trying to purchase Figma, it was, I believe, the UK that put up the biggest fight. So it's on like Donkey Kong. We're going to see massive M and A, and as I said, I think we're going to see $100 billion company get bought in 2026, and I think we'll see multiple $10 billion or billion dollar companies get bought as well. So it's coming, folks, and that's a good thing for the playing field. Consolidation is good. It creates disruption, it creates opportunities. The only time it's really bad is if it limits consumer choice or, you know, it creates a monopoly position where that person's able to raise prices. That's really the only two things you have to think about. Netflix buying hbo, you know, yeah, that has a detrimental impact on directors. And the catering companies and the lighting people. Sure, you know, they have less parties to negotiate and less customers, fragmented markets better for them. But the truth is these are just magazine companies buying each other, cable companies buying each other's equivalent of magazine companies buying each other, studios buying each other as the equivalent of newspapers buying each other. In other words, they're all going to lose to TikTok and YouTube or are losing to TikTok and YouTube and it's.
- [53:54]
A
Or have lost to TikTok and YouTube.
- [53:57]
C
Yeah, to a certain extent, you know, they have lost a lot of audience to those platforms, but they've still built a great business with hundred, you know, hundred this, 300 million plus subscriber bases for streamers. Put that aside. People are going to knock on your door. Yes. And typically the way this goes is partnership to purchase, partnership to purchase. You do some sort of partnership. That's what happened to me with Weblogs Inc. And aol. They came to me and said, hey, we want to invest, we want to learn from you. I took the meeting and then it went from them putting an investment of a million dollars into buying 49% of the company, buying 51% of the company to just making a $30 million offer. We'll buy the whole kit and caboodle. And so that was great for me. We decided to do it. And that all happened over like a six week period that they went from courting us to buying us. And last minute News Corp came in, was like, hey, can we put an offer in? I was like, yeah, you know, we're, we're already down the road and they wanted to put in a low ball offer anyway. You, you're not obligated to tell the corporate dev people all the top stuff. You just say, hey, yeah, tell me what you're thinking. And they say, oh yeah, tell us about your business. And yeah, you can actually talk your business down. You could tell them the opposite of reality. You could tell them, yeah, we're figuring, what I would say is things like, we're figuring it out. And they said, well, what's your average cost per content? And say, you know, you know, we, we pay the bloggers well and you know, tens of dollars. But I didn't give them the data and give them the list of bloggers. They could get that off the website anyway.
- [55:24]
A
Take the meeting. Don't be overly specific. That makes good sense to me.
- [55:28]
C
Just ask them questions, let them talk.
- [55:30]
A
Ah, okay. So make them pitch you. Essentially, yeah.
- [55:33]
C
Why are you interested? So that's what I said to them. And they said, well, our content cost is $300 per content object on AOL, 2 or $300. And we said, oh yeah, that's pretty high. And they said, yeah, you know, we have full time people. They get, you know, a story done for AOL Finance every three or four days. And so, you know, the $60,000, $70,000 writer is costing us, you know, $500 per piece. And we're like, okay. And they were like, what do you pay bloggers? And you know, we're like, well, they get paid the equivalent of like 15 bucks an hour, which 20 years ago for a writer was pretty good, to be honest.
- [56:05]
A
Like, salaries haven't moved since then, Jason, for a lot of. No, no, they have.
- [56:08]
C
You know, I think maybe they would get. What would a blog post, a 400 word blog post might get you 50 bucks today or 100 bucks today in the age of AI. So, you know, I'm not saying for, for your level of writer, I'm just saying for like the average writer. And so you take the call if you want to sell your company. If you don't want to sell it, you just say, yeah, you know, we're not interested in selling, but I appreciate it. Or you can take a, a brief call and you just ask them questions and you collect information. You get to ask them, who are you talking to? Who else are you talking to? What are your goals? Are you releasing any new products that would be competitive with ours? What's your strategy here? And you just ask them those questions and you know, they're, they're in all likelihood making a decision to buy one of your competitors or you. So they're going to meet with all 10 competitors and then figure out which one they want to buy, or they're going to meet with all 10 competitors and then decide to build anyway. And the other thing you can do is look at their track record. If it's Google or Meta, you know, they do build their own features into their products and they both also have a track record of buy. A track record of buying. If you look at Apple, they don't buy stuff. They do. They buy enabling technologies, but they're not in the business of buying YouTube, WhatsApp, they never, they, they had the opportunity to buy those things. I'm sure Instagram, they didn't. So know your buyer. I knew AOL had bought tons of companies. They had a history of buying two dozen a year. They were constantly buying companies. They were an acquisitive company. Zuckerberg acquisitive and build. So if he wasn't able to buy Snapchat, he just stole the feature. So know your buyer and it's totally fine to take the initial meeting and pump them for information and to, you know, treat them like mushrooms, which is keep them in the dark and feed them shit. That's what most people would say is the best thing to do. It's been another amazing episode of Twist and. Have a great weekend, everybody. We'll see you on Monday. Bye. Bye.