Behind the scenes at the ALL-IN AI SUMMIT w/ POTUS, JD Vance, Jensen Huang, Lisa Su, and more… | E2156

Here's the thing about H1B visas versus American college graduates. H1B visa are going to get paid 30, 40% less. And because you have, they essentially become indentured servants. If they don't find a new job within 30 days of being laid off, they're kicked out of the country with their family. It's incredibly cruel and unnecessary. They should have at least six months to find a new job. A minimum. Minimum. Like it's ridiculous. So the people I know who are abusing the H1B visa system and other systems like this, I have heard people explicitly in my career say we're going to get these people because we can work them harder and they don't have any recourse, you know, to call them indentured servitude or slaves may be a little bit hyper. It's obviously hyperbolic. But they are less empowered workers and I would say significantly cheaper and less empowered. So JD makes a good point.

This Week in Startups is brought to you by Inbound. Connect with visionary leaders like Dario Amadei and Amy Poehler at Inbound 2025 September 3rd through the 5th in San Francisco, the epicenter of tech innovation, and transform your business Strategy for the AI era. Use code TWIST10 for 10% off your general admission ticket@inbound.com Register valid through July 31. Bolt don't be left behind. Build apps quickly without knowing how to code with Bolt New. Try it free at Bolt New Twist and Public. You take investing seriously. Public does too. Build a multi asset portfolio and earn an industry leading 4.1% APY on your cash with no fees or minimums. Learn more@public.com twist all right everybody, welcome.

Back to this Week in Startups. Alex Wilhelm, Jason Calacanis here on the Ones and the Twos. I just got back from D.C. after being in New York. We're supposed to have a little vacation. See my family and then my friends decided 10 days ago during the middle of my New York vacation to see the family that the president needed to do a summit with all in. And we had to actually make some big announcements. So we'll start with that, I think talk about those executives orders. I'll give you some behind the scenes of course and then there's a ton to talk about and we got a lot of great founder questions and stories to process because hey, I haven't been here for two days or for two shows. You did six excellent interviews which was the highlight of the interviews, I think Xtropic.

Now Jason, you're familiar with The Twitter account Based Jeff. Based Jeff Bezos, whatever. That guy, he's the founder. So we had him on and we got to talk about thermodynamic computing, future of AI chips and power. It was just one of those interviews in which I didn't know much about the deep science, but I got to learn a lot. So it was just an absolute treat.

We're trying to make these 20 minute interviews not like to take up a ton of your time, but you need to know about these companies and get the update. They're also very busy. So three 20 minute interviews and we want to get into the business models and looking at these through the business lens. You know, companies can spin what they're doing, their mission, all that. I want brass tacks included in these. They can give the spiel about their mission and. All right, let's get to work here. I think. I don't know if you got to catch all the announcements, but it was a pretty exciting day. Many of you may have seen the President gave a talk. He went massively off script. It was a very interesting, surreal moment to be in the first row. First time I've met the President, took a picture with him, got to see JD Vance on stage. We did a little round two on our immigration debate. He loves to debate, so that was awesome. We went a little off script. You know, going into the immigration stuff was tangentially involved with AI, but he wanted to have a freewheeling. He decided at the last minute he just is a fan of all in, so he wanted to come and just, you know, chew the fat with the team. And that was a lot of fun. Had a little debate about energy, which got a little spicy, which was great. Spicy moments can kind of give you the roadmap to what's important for folks who are wondering, like, all in is not endorsing Trump. All in is a podcast with four members. Each person has a different opinion. I would say three of the four people are very much in the Trump camp and super big supporters. And then you have one moderate moderator, me, who, you know, as I've said many times, I think whoever wins, I want to root for them. And I will give my exact percentage of agreement with this administration every week if you all want it. Right Now, I'm about 80% in agreement with them, which until the Democrats or an independent puts up a better product that I can agree with half of what they're saying, hey, I got no choice but to support this administration and what they're doing with AI and crypto is a lot more than the last administration, which wouldn't even meet with us. So important note for the Democrats, you might want to meet with the business leaders in the world and at least give them a seat at the table. And this was an unprecedented seat at the table. I will say that Lisa from AMD showed up and we also had Jensen from Nvidia. I mean, this was. Big guns came out and it was quite surreal. Where do you want to start? I'm sure you have questions for me.

All right, so we are going to watch a video here. This is with the Energy Secretary Wright with Jason debating solar power intermittency and kind of the future of the grid. Let's take a listen.

And just a crazy love affair with intermittent unreliable energy sources. I'm talking about solar. Why are you so down on solar? This is the cheapest thing you can install. Batteries are here and they're being produced at an incredible rate. Why are we so anti solar or why are you so anti solar? I'm not anti solar. Why do you keep saying then that like this unreliable solar, if you put batteries on it, it's totally reliable. If you take all the batteries in the United States, you can store five minutes of power, five minutes of the entire country. But if we've had many days in California and in Texas where solar has been the majority of it. So why are you so down on solar? It can be the majority on a sunny day in the summertime. That's not what matters. In pjm, where we are right now at peak demand this year, 97% of electricity, wind, solar and batteries delivered 3%. That's when it matters. If you're 20% on average, cherry picking DC.

All right, so this is so interesting to me because it feels like you're making a very obvious point, which is during the daytime there's sun. We can collect that power, we can put it into a thing called a battery, and then we can use it later on.

Yep.

This is basic stuff. And the response from this guy is, well, currently it doesn't power dc. Therefore, that's just that to me is not.

I think in fairness to him, he was probably expecting no pushback and that maybe he would just be able to give, you know, a standard spiel. But, you know, I'm paying attention and I wasn't trying to be super adversarial here, but he wasn't expecting the pushback. You know, I'm not saying my job on all in is to give pushback, but the spirit of the podcast is free dialogue Unconstrained, we're going to mix it up. And so whenever I hear somebody like a little too much spin and I felt like if you're going to give solar this big dig without actually saying why, I need to know why. The audience is wondering why. So we had to get into it. He says later. He circled back around later as you may have seen and said, I just want to let everybody know I'm not anti solar. These guys are just, I think, very pro fossil fuels. And so we probably had too much emphasis and too much of the government for many people's taste. Spending money massively on solar and wind and not doing enough drilling. And now we're kind of got the other group in office, they're going to be like trying to get more, more gas out of the ground, et cetera. And it's all of the above. Let's call it what it is. We need all of it. If you're going to look from this, from first principles, solar is the best solution for the short and the midterm. Nuclear and eventually fusion is the like the 10 year, 5, 10 years from now we might have more nuclear online. For now you do need NAT gas for sure and you do need a lot more solar. And solar, you can just lay solar panels on the ground and when you need more power, you lay more on the ground and you don't need to put them on roofs. As long as you're not in a populous city where there's no space. It is the best solution. It's obviously the best solution. If you look at the cost to install coal even, and clean coal is I think a significant premium on top of coal or to do NAT gas. Those new plants are much more expensive than solar and worse for the environment. Yeah, and if you bring up environment, I think you have to bring up two different pieces because it is a trigger. Some people don't believe in global warming. So if we just take the global warming debate out of it, let's say for argument's sake that there was this grand conspiracy and global warming is a non issue and the whole thing was made up. Okay. Now I certainly think it's been exaggerated. I don't think it's made up, you know, like you can see the global warming, but we can have that debate. But if you put that debate on the side. Nobody wants to live next to a coal plant. That's not good for your lungs or your kids. Nobody wants to rip the environment apart. Of course not. But we do want to compete with China and other places inbound 2025 is almost here. This is not just your standard user conference. No, you're going to get opportunities to check out the latest HubSpot updates and demos, of course, but. But you're also going to hear inside level insights from visionary leaders and personalities. Dario Amodei will be there, the CEO of Anthropic. They may Claude and I am obsessed with that product. And this is how you're going to level up as a startup founder. You're going to level up by going to inbound 2025. Work from home is over, folks. It's time to get back in person. You need to meet new people, interesting people, kick around these big ideas, do a jam session. So if you want to hear from these experts like Dario, like Victor Ripparbelli of Theseia, HubSpot founder Dharmesh Shah, my favorite guy, and many others, we're going to see you there. Inbound 2025. It's all happening September 3rd to 5th in my old stomping ground where I am right now, San Francisco. Use the promo code twist10 at checkout for 10% off your GA ticket@inbound.com register. Inbound.com register. And so pollution is the better way to say it. So if you just say pollution, forget about the word environment. It's triggering for people on a pollution basis, nobody, 100% of people don't want to pollute. So that's, I think, the way to frame this. There's pollution and there's global warming. I just say take global warming off the table. We're going to solve it with technology and we're well on our way. And the people who have to solve it, frankly, is not the United States because we have great standards. Whether it's Democrats, Republican standards or somewhere in between the two, it's India and China. That's the. No matter what we do in the United States, it'll have zero impact on this long term if China and India don't get their stuff together and they're going big time solar in China and big time nuclear. So they have already got the memo. So this was, I think, just a little bit of a political statement from him. I think this administration is anti renewables, generally speaking. They don't want to give them subsidies. Okay, fine, don't give them subsidies. Solar still wins. Now wind, you do have a pretty good case for. Wind is not actually very efficient. It's very expensive. You have to maintain these things. It does kill birds. It is an eyesore. Nobody wants to live next to a wind Farm. I get it. Yeah, you do?

I do, yeah. Hell yeah. I think wind turbines are gorgeous.

Oh really?

They're fantastic. Yeah.

Esthetically if I'm driving over like the rolling hills of Northern California, I don't want to see a bunch of windmills. I don't mind it in the desert, I suppose. Yeah, I don't think they're good looking. But just cost wise, everything comes down to dollar and cents. Solar's already won. It's a meaningless debate. But I'm glad we had it because it kept people very entertained during a very boring panel. Let's be honest, energy can be very boring. A lot of people came up to me and said hey, thanks, thanks for making that interesting and speaking up about that. And you know what, it builds their credibility. So whenever the administration faces some good fastballs, which you know, I think I'm uniquely qualified to do, my background in journalism and also I'm not part of either party. I'm a moderate, I'm an independent, I'm not a Democrat, I'm not a Republican, I am a full on independent moderate. So.

So a couple notes about this one. If you go back to May 13th, 2025 to twist episode 2125, we had the CEO of Terrabase on. Terrabase is a company that's helping people do mass solar installs and if memory is served, Jason, they built a on site robotic factory to, to actually finish the construction of these panels so you can do even faster, even cheaper updates to them. So that's one thing. So cool companies are out there, are working on this now and just on the count of, you know, people that want to drill more, the market is talking right now. The market is saying that what we are not going to do is keep drilling at the pace we have been. This is a chart from Y charts and it shows just the number of oil rigs that are currently active and as you can see they peaked in 23. And, and then this is not a regulatory point because this is 2025 starting here and it's been dropping pretty quickly. So I wonder what the driver of that is.

I guess it's just the economics. If you have, if oil goes down below $60 a barrel, it's not worth activating certain mines or certain plots because, or rigs because it's just more expensive to clean it, et cetera. That's where like the oil sands come in in Canada. Canada just too expensive.

Oil sands is very expensive and the direct opposite of that is what Saudi Arabia has believe is light sweet crude that essentially, if you stuck a straw on the ground, it just came shooting out. So it wasn't hard to produce, as they say. Anyways, all this to say that I think you're dead right about solar. And I thought that was a funny clip. And I was like, why are we arguing with the past? This is a thing about the future.

Oh, you know, it's. We live in contentious times and I look at how it's all framed and what is the truth? You know, I think as a person listening to this podcast, are all in. Just go to ChatGPT, go to Grok, go to Gemini and say, what is the cost of a solar installation? And what you'll see is solar installs now are so much cheaper than coal to install a new plant that even the retrofitting of an existing coal plant is more expensive. And you can ask producer Claude and he'll tell you, and she'll tell you. They'll tell you. They'll tell you. Producer Claude, they will tell you. They'll tell you.

The truth is, assign a gender. They're going to call us and be like, what's that?

No, it's them, it's they. Producer quad is a they. And they will tell you very clearly that, you know, unless you're right on top of the gasoline, you know, right on top of the nat gas, much better to just install solar. There is, you know, a challenge with who makes the solar panels, and that's China. So there is a little geopolitics sprinkled in here. I think sometimes you got to tease this out. There could be many things at work. There could be lobbyists who want more natural gas. There could be people who want to stop the subsidies. Okay? They feel like the left went too far, but end of the day, we have a ton of oil, but we need more demand, and we're going to need all sources. The biggest win. And I'll say, listen, you could give Trump the Nobel Peace Prize or whatever prize you want to give him. I told some of the folks who I met in the agencies when I was there and some people working on nuclear. I said, if you guys stand up a nuclear reactor before he's out of office or within a couple of years, or, you know, he wants to do dozens. I was talking to Lutnick who spoke at the event, and I was talking to him offline, he said, they're. They're going all in on nuclear. They want to get dozens. So if they do that, I got to tell you, that makes them heroes to me. Because if we can get more nuclear on. Oh, my Lord. That's going to be amazing for the abundance of this country in the AI race. Okay, moving on hard.

Agree with that. You did mention two CEOs that drop by, Jason, both Jensen and Lisa. Do you want to take a look at those clips real quick?

Sure, yeah. Cue us up, whatever you got.

All right, so we are going to hear from Lisa Su. She is the leader over at amd, and we're going to talk about the potential for AGI. Jason's clip is about one minute, and let's give it a listen.

The best case scenario for this technology and this trajectory we're on, which is accelerating and you're enabling, what could the world look like in 10 years? Let's say pretty obvious we're hitting artificial general intelligence at this moment. I think we'd all agree we're starting to see that, but superintelligence can't be far behind that. I assume you agree with that. Soon we hit that super intelligence. What would the world look like in 10 years in the most optimistic scenario, if we do this right? Well, I think the exciting part about it, and I can say this very sincerely, this is the most transformational technology in our lifetimes. I mean, that's the way we should think about orders of magnitude. Orders of magnitude. The reason is it's not just going after one aspect. You can actually take A.I. and make science better. You can take A.I. and make medicine better. You can take AI and make manufacturing better. You can take AI and Make every aspect of your business better. And so, in my mind, 10 years from now, we'd like to believe that we are really leveraging it to solve some of the world's most important problems. I like to say Amders get up in the morning and they say, how can I use technology to solve some of the most important challenges in the world? And AI is really our mechanism for doing that. I have a business strategy question. If we went back 20 years and we were okay.

So I think that's very optimistic because she's drawing a very wide picture, Jason, of where AI could impact.

Yeah, this is the natural tension of any discussion with AI. We're in the industry. I think we have a bias in the industry towards having seen almost every technology make the world better on the margins. How people use technology is up to them. We are now at the lowest unemployment of our lifetime, while people have become more efficient. In my 50 years on the planet, 54, unemployment's at the lowest. And I remember every discussion about, oh, my God, Job displacement. Job displacement. Microsoft Office, the Internet, bike messengers were going away. FedEx was going away. The mailroom was going away. Operators on the phone. Every career you could imagine was going away. Yet here we are with the most unbelievable robust employment of our lifetimes. I know that's unpopular for people to recognize that. So then we're left with the discussion. Is there going to be job displacement? That came up a bunch of times. I do think everybody agrees there'll be displacement and I don't think anybody knows if it's going to be conquerable or not. The vibe coding revolution is here. The days of needing a team of developers to build your mobile app are over. With Bolt New anyone can build web applications quickly and without knowing how to code. Getting started is as easy as visiting Bolt New and then telling their AI powered system what you want to make in plain simple English. And this is not some template or WordPress blog. No. You can use Bolt New to design a stunning new landing page for your company. You could upgrade your CRM, build a consumer facing app and more. This is the future folks. Your competitors are already using cutting edge tools to ship their products faster than ever. Don't be left behind. Try Bolt New for free at Bolt New Twist. That's right, Bolt New Twist.

Yeah. I want to just underscore your point Jason, because I think this is actually a useful thing for people to see kind of in chart form. So if you're on the audio version, what I have here is a historical unemployment chart from the United States tracking the unemployment rate over time. And if you're looking at the video version, the shaded areas are recessions. And so if you take a look at this Jason, here is the the 90s and the first Internet boom in personal computing and unemployment went down. And then we here have the post grade financial crisis. Great financial crisis following this unemployment spike was the pandemic to your point. And now we're back down to 3.6, 3.5. But it doesn't matter what era of the country's economic change you look at, we get back down to 3,4% unemployment and the only thing that really spikes it is a recession. So I think you're correct in saying that yeah, all this time we have seen consistently a return to full employment be a change in historical patterns to actually end up with mass unlimited from AI. So I'm long term bullish, short term scared.

Yeah, I mean if you look at self driving and you look at humanoid robots in factories, I gave a statistic and people ask me where'd you get that statistic from? And I said, well, I did it in my head. For every robo taxi on the road, which will work 20, 22 hours a day, that's basically two shifts, isn't it? Two eight hour shifts, maybe two and a half shifts. In New York City, the taxi drivers rent taxis for 12 hours. That's why they're so exhausted and on the margin. Sometimes they're a little cantankerous. You can give them some grace and empathy. A 12 hour shift in a New York City taxi in traffic, I mean, come on, give these people a break. It's a hard, hard job. But if we think of like a basic 10 hour shift, there's two shifts, there's, that means there's seven shifts a week, that's 14 shifts. It's basically three jobs, three full time jobs, right? 15 shifts, five shifts, eight, 10 hours each. Okay. Each robotaxi takes three full time driver's jobs. Now most people drive halftime for Uber or Lyft, so it's taking six gig economy jobs, three full time equivalents. Now you go to a factory, you can't work somebody 12 hours in a factory. It's just too much. It's eight hours. Maybe they get seven hours out of them with breaks or whatever. So now you're talking about three shifts of hard labor times seven is 21. It's really four jobs. So humanoid robots will take four jobs when they hit scale and they're like in an Amazon facility sorting packages or delivering them self driving car, let's put it at three. So maybe it's like it's going to happen fast, folks. I think it'll be slowly than fast. And if that is the case, you're going to see millions of factory workers, drivers and maybe some painful chore jobs come on the market. Those people will be able to be deployed into other jobs very easily, I believe. Construction, we need more homes. Education, we need more teachers, Healthcare, we need more nurses, doctors, orderlies, EMTs. Just pick. Anybody at a hospital anywhere had this epiphany, Alex, during the programming. Anywhere you have a wait time in weeks, days to weeks is an incredible opportunity. So just think that through. When's the last time you had to wait two weeks, three weeks for an appointment? When's the last time it took two or three days to get through to somebody? Healthcare, education, construction, renovating your home, you know, pick one of those categories that means there's demand. You try to get a plumber to redo your hot water heater, you know, they might say Check with me in two weeks. I might be able to fit you in in August, end of August. Right. That's where all those jobs are going to emerge. And those, by the way, super high paying jobs, super well paying jobs. So how quickly can somebody learn to put up drywall if they were working in a sorting facility, if they're driving an Uber or a taxi, how long before they can work as a nurse or an orderly in a hospital? I don't think these are terribly different jobs. I think they're well within people's capabilities. And then it just becomes reeducation and training. At the AI Summit, we also had, I think three things got passed. Maybe we can just get into. Before we wrap up here, we can get into the details of what the administration proposing. Some of it I agree with, some of it I'm unsure of, and some of it I definitely don't agree with. So I can get into all of those points. And by the way, I had no idea what the platform would be. I mean, I know David's positions. David Sachs are AI crypto czar. I know some of his positions on this because we've debated them before, like training data, states rights, training data, woke AI. And yeah, so maybe we'll get into an executive order, by the way, you know, presidents give these things all the time. Then they have to somehow go become law. And so I was talking to somebody who's on the inside of all this and they're like, by the way, it's like that's the starter's pistol. That's like somebody doing a spec and saying like, okay, now you guys have to do a ton of work. So some of these executive orders never materialize into law or into a policy. This one is going to, this is I think, signature legislation. So let's get into it.

All right, so there were three executive orders that came out, Jason, in the order from our notes. The first one is preventing woke AI in the federal government. I gave this a read and it essentially sets rules for AI models procured by the federal government to not include or not focus on certain things. Just keep in mind, this is not saying that you can't make an AI model to your own spec. It's just that if you're selling to the federal government, these are the policies that they are going to put into place. This is the one that I was most curious about, how they were going to phrase it. And just to cite the actual text a little bit here, one of the most pervasive and destructive of these ideologies is so called diversity, equity, inclusion, DEI in the AI context, DEI includes the suppression or distortion of factual information about race or sex, manipulation of racial or sexual representation in model outputs. And this is the part that really threw me. Incorporation of concepts like critical race theory, transgenderism, unconscious bias, intersectionality and systemic racism. And to me, incorporation is a weird word to turn on. So I'm curious if you can elucidate what they're trying to say here.

I don't know. I didn't talk to them about this. I think this is the least important part of all this. I think this is just red meat for the base. Let's call it what it is. When politicians get elected, they have a base, right? And the base helps them get the nomination. We all know the horseshoe theory. Two extremes woke lunatics on one side, MAGA lunatics on the other, socialist communists on one side. We're going to have breadlines and we're going to have whatever mom Donnie is talking about with supermarkets. That's an example of feeding the base. When he goes into office. You always move to the middle. And the middle is maybe we can create more housing supply, maybe we can make it more affordable to take the subway. There are things you can do that are easier to execute. I see whenever they do the ice raids in Los Angeles, that's performative. It's not productive. This is another example of I think performative as opposed to productive. I don't think any of the folks who are in, who are building these models are saying, how do I sneak in? Gender identity politics. And by the way, you can prompt them. So I could put a prompt into any of these large language models, say my politics are X, Y and Z. You can just put all that you're crazy ideology, whatever you like. I think that Mexican food is better than Japanese food. It's not. It's not true.

No, it absolutely no.

See, and here we go. So that's your.

I care.

You could put in your model that I think the seven, the same five ingredients mixed 20 different ways is an incredible cuisine. And I can say, you know what I want to. I want incredibly refined uni and viva la difference. So the a person could tell their AI to do this. You could tell it to be woke, you could tell it to be anti woke. So it's kind of all meaningless. Let's put that one on the side. I think it's performative. It's for the MAGA base. It's like the ice raids at Home Depot. It's performative and it's not like the real thing, which is closing the border and not letting criminals into the country. Everybody agrees on that. What else is in there that we should discuss Every time you turn on the news, it's about AI getting smarter and smarter. These models are writing perfect code, they're acing the world's most difficult exams, and it turns out they're also pretty good at making money. That's why there's Public.com, the investing platform for people who take their money seriously. What really sets Public apart are their AI powered tools and features. Check this out. You can ask Public a question about any stock in plain language and get a straightforward answer. If you want to see why a particular stock's price jumped or crashed, they've got got a clickable AI generated summary right there in the performance chart. You can build multi asset portfolios on public stocks, bonds, options obviously. But hey, maybe you want a little crypto in there. Some ETFs. I will advise you to try public.com and access industry leading yields like 4.1% APY on your cash with no fees or minimums. Plus for a limited time earn 1% match on all IRA deposits, IRA transfers and 401k rollovers. Never too soon to get that retirement dialed in. So go to public.com twisttoday to fund your new account in five minutes or less. That's public.com twist disclaimer paid for by Public Investing. Full disclosures in Podcast Description yeah, so.

The next one is, quote, accelerating federal permitting of data center infrastructure. This is kind of meat and potatoes. Three things that I think really matter, Jason. There's a. There's work inside of this to ease regulations. The quote here is that the EPA shall, quote, assist in expediting permitting on federal and non federal lands by developing and modifying regulations promulgated under the Clean Air act, the Comprehensive Environmental Response, Compensation and Liability act and the Toxic Substances Control Act. Then they also want to make federal land available for construction, essentially opening up public land for data center buildouts. And then there's also this surprised me, the potential for subsidy from the public purse. So the text reads that the Secretary of Commerce who you were speaking to shall quote, provide financial support for qualifying projects which could include loans and loan guarantees, grants, tax incentives and off take agreements. So essentially we're going to cut the red tape. We're going to let you use federal land and we may even subsidize it.

Great, Let me think that through. Is there any if we believe, if you believe AI is a race and America needs To win it, then using some federal land and subsidizing it is what China's already doing. So this is all us versus China. Let's be honest. China is pulling out all the stops. If you want to build next to this hydro dam, they will block the equivalent of what's the most important river.

In the United States, the Colorado.

They would literally, in China, block the Colorado river and be like, sorry, we need to win AI and put all that power into a giant data center that ran across the most important wetlands for any creature. So we're not going to do that here, but easing up the regulations, probably a good idea. Devil's in the details, and I think probably a decent amount of what's hidden in there that's important is AI. Data centers in the short term are going to use natural gas. There's no way around it. There's just not enough time to get solar, batteries, nuclear and other sources on. And they have to be off grid. You're going to have to bring your own power. You're going to have to move your data center to where the power is. And so that seems to me to be wise. And I do think we have to win this race. I do think there'll be many winners. And so. And I don't think there should be any, by the way. I'll just say one final thing on that. I don't think you should give grants to anybody. Everything should be in the form of a loan. It should be a senior debt device. The taxpayers get paid back before anybody else, plus they get warrants. So whatever amount of money you take from the government, 10% of that amount should be given to you in warrants. So if we have a billion dollars to OpenAI or Google to build a data center, they should pay market rate or some percentage of market rate for the loan. It could be half market rate. It could be 1%. Who cares? The interest isn't the thing they should give. If it was a billion dollars, they should give 20%, $200 million in options. How's this for a framework then? If OpenAI goes from $350 billion valuation to $3 trillion, which it probably will or could, then we would 10x that 200. We would get 2 billion back plus the original billion, plus some interest. That means it would be profitable for the American taxpayer.

I just, I get a little heebie jeebie about using public funds to support these companies because I feel like putting aside all the worries about government intervention in the market and all that, I just don't Want to cement the current winners as the governmentally selected long term winners. Because if there's a company that's going to come next year and build something that blows OpenAI out of the water, I don't want OpenAI to be able to lean on the fact that the government gave them cheap land, regulatory credits and whatever to build other data centers and then use that to quash other sure.

Competitors. So that's where the details come in. It's a great punch up. Again, executive orders are the starting pistol, not even the rfp. So here we go. I think we're in alignment. Let's keep going.

All right, the next one is the promoting the export of the American AI technology stack. What this does is quote, build an American AI export sports program. And this is a package of things like AI hardware, AI models that you can sell as a unit to countries around the world that we are friendly with. There is a little bit of public money here as well. The EO says that the Economic Diplomacy Action Group didn't know we had one. Quote, shall coordinate mobilization of federal financing tools in support of priority AI export packages. So essentially, if I don't know Ukraine wants to buy the American AI stack, we might subsidize that by offering a loan in some capacity that's more of an import export action. So I'm a little bit less worried about use of public funds there. But to me this makes pretty good sense. Jason, let's create a unit we can sell out to other countries because not everyone's going to have the Kingdom's budget.

Yeah, this all comes back to Huawei. If you look at Huawei, which is a company that builds like 5G infrastructure, infrastructure for telecom. They are a Chinese brand, they went around the world and they were really cheap and countries, you know, China, uae, Africa, Latin America and even in Europe, a lot of countries use Huawei's 5G. This is a significant issue because the Chinese are don't have everybody's best interests at heart. And you can be sure every piece of Huawei equipment has a backdoor for the Chinese government. So if you are Germany, Italy, Spain and you're using this stuff for your infrastructure, well, you are basically saying please spy on us. And if you don't, then you're naive. So I think America, this is the equivalent that could happen with AI. We don't want to see all of Europe using Huawei chips, Huawei AI models, because then all of that power would accrue to the Chinese Communist Party, not United States. So that's really what we're talking about here, and it starts in China. If China has an incentive to make chips that compete with Nvidia and amd and we don't let them on those and we don't let them have access to this technology, they will. So we want our platforms to win, but we don't want to give them the latest and greatest. And as Howard Lutnick said, part of this negotiation that they're doing is like, oh, well, give us your top hypersonic missiles. China. China's like, are you crazy? Like, we're not giving you our best hypersonics. Oh, just sell us like 10 of them so we can deconstruct them, you know?

Yeah. And every time China wants to open a business in the US they should have to form a JV entity and only own up to 49% of it and commit to technology transfer. Oh, they don't want to do that.

Exactly. So, I mean, just reciprocity is what we're looking for here. It's a no brainer. I've had this conversation 100 times here and on all in the tariff concepts that the Trump put out there to begin with were shock and awe. And right now, the ones that are being executed on by Lutnick and Besant and the whole crew over there are boring and mundane and effective and in America's best interest.

So you're a tariff guy now?

No, I'm a reciprocity. I've always been a reciprocity guy. I'm not like, we should do, oh.

Are you referring to China in particular or the overall package of tariffs around the world?

There are 10 countries that we trade with. They equal 99% of the issue. It's a major power law here. It's a big Pareto principle. We just have to work it out with the top trade imbalance folks. And even Trump, during his speech, he conceded that he's like, I don't know if we need to do many more of these. I don't know if the 108th country, which I don't know the name of and I've never heard of, we need to have a trade agreement with. So again, how Trump does business? Just boom, explosive, crazy, everybody covers it and then whatever, 72 hours later, it doesn't seem as crazy. And then six months later, the people who are executing it are going to execute it in a very mundane, realistic fashion, which is reciprocity. That's where we found up. So again, you may not like the style, but at the end of the day, reciprocity no, no, I don't agree with that. You don't agree with reciprocity?

No, I'm a Chicago School of Economics guy. I'm a free trader.

Okay. I mean, if you. So you think they should be able to put 100% tax on cars going into Japan or China and Korea, and then we allow their cars here with no strings attached.

If they want to have a less effective trade policy than free trade, sure, they're willing to. They're able to hurt themselves. But if you want to think about why you might want to agree with me here is that melee down in Argentina made very compelling arguments before he was elected president discussing why protective tariffs for domestic industries lead to more poverty in your nation. So, no, I'm going to stick to the current Silicon Valley favorite outside of the United States, Melaye, and state of my free trade principles.

Okay. Yeah, I just. If they're going to put this huge tax on cars coming in to China, Japan and Korea, I think they should face whatever they give us. Just we'll do the same thing, you know, and that's why if you look at Japan or you've been to Korea or even China, like, there's nobody buying American cars, except once in a while, you'll see it. I saw like three Teslas in Japan. You know, this is years ago. And it was just rich people who paid. I don't know what is. What was the current tariff on American cars in Japan? 50%.

It's very high. Yes.

I think it might be 50. So just how do you ever compete? Because you've got to ship the car there, too. There's, like, that cost and there's like the, you know, maintenance of the car. So who buys. When I was in Singapore, like, who buys American cars there? Somebody was showing me their Jeep and they're. And I was like, oh, you have a Jeep? And I said, yeah, you know, it cost me like $150,000 to have this tricked out Jeep wrangler. I'm like, a Jeep wrangler is like a $50,000 car in the United States tricked out. 70,000. So there you have it. Okay. Anything else in there I think I can remember is the content and the training one.

Yeah, yeah, that's exactly where I was going. This was not in one of the EOs, but it was in Trump's speech. And we don't spend a lot of time on this. Jason, we've talked about it a lot on the show, but a couple of quotes from the Trump riff on this quote. You can't be expected to have a successful AI program when every single article, book, or anything else that you've read or studied, you're supposed to skipping ahead a little bit. When a person reads a book or an article, you've gained knowledge. That does not mean you're violating copyright laws and have to make deals with every content provider. China doesn't do that. And I really kind of. I don't think Trump actually has a viewpoint on this. I think this is what his speechwriters have put together for him. And so this, to me, is one industry setting the rules for everybody here.

And also the statement, right, let's break down the statement nice and slow. So first, he said, when a human being reads a book, right, they're not breaking copyright if they learned from it. Okay, that's true. And China doesn't respect ip. Also true. What's not true is what's not said in there. What AI is doing is taking somebody's book and then making it so nobody ever has to buy it again, and that they can make money from it. That's what's left out of it. And that they didn't have permission. And when OpenAI was stolen by Deepsea, Sam Altman was not happy. And you can be sure anybody who has their IP stolen, if you were Google and your algorithms were stolen, you might not be too happy about it and you might file a lawsuit. So we need to look at this issue and say, this is something new. So how do you solve something that's new? Very simple. You come up with a new plan. And the new plan that I put out is super simple. If you want to explicitly use somebody's content, make a deal with them, and then you can have it be an exclusive deal or not. And you can very easily have somebody authenticate their New York Times subscription when you log into any of the. Any of the different language models, right? They already asked you to log in with Google, and they already asked you to log in with Google so you can search your Gmail. So why can't I log in with my New York Times or my Disney account? And then it says, okay, you have a Disney account. Now you can make Jedis. Now you can do generative art, right? And so we all agree on the output. So that's good. I always look for where we have consensus. We all agree on the output. You don't get to make Disney characters if you're not Disney. You don't get to leverage. You don't get to take my entire book or this podcast and republish it. Okay, so then what could you do? You could talk about what we talked about on the show, but I don't think you can make a super clip of it or use the entire archive. But if somebody said, hey, this week in startups is going to get paid. They have a subscription program. It's a subscription podcast. Great. Pay for it. I pay for Financial Times. I pay for the Economist. I would love to be able to go to the Economist and have the Economist say, log in with Grok, log in with ChatGPT, log in with Claude. Can you imagine if I could log in with my Claude account to those services and then use Claude against the FT and E, then that's really actually where this will wind up. And the people who do that in a reasonable way will have an advantage against the people stealing. That means it's an opportunity, It's a business opportunity for everybody. That means the financial time and the Economist, Disney can invest more in content and get more out of it, and that's where we'll wind up. Another option is I asked to have him on what's cloudflare is not letting you crawl. I don't know if you saw this story.

Yeah, Paper crawl is the new cloud.

Paper crawl is his new concept and he came out and said he's against what Trump is proposing. Or the administration. I should say. We should just say the administration, because this is a full effort. The administration's position he disagrees with and he's gonna block crawlers and he's gonna have them pay and the crawl is gonna be in there as well. And these two ideas. What's the guy's name? Prince. Is it Matthew Prince?

Matthew Prince.

Matthew Prince's idea can stand alongside mine, which is you could have your website. I could take the this Week in Startups website. You could take cautious optimism. You could put it on there. You can crawl this. It's a dollar a page. Every time you come, it's a dollar a page. That's it. And they might say, well, we're not crawling under $1, but if it's somebody with a $200 account and 2,400, they could say, yeah, we'll crawl up to X amount. And now you've got a royalty system. So I call bullshit on 100% that this is not technologically possible. It is 100% technologically possible. And is it possible to unravel it and be China? No. So since that's also true that you can't unravel it and that you have to be China, then that means we all have to work together, not run over each other.

Let's move on to how to sell your company. Or Jason, as I like to put it, how not to sell your company. Do you recall the windsurf OpenAI fiasco that turned into the Google buying part of Windsurf and leaving the we should.

Do the one thing was the J.D. and I, J.D. and I talked about immigration. I don't know. It's worth covering here. Was there any piece of it that was worth covering about it? Oh, we did get into Microsoft. Let me just, I'll just tee it up. I don't know if you have that there, the Microsoft discussion.

I don't have that clip.

So there was a moment where I was talking to JD about like, hey, let's put the border aside. We all agree on that. We're never going to agree about dragging 20 year immigrants out of the country that were let in here under Bush or Clinton. Like I'm going to hold the line on that. They should be given a path to citizenship. Sorry, we brought them in here. We had a bunch of joking. But the one where we had like an interesting debate was well, what about recruiting talent here? And JD made a very good point. Hey, I got wind of this company laying off thousands of people, like 9,000 people. And I was like, oh, that's Microsoft he's talking about. And then they're also filing for a bunch of high skilled immigration visas. What's going on here? I need to look into it. He didn't say he had it all wrapped up. Of course Microsoft's layoffs were across a number of countries. It wasn't just America, it wasn't just high skilled, it was salespeople. Right. Product manager. So it's not necessarily high skilled, but he does make a great point. And it goes back to the static team size or doing more with less. If you're laying people off, do you really have the ability to then lobby for high skilled immigration? I do think that's a valid point. Even though I believe in recruiting the most high skilled people here. What did you think of that moment? I don't know if you saw it or not.

I think the important thing to keep in mind because I've seen people make that point both online and of course at the AI Summit. I think these are not one to one replacements for the people that were laid off. These are often H1B being re upped for another term. So it's not like Microsoft is out there being like we're going to fire 9,000 people and bring in 9,000 other people that are being paid less. But I do think this conversation is becoming toxic enough for Microsoft's brand that Satya Nadella, the CEO, had to yesterday put out a memo entitled Recommitting to our why, what and how. And in this he discusses this exact thing. So I think they're getting beat up pretty badly in the media, in political circles for being so what would we call it? They're just, it's shedding people as they make record amounts of money.

And here's the interesting quote. You see some Silicon Valley technology firms, especially the big firms, this is JD Say that they're desperate for workers, they can't find workers, that they have to use overseas visa programs to find workers. And yet at the same, the college educated employment rate for STEM graduates in this country seems to be declining. Oh, wait a second. If you're not hiring American workers coming out of colleges for these jobs, then how can you say you have a massive shortage in these jobs? And by the way, you see some of these tech companies where they'll lay off 9,000 workers and then they'll apply for a bunch of overseas visas. It doesn't make sense to me. That displacement in that math worries me a bit. He makes a good point here. And I wonder, and I kind of know from being an insider for so long, here's the thing about H1B visas versus American college graduates, H1B visa are going to get paid 30, 40% less. And because they essentially become indentured servants, if they don't find a new job within 30 days of being laid off, they're kicked out of the country with their family. It's incredibly cruel and unnecessary. They should have at least six months to find a new job. A minimum. Minimum. Like it's ridiculous. So the people I know who are abusing the H1B visa system and other systems like this, I have heard people explicitly in my career say we're going to get these people because we can work them harder and they don't have any recourse, you know, to call them indentured servitude or slaves may be a little bit hyper, it's obviously hyperbolic. But they are less empowered workers and I would say significantly cheaper and less empowered. So JD makes a good point.

I would say that there was a, there was an interesting interview on a show called all in back before the election. And one of the hosts, I forget who it was, asked the then presidential candidate, what should we do about high school immigration and I think the thing was I'm going to staple a green card to everyone's diploma. That is my vision. Yeah. Now I know that that has not yet manifested in current policy, but I think that you're right. I think everyone that I've talked to, who hires people or who has been an H1B worker, because I have a lot of friends that have had those visas, confirms what you say. So what's the solution? Well, take away the stick that companies hold of these people's heads and give them more ability to come here, work at a fair market rate and stay. Because I think we should have every fricking smart person in the world here, and that's my policy. So I think JD here makes what I would call a classic JD point of not actually saying we shouldn't have H1B visas, not actually saying Microsoft is bad, but appearing like a populace that won't actually do anything once in office. Because he's currently the vice president, baby, do something if you don't like this.

The vice president is, as we've seen, it's hard to. You're not the show. But we'll see. I think he's gonna be, I think he'll be president. You have influence.

You had him on that. You had him out for a reason. You didn't invite a nobody on stage.

Actually, he wanted to come. I think he asked to come. If I, if I understand correctly, I think he's like, hey, you guys are in town, can I come on and we'll talk, you know, and continue the discussion. I thank him, but I wasn't involved in the invite.

I just, I don't like it when the problems and inequities in the H1B visa program become a cudgel to, to argue that we should restrict high school immigration in general to this nation. Because I think we should not. But again, my personal politics, yours, everyone listening may be different.

Yeah, I mean, probably 90, 80% or 90% of the country agree on immigration. We just don't talk about it in buckets. If you said we're going to bring 25,000 high skilled people here a year who start companies and if they don't start a company and create 10 jobs, then within two years they lose their visa, their golden visa. And it had that condition on it, you'd be like, okay, sounds reasonable. And if I said, hey, we should close the border so coyotes are not, you know, sending bad guys and charging people $10,000 and you know, all this horrible stuff that happens, you'd probably say, okay. And that'd be 90% of the country. And if I said, hey, you want to deport the like felons from the country who are here illegally. But yes, I agree. So there you go. Like it's like 80, 90% agreement on 99% of the issue. And the only disagreement is on what to do with people who've been here for over five or ten years and have made a life here. And that's probably 50, 50.

So from our friends over on the startups subreddit, which I read each and every single day, this is a question from a person who has worked at a couple of early stage companies and they want to know, is it possible that adding AI to your startup too early actually slows you down? And they write that having worked at a couple of companies that rush to add an AI feature E either to attract investment or just to stay current with trends, but it actually slowed them down so product decisions became more slow, MVPs became bloated. And I think that this question boils down to, Jason, when should you add something hot in buzzword terms to your startup to ensure that you maintain your place in the conversation and at the investor tables, but also avoid wasting your time and shoehorning crap.

You don't need a couple of different pieces here that are important. One of them is, should I frame my company as an AI company? Should I market it? Should I pitch it as an AI company? Well, absolutely, because VCs are looking at the market and they're saying these are the companies that are growing the fastest. Now, if it's just AI dressing, if you're AI greenwashing and there's no substance to it, well, you won't actually get the gains. Therefore why are you doing it? Doesn't make any sense. So you should authentically be thinking about that. And if you have a previous company that was a SaaS based company and you were pitching it as such, reframing it and in fact figuring out how AI would disrupt you and going there, not a bad idea. Let me give an example. Let's say you had a sales automation tool, Salesforce, HubSpot, something that does marketing and sales in a company. Well, you go and try to raise money for that concept, people are going to go, well, that exists already and it's going to get disrupted by AI. And they say, well, how is it going to get disrupted by AI? And you say, well, I'm an investor, it's your job, you're the founder. Tell me. So they have 10 people coming to them Saying, we're going to do to Salesforce and HubSpot what they did to legacy client server enterprise software before that. We're going to make a better version of that than the per seat model. Here's how it works. So you should absolutely be thinking about that. Take your legacy product making five or $10 million a year, let it cook. You don't have to change it. But then you launch the AI interface, the version, et cetera, in parallel, and you start marketing that immediately. Which, by the way, is all Benioff's doing. All Benioff does is go on TV and talk about agents and how a third of the work at Salesforce is being done by agents. There is some truth that AI is not ready to do certain job functions. We saw that with, I think it was Klarna that said they were getting rid of all their customer support, going 100% AI and six or seven hundred people were losing their jobs. And it was incredible. And then they said, oh, you know what, we need to have some humans in the loop here so you can jump the gun on it. And that's your responsibility as a founder, is to actually assess if the technology's ready. I looked at a lot of these AI solutions to make versions of podcasts in different languages. Not ready.

To translate essentially to us to Swahili or not ready.

You could do it as a proof of concept. I don't think it's ready yet. I don't think the people listening to us in Japanese as done auto dubbed is ready yet. Now it might be ready later this year, but it wasn't ready last year when I took the pitches. I don't know where it is now.

That's such an interesting idea. I love that.

Who doesn't love it? We should have it this week in startups in the top 10 languages.

Yep.

And they should automatically and you can be sure, YouTube. I know they are working on it. YouTube already dubs. If you go and type in Mr. Beast in Spanish, you can find the dubbed version of Mr. Beast officially done by YouTube. People don't know this, but YouTube is doing that for Mr. Beast. I don't have any inside information here. I just have what some YouTubers were telling me. So maybe I do have inside information. What some YouTubers were telling me that they were not using AI for this. They were using voice actors. Now, I don't know if that's true or not, but let's play play it and see if you found it. Did you find it?

So, yeah, so this is Mr. Beast in Espanol.

How did you find it, by the way? Is it on the YouTube Espanol site or is it in the drop down menu? And you just change the language.

So I literally just said Mr. Beast in Spanish and this was the first thing that came up. And it's completely translated. So the description is in Spanish. This is. I lived underground for 50 hours and listen to. So this is not just a translation. I don't think.

Jason.

This does feel more like voice acting to me.

It's voice acting. So they got. That's how big of a franchise Mr. Beast has is that YouTube is apparently paying for this experiment. And I guess for YouTube, it's super important for their growth in those countries, therefore. And type Mr. Beast in Japanese. Mr. Beast in French. I wonder if that even exists, if they're up to Japanese and French because those would be culturally very different markets and much smaller markets. Sub 100 million or way under 100 million because it would be. The subset of people in those countries is 100 million people in Japan and France might be 70 million French people.

I don't see anything similar for French or Japanese. I think it's just because Spanish is such an enormous language.

Yeah. So I don't know if it's public or not. But there you have it, folks. So to this person, back to this great question. You could look at the solution from first principles, test it with customers and then just decide. Right. You can just decide if the market is ready for that feature. I do think there's another point he alludes to here, which is distraction. As this AI comes in, it is so distracting because an intelligent person can look at what's happening and very easily extrapolate that it will work. The way I got to podcasting early, the way I got to blogging early, the way I got to zines early, and the way I got to angel investing early and syndicates early in my career is that when presented with a new idea, and this is my big unlock or hack for people, sometimes I just say yes to an opportunity and I assume it's going to work now. I said yes to Friend Feed. I spent a lot of time on Friend Feed by Paul.

Brett Taylor.

No Paul Boucher Bouquet. Like Booheid. Booheid. Thank you. Like Fahrenheit. Paul Buche, who created Gmail, worked at Y Combinator. He created Friendfeed. He sold it to Facebook. It was like Ajax. The page would automatically be updating as you would be posting to it. I spent like a ton of time on that for a Year I did a ton of time on Google. As those things grew, they fizzled out. They didn't work. Google Buzz, I spent a lot of time on that. Google Buzz could have beat Facebook if they weren't so concerned about the privacy issues. And they were a little more freewheeling Google, but they folded because they got too many privacy issues and Zuckerberg didn't care about privacy. The people at Google rightfully were like, oh, yeah, this is crazy. There's all these fringe problems with Google Buzz. And so you just have to think these things through. But if you assume something's going to work, you could hit, but most times you won't. AI presents the ability for you to believe anything is possible. We're sitting here talking about these robots, like, are those robots going to go pick up your lunch and bring it back to your house right now? If you said, I want a nice coffee. That robot you were showing earlier, the $5,000 one, is that going to go get it and successfully do that?

No, not. Not this year, but.

But you immediately assume next year it would. But it might be five years from now. So the timing for entrepreneurs is so critical. Is it one year? Is it five? Can your company survive for five years? Oh, by the way, scholarships are open. A lot of people have been asking me to come to the All In Summit. As is tradition, I set up the scholarship program in the early years and the team agreed to keep doing it. You can go to the website and apply for one. If you're one of my founders and you know me, I might have a little juice and be able to help a friend or two get their scholarships approved. It's probably a 1 in 10 chance that your scholarship gets approved because there's a limited number of those tickets. But if you are one of my peeps and you did apply for one and you can't afford the full price ticket, I might be able to help. So if you're one of my founders, go ahead and apply for a scholarship and let me know you did.

What is the scholarship? Is it like 80%?

I think it's like 80% off. I think it might be $1,500 instead of $7,500. So the event loses money on the scholarships because I think, yeah, no, I'm not surprised.

But it's profitable overall. You and I know the events game, and this is probably an overall good event for the business.

But All In Summit has been profitable since year one, and it's significantly profitable. And then all in, plus the tequila, plus the other events you see us doing. We don't have advertising on it, but it's a significant business. And this Week in Startups and I'm going to probably do two more podcasts on the this Week in Startups brand in the next year, so stay tuned for that, folks. He's cautiously optimistic. Cautious optimism.substack.com news news. Oh, you have your own domain name?

Yes, sir.

And you're still on Substack, not Beehive.

I'm on Substack for a little bit, but I'm working on some cool stuff behind the scenes, so that might change in the next couple of months.

What? I mean, you're thinking of your own platform or being part of something.

Let me sign, let me sign some contracts first.

Oh, okay.

Then I'll tell you.

Excited for you. There you go. Sign up. It's 100 bucks a year. You give me the support. Let's get him 100.

Or it's free. If you want to just not pay.

I think you should make it not free. I think you should just make it only pay because you got good insights and it's better to have a thousand people paying a hundred bucks than to have 10,000 people paying nothing. And, you know, whatever.

Stay tuned, Stay tuned. We're moving that direction.

We'll see you next time. Bye bye, X Com Alex X Com Jason X. Com TWI Startups Bye bye.