IBM Nukes, Demis Govt Plan, Paramount WBD Deep Dive | Dylan Byers, Noah Schochet, Saam Motamedi, Ioannis Antonoglou, Jack Dent, Evan Burns & Jamie Seltzer, Tyler Page

You're watching TVPN.

Today is Tuesday, July 14th, 2026. We are live from the TVPN El Dorado. The temple of technology. The fortress of finance. The capital of capital. Let me tell you about ramp.com. Time is money, save both. Easy to use corporate cards, bill pay, accounting, and a whole lot more all in one place. IBM is absolutely nuking. The stock is down 25%. Let me see if I can pull this up.

I got it.

Boom.

IBM. Well, that is a crazy chart. What is that?

The—

that's the 1-week chart. It looks better on the 5-year because the stock is actually way up in the AI era since the launch of ChatGPT. IBM has done really, really well. The stock has basically doubled since the introduction of ChatGPT during the AI era. You know, are you going to be a winner or a loser? Are you going to get steamrolled, slopped, something like that? But it's been doing well up until today when the company reset the narrative around their server business specifically. So the high-level reason that IBM is not well positioned in the token path, to use the Brad Gerstner and Gavin Baker parlance, is that AI spending is currently flowing into GPUs, memory, networking, hyperscale cloud computing, and frontier model inference. IBM is not a major winner in those categories. So just to refresh on IBM, 'cause interesting business, with a great name, International Business Machines. Um, the first business machines they made were punch card systems. They made clocks. Like, it's like, you're running a business, you need a machine, you need a great clock, you're going to need a clock. No, really, not just any, like a clock that works really well.

That's right, professional clock, Clock Pro Max, on time.

Exactly, Clock Pro Max, lighter, thinner.

It's the lightest, thinnest —best looking, fastest. Not the fastest.

You want a fast clock. But tabulating machines, basically a bunch of different ways to process information mechanically. And that foundational insight, you know, was pretty simple. It was businesses will continually pay forever to automate record keeping. And at a high level, that's sort of been working forever. And they're continuously paying.

You know, if they ever tried to sell a clock as a SaaS product? Time as a service? Hmm.

If you really, really squint, Red Hat Kubernetes, it's keeping time between distributed systems. Maybe there's something there. But when you're running a database across a bunch of different servers, there's some timekeeping aspect that's important. But no, I don't think they ever did. The IBM that people know, the mainframe business, that started in 1964. System/360. It was a compatible family of devices, which is interesting. It's not just one— people think one mainframe, but it was actually a whole bunch of different systems that you can upgrade piecemeal without redesigning the entire workflow. So you need a little bit more storage, you upgrade that. You need a little bit more compute, you upgrade that. And this turned IBM into the dominant supplier of corporate computing. Banks, insurers, airlines, manufacturers, governments, they all used IBM as the central system for their hardware and software. This was the mainframe era. —then, and the whole reason that IBM in particular became dominant in mainframes was they focused on high reliability, long customer relationships, expensive switching costs. It's very difficult once you're in the IBM ecosystem to weed your way out. Proprietary software tied to the hardware. Certain software would only run on IBM hardware, so you couldn't replatform. You had to rip everything out, which is very difficult for a large bank or a large airline in the '60s and '70s.

Good for business. And they also had huge—

huge support and consulting contracts associated with all the software and the hardware that they were delivering, sort of a precursor to the forward deployed engineer if you squint a little bit. But the PC era was the real turning point. So the IBM PC launched in 1981. This legitimized the personal computing market and set up two new companies, Intel and Microsoft, for— to capture immense value during the next computing boom. So the IBM PC ran Windows. And used an Intel chip. At the time, IBM was doing $30 billion in revenue. Intel was doing less than $1 billion, and Microsoft was only doing $17 million in sales. And so I think Microsoft had like 120 employees. And all of a sudden, those two companies became ultimately way, way, way bigger, like 10 times as big. So the market eventually fractured, and proposals to break IBM into separate companies started to pop up. The market fractured because once you had, you know, an Intel chipset and Windows operating system, you could run Windows on a different chipset and you could have a different chipset with different operating system. And the value capture piece, there were just other PC manufacturers that came in and then obviously Apple with their, you know, anti-IBM, like challenge the man campaign. So the market was fracturing and there was a bunch of proposals during the '80s and '90s to break up the country, break up the company into separate units. Lou Gerstner, who became CEO in 1993, rejected that idea. And he said, quote, "We do not necessarily need to manufacture every piece of technology. We need to be the company that makes all of it work together." So we have to work together. We're going to be the integrator, the systems integrator. His strategy ultimately produced 3 things— IBM Global Services, large outsourcing contracts, and a vast consulting organization. And that's a lot of what we know about IBM today. So services businesses do have limitations, though— lower margins, higher headcount, slower organic growth, price competition, et cetera. In 2019, IBM IBM acquired Red Hat for $34 billion and spun off its traditional managed infrastructure outsourcing business in 2021. So today, you can think of IBM as sort of 3 key businesses. They have software, which is 44% of the business. That's at 80% gross margins. Great business. 31% of their business is consulting. That's under 30% gross margins, though. And then 23% of the business is infrastructure, which is just shy of 60% gross margin. And so for the last 3 years, the stock's been doing really well, up 77% before dividends. And the Red Hat acquisition started paying off and the Z17 mainframe cycle was surprisingly solid. But the problem is, is that they just called out a shift away from mainframe spending with customers shifting capital spending towards the physical AI buildout. Demand for AI and associated hardware is strong, but IBM is losing share of their customers' technology budget. IBM still does have a strong asset for the AI era, Red Hat OpenShift, which is their enterprise Kubernetes platform for orchestrating workloads across multiple computers. But there are so many other companies offering AI capabilities up and down the stack that they're getting a little hammered today with the worst— the biggest share drop in its 115-year history. Rough day for IBM, but an interesting story nonetheless. Let me tell you about Shopify. Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in-store, on mobile, on social, on marketplaces, and now with AI agents. We have a great show. We have 7 guests coming on, lots of fundraising news across VC firms. And I'm thinking we're back. We're totally back. Dylan Byers is also coming up, coming on from Puck to break down the Warner Brothers Discovery.

Yesterday it was over. We had no fundraising news. Yeah, today we're back.

I think it's just like slow weekends.

People like— no one wants to announce Tuesday, Wednesday, Thursday news on a Monday.

Yes, but Demis Hassabis from DeepMind, the DeepMind chief, he has called for a US-led body to test frontier AI models. He says society has a precious window to prepare for technology advancing at historic speed. He's a Nobel laureate. The Financial Times has the story and there's an article that he posted on X that will sort of click through and give you the takeaways from the Financial Times. Google DeepMind Chief Executive Demis Hassabis has called for the creation of a US-led standards body to test new frontier-class AI models for national security threats, arguing that urgent action from international regulators is needed to address the risks posed by rapidly advancing technology. I'm surprised. Has he never proposed this before? This feels like something that has been proposed many, many times, but maybe I'm just I'm just misremembering the AI 2027 people and the AI 2040 people and the OpenAI white paper and what Anthropic said. Like, it feels like we've seen this before of like, we need to have a regulatory body of some sort. All the way going back to when the All In podcast was talking about an FDA for AI back like 2 or 3 years ago. But it's now here and it's coming from a DeepMind executive, which hits a little harder. The warning from Hassabis, a Nobel laureate who leads Google's AI efforts, follows the White House's abrupt export ban on Anthropic's most advanced models last month alongside a fresh wave of warnings about the potential for AI to disrupt the global economy and financial system. We talked a little bit about The Economist. They got together with a much more moderate proposal, I think, because it wasn't actually— it wasn't actually calling for any sort of change to the development of AI whatsoever or the rollout.

Just saying it could get— AI could get better in the next 10 years. Yes. Which is a very sort of—

Could get a lot better. That's what they said. They didn't just say that. They said a lot better. But, but the actual pitch from, from the economists was we need to have economists and government officials think about responses. If there is job displacement from AI, what is the impact? What will the reaction be? So to sort of like prep the legislation. So you can be more ready when things start to happen, whether that's retraining or stimulus or jobs programs or all sorts of different things. So this intervention from Demis is the most detailed proposal yet for AI regulation from Google, which is vying for AI leadership with Anthropic and OpenAI. Quote, we've already seen the challenges frontier models pose for cybersecurity. Good point. And other threats, including nuclear and bio risks, may soon emerge as capabilities continue to advance. The rapid progress we're seeing in AI requires a new approach to testing frontier AI model capabilities that is dynamic, adaptable, and rigorous. The US is well positioned, given its economic and technical standing, to take the first step in developing such a framework. My big question is, it seems pretty easy to go to the leading labs and say, hey, we— you have to go through this process. But do we have a good framework in the United States for reviewing code that China just sort of throws over here open sourced. Because if you, I mean, as we've seen with the KIMI K2 and GLM, like if you tie someone up in an FDA-like review for even 6 months, let alone a year, let alone what the FDA timelines are for drug development, 5 years, 10 years sometimes. You are gonna have open source models that are way way more advanced.

So like, well, you either have to review this very quickly. New Frontier model. Yeah, it's going through a, a 6-month review. Yeah, say maybe it's really like a 1-month review, and then there needs to be—

with AI, it could be a 2-minute review.

Yeah, I would hope. But let's say it's like a 3-month, yeah, a 3-month delay or 6-month delay. Then when it eventually does get released, it gets distilled likely within even less time than that. Publicly available. Yeah. Open weights. Yeah. I would like to see— so we keep seeing these like letters and proposals. Yeah. And they always come, one, with a request for urgent action, but they rarely come with super concrete scenarios like near-term scenarios. I want— here's what's going to happen in 6 months. Here's what's going to happen in 12 months.

Or, or even just a, like, a trigger. Like, it would be interesting if somebody said, if the unemployment rate goes above 10%, I would recommend a stimulus check of $1,000 be sent to everyone and means test it so it only goes to the middle class and lower class. Like, that is a very reasonable thing. That's basically what happened during COVID right? Like, the unemployment rate went to 15% and then boom, there were checks in the mail for everyone and That's a very concrete proposal that you could say, if this happens, then this happens.

Yeah, I want, I want someone like Demis. Yeah. Basically the world of LessWrong and AI 2027 and 2040. Yeah. They're willing to lay out super— sure, sure, sure. And they can at times come across as very sci-fi. Yeah, but there's always, at least so far, been some element of of reality in them.

I hear what you're saying.

I want somebody who's generally more moderate to come in and just say, here's a few potential scenarios and this is what I think. And because I don't believe it's— Demis could suggest what he thinks that the government should do, the US government in this case, he's encouraging like US watchdog. He's in London, though. So I think it's going to be on our lawmakers and our government to understand in these different scenarios, at least start thinking through in these different scenarios, how would we approach them? Mm-hmm.

Mm-hmm. Yeah, I just, I always have a problem with the, with timelines and predictions because those can get so nitpicked and they're so hard. I'd be more interested in less of Like, no, but don't you think that'd be helpful if— I don't think it's helpful. No, no, no, I actually don't. I think it's much more helpful to say if the unemployment rate goes to 10%, create a new government body that hires people to do something like create the next TSA.

I just think—

I just think send out stimulus checks or lower interest rates. Right.

You tell Washington, D.C., AI models are very good at hacking computer systems and they're going to get better at hacking computer systems. There's not really much for them to do with that because hacking computer systems are already— it's already illegal. Yeah. And the solution there is for companies to beef up their own cybersecurity, make sure they're using the most advanced models, right? Yeah. And so if you play out again, more concrete scenarios where like, here's, here's a timeline for the trucking industry. And potential job displacement within trucking or any of these other categories. I just think it allows people in Washington, actual lawmakers, to start thinking about it.

I just think that's always wrong. Like, they're always wrong about those predictions. Yeah, it's so much better to just say, look, if the trucking industry goes through mass job displacement, then here's what I actually propose. Here's the solution. As opposed to just saying, like, there might be a problem. And I think that there's a problem coming down the pipe. I don't know. Like, it's like, what are you actually advocating for other than just being like, the, the sky might fall, I have a P-Doom of this number, and it's your job to go figure it out? It's like, you're smart. What do you suggest? UBI? Higher taxes? I just don't think predictions are always wrong.

There's been so much, so many examples over the last decade where people have gotten predictions like dead on.

Yeah, situational awareness. What do you think about this?

Yeah, I mean, I'm probably in the camp of like proactive regulation. Like usually it like has some bad consequences. It doesn't really work out. But also I was going to say like what he's describing is basically just Casey, the Center for AI Standards and Innovation, which is under the Commerce Department. Yep. And it's like a slightly beefed up version because right now Casey is like very much you opt in. Yeah, but it seems like he should just have said like you should specifically be beef up Casey, add these policies, do these specific things. And I think that would be much more palatable or well received or something. What do we actually do with this letter? It's kind of very, I'm not sure what we actually do with this.

Yeah, it feels like if, I mean, to go back to cybersecurity, it's like if it's a national issue, like the NSA works on this stuff, increase their budget maybe? Raise taxes to increase their budget? Issue more debt to increase their budget? If it can be solved by the private market, it's like, go support CrowdStrike or start a new company that can help with cybersecurity. I don't know. The actual concrete recommendation boiled down from what Demis wrote is something along these lines. Create a US frontier AI standards body that's like Casey, but probably more beefed up. He's also advocating that it's overseen federally, but funded by AI companies. Define and regularly update benchmarks to determine which models and labs qualify as frontier. So that's something that doesn't exist yet. Require frontier labs to submit models for testing up to 30 days before release. That's sort of nice because that would allow someone who's just going and building a recommender system on Netflix that's actually, it is using AI technically, but it's not a frontier model. Because it doesn't qualify for that. So then they can just go and ship the latest recommendation algorithm on Netflix, no big deal. Um, test models for cyber, biological, nuclear deception, autonomy, and guardrail-passing capabilities require strong cybersecurity, personal vetting, model cards, watermarking, and substantial safety research. Use national labs, federal agencies, and independent third-party auditors to conduct evaluations. Develop independent confidential tests so labs cannot train specifically against the benchmarks. Require labs to fix serious vulnerabilities discovered after release, apply the rules to all frontier models deployed in the United States, including foreign and open-source models while exempting smaller models. Okay, so he wants to apply it to foreign open-source models. That feels very tricky, but I guess you could get like sort of DMCA notices to Hugging Face and GitHub so that it doesn't proliferate across the web.

But yeah, I mean, there's probably some power law to like where people are actually downloading and inferencing the model.

And I guess if you go to all the Neo Clouds and all the open-source folks and you're like, okay, This model is actually a bad model. You gotta go, you gotta stick to GLM 5.2.

Yeah, it's much easier to regulate the compute.

This is gonna be very controversial to the open source fans.

Yeah, this is kind of like the George Hotz nightmare take.

Yeah, for sure. Attacking every single GPU.

For sure, for sure.

Coordinate a slowdown among frontier labs if testing reveals sufficiently serious risks and turn the US framework into an international system of shared frontier AI standards. I like the general direction. I like the idea that he's just sharing his viewpoint more broadly. I think all of that is good. I'm not sure that there's enough to dig into here exactly how this would— like where the rubber meets the road, how this would be implemented or what effect this would actually have on the industry. Like this could be really good for open source because it could just slow down the frontier closed source labs. It could also be really bad for open source if it's much more cumbersome because an open-source project might not have a regulatory budget to actually massage a model through the approval process. Like, there's a reason why small biotech companies get acquired by big pharma before they launch their drugs. It's because, like, the big pharma companies have offices in Washington, D.C. and can walk the legislators through the whole process. So in general, I'm sympathetic to the view where People say, oh, regulation benefits the biggest companies in the world because historically that's how it's played out. Maybe that's different this time. Who knows? It could just slow down the frontier. But, you know, he does work for a leading lab. So interesting. Let me tell you about MongoDB. What's the only thing faster than the AI market? Your business on MongoDB. Don't just build AI, own the data platform that powers it.

And what's going on in New York? I will tell you what's going on in New York today.

New York Governor Kathy Hochul signed an executive order placing a 1-year pause on new AI data centers in the state. This is the wah wah for everyone except 7% of Americans. People that are probably outside the state. But the order establishes a moratorium while New York develops a regulatory framework and conducts environmental impact assessments examining data centers' energy demand, water use, water quality, air quality, and effects on the electric grid. You would think there's a decent amount of oversight around those things generally already, like air quality, like whether you start a new barbecue restaurant or a coal plant, you would imagine that there's just a general rule about not polluting the atmosphere that would apply to data centers by default. But it seems like there's a little bit of a special case here. And so they're working on that. This in particular. The move immediately drew criticism from the tech industry, which argues that restricting data center construction will cost local communities jobs and weaken America's position in the global AI race. Earlier this year, Maine considered a similar moratorium, but Democratic Governor Janet Mills vetoed the proposal after concerns it would block a major data center planned for a town still struggling after the closure of a local paper mill. Hochul's Republican challenger, Bruce Blakeman, also opposes the moratorium, arguing that local governments, not the state, should decide whether to approve projects that promise significant economic benefits. If it stands, the order would make New York the first state to impose a broad moratorium on large-scale AI data centers. The Associated General Contractors of New York State is already condemning the decision. So the, the contractors who are going to be working on this project and see it as a form of job creation are upset about this. But let's see what President and CEO Mike Elmendorf had to say. He warned that halting permits for as much as a year in this fast-moving sector will not simply delay projects. It will send them permanently to Virginia, Texas, Georgia, and other states actively competing for these investments. Once those projects break ground elsewhere, he argued, the jobs, tax revenue, and economic opportunities are unlikely to return. So there hasn't been that much of a data center boom in New York State that I'm aware of.

I'm trying to find the, the largest campus that I can find is the Lake Mariner campus in Barker or Somerset. Yeah. And it's around, it has around 205 megawatts active. Oh, that's pretty big. Wow. Yeah, that's pretty big. And they, there's a proposal out to do another 500 megawatts. Yeah.. But it hasn't been approved yet. And, and certainly that the new ban would, would stop that as it's any data centers requiring 50 megawatts or more.

I know a lot of, uh, a lot of IT infrastructure data centers, if you can call them that, they're usually smaller scale for, from financial institutions in Manhattan, often are built in New Jersey, but that's basically purely for economic reasons that the land is cheaper and the buildings are cheaper. But I do wonder if this will put have any knock-on effects where the Netflix content delivery network that was just planning to build a small data center to route video streams to Manhattanites would be delayed as well. Data center, it'll be interesting to see how they define AI data center. Will they do it on energy or what type of GPUs you're racking or what's going on there? More to dig in.

Ken Griffin was on Goldman Sachs's podcast, The Exchange— Exchanges podcast, and it was circulating this week, even though it was, I think, recorded last month. And he was talking about how, yeah, in his view, what an error this would be to— his view, the data centers are going to get built. And if they're not built here, that means hundreds of billions of dollars of revenue basically flowing through other countries, right?

And other countries, I mean, it'd probably go to other states first, right?

Well, he was talking about it. Oh, if it goes national. If New York does it, there's going to be a lot of other states.

Yeah, the meme is like China wins in this scenario. US Senator John Fetterman.

Yeah, we did this with nuclear. We did this with manufacturing, both of which we can all agree were were mistakes. I was trying to find if any of the hyperscalers operate our own data centers in New York.

Well, the funny thing is that when Mark Zuckerberg proposed the Hyperion data center, the visualization he used was we're building a data center the size of Manhattan, which was very cool to see. But now Dylan Patel is sharing an image of, you know, a proposed data center that takes up all of Central Park. Of course, the most controversial data center you could possibly build in the entire world. Tyler's in favor of it, but it's funny because that image probably stuck in some people's mind as like, "Oh, Mark Zuckerberg's trying to build a data center in Manhattan. I don't like that idea." Even though, no, he was always going to build Hyperion in a very remote location for a variety of reasons. There's also an article in the Wall Street Journal, "Can a prettier data center curb the community backlash?" People have been been batting this idea around for a while, but let's pull up this image and you tell me, would you be okay with this going into Malibu? The Malibu Compute Company. Would you be cool with this if it was puking out diesel fumes 24/7? If it was clean and it didn't drive up energy, didn't use any water, it was all closed loop and it looked like this, tolerable, right?

I wouldn't just be okay with it being in my town. Oh, demand. I want it in my backyard.

Yes. True Yimby over here. That's right. In an effort to soothe local opposition, architects plan data centers that resemble tech campuses or art museums rather than bland boxes. You have to imagine that the money that they're spending on the data center for a facade like this has got to be very, very cheap by comparison. It looks like a slab wall. Yeah, half a percent. Less. And all of a sudden, just every time it's screenshotted, Like, there was that odd Google presentation where they were in front of those crazy tanks and they put the logo on there and it made it look like they were taking like a brewing facility and turning it into a data center. But it was just for the press release. Like, the data center was actually somewhere else, but it was just sort of like an odd image. Americans are up in arms over data centers. Of course, we know this. They worry how much water these buildings use and fume at the amount of electricity they consume. People hate the way they look too. Says the Wall Street Journal. Now a small number of builders are on a mission to ensure that new data centers don't have to be eyesores. Gensler, one of the world's largest architecture firms, is leading the charge. It's drawing up plans for data centers that look more like Silicon Valley tech campuses or art museums rather than windowless rectangles that neighbors often grouse about resembling prisons. It's no different than any other building, and it doesn't deserve to look any worse than any other building, said Jeffrey Diamond, a design director at Gensler.

Yeah, see, this is just very rough. Yeah, not good. That is underground. Objectively, you got to do more.

Yeah, yeah. You get— you, uh, people will push it to the limit unless there's some pushback. But, uh, in other aesthetically pleasing AI development news, Clanker Media shared that Researchers built a soft floating robot for indoor interaction. And for so many of the AI robots, of the humanoid robots that we see on the show are Lovecraftian and horrific. This is so cute. I want one. Don't you want one just floating around answering your questions?

These have the— this has the potential to be a like a massive hit. Yeah, consumer product.

It uses helium and flapping fins instead of propellers. Extremely cute. The result is quiet, lightweight, and safe to touch. It can follow people, give reminders, and act as a study buddy. So you can be studying and this whale can come up next to you and answer your questions about your math homework.

See, I don't even need it to be smart. No, I just wanted to fly around.

Yeah, load it up with GPT-2, it's good enough.

No, we don't need any, we don't even need any local models. Just have it fly, just have it fly around.

Oh, you don't even want to—

I just want it. Yeah, I just want it. You just want it, uh, for the—

I, I personally would demand that they install Codex on this thing. Let me tell you about Codex before we bring in our next guest. Codex is a powerful workspace for getting work done with AI agents, whether you're writing code, analyzing data, creating content or automating business workflows, Codex helps you move projects forward from start to finish. And our next guest is in the waiting room. We have Dylan Byers from Puck. He's the founding partner, and he's here to get us up to speed on Paramount Warner Bros. Discovery and the acquisition. How you doing, Dylan? I'm good, gentlemen. How are you doing?

We're doing— have you been on vacation?

You look incredibly tan.

Very tan. Yeah, what's going on?

What's the secret? I'm tan, I'm unkempt, uh, I, I eat No, I'm in the Pacific Northwest. I've been traipsing around the Pacific Northwest for the last month.

For— you've been traipsing?

Business or pleasure? Yes, like a logger. A little bit of both. Business and pleasure. I was in Sun Valley during the Allen Company conference to meet with a bunch of folks who were there. And then I— thank you. And then I am— I'm from here. I grew up in Seattle. So, you know, what better place to spend the summer?

We had a question about Sun Valley. Who had the bulkiest security guard? Who's the security guard who's the strongest physically? Oh yes, bigger.

Historically, it's Bezos. I know, I know Barry Weiss has some beefy security guards. I read about that in the New York Post. Yes, I did hear that. I didn't see them. I didn't see any security guards there. You know, here's what I will say about Sun Valley, about the Allen Company conference. It is remarkable how little— I mean, they create a security perimeter, but it's very subtle. And it is remarkable how comfortable all of these executives who are used to security details how comfortable they are just moving around the property. They really feel like there's minimal, minimal attention even to the fact that they're there from the locals other than like myself and reporters from Bloomberg and CNBC. Yeah, they are largely left alone and they feel that.

Sure. How— what is the interaction? We always see those photos of like the line of microphones at the fence for the ad hoc interview. Obviously, there are scheduled interviews with CNBC and Bloomberg sometimes. Sit-down interviews. But what is the day-to-day of actually reporting on Sun Valley like? Are you grabbing quotes from people or are you meeting with folks?

I did different strokes for different folks, right? I mean, there are historically, there have been like New York Post reporters who go there and once in a blue moon try to breach a security line and get color and quotes. I go there to meet with people off the record. That's like, for me, like that is the world I cover. Those executives, be they in media, tech, entertainment, whatever. Yeah. And I, I have off-the-record meetings with them at a café on the property. There are others who go there and do get interviews, like Julia Borstein of CNBC always gets 3 or 4 interviews. Yep. It depends. But what I do not do, thankfully, gratefully, I do not stand there at the front and shout questions at people who then just sort of like wave at you, which is really their way of giving the finger. Save for David Zaslav, who's like a moth to the flame of a microphone and will come over and always give a good quote.

I love it. How real is the, the meme that like this is where the mega mergers are birthed? This is the place where the ideas come together. The merger that you'll hear about in a year for $100 billion, it all starts in Sun Valley.

Well, look, here's the deal. I have, I have I've tried to set the record straight on this. Deals do happen there. Oftentimes they happen as the culmination of conversations that happened well before Sun Valley. And occasionally, like, conversations start there. And I don't want to downplay that. Like, it is true. Like, if you just, you know, this year, there's not much that I know of, at least yet. Sometimes we don't know about it till later. If you look at last year, like, yes, Apple got the rights to F1. Yeah. At Sun Valley, and there were meetings between Liberty and ESPN and Apple, and Apple won Sun Valley last year was where not only Larry— or sorry, David Ellison, but also Matthias Döpfner at Axel Springer and the Murdochs came and tried to court Barry Weiss. And of course, David Ellison won that one. So, yes, deals do happen. I think what I reject, and I've written about this, I reject the notion that all of these people with their private jets who basically run on similar circuits throughout the year, that somehow, like, if Sun Valley didn't happen, they would have no way to make deals. Sure. And that somehow these brilliant people, these brilliant entrepreneurs and executives who have managed to, like, run their businesses successfully and are always thinking not just months but years, if not decades ahead, are like, oh, I happen to be in Sun Valley with this person and therefore I will buy The Washington Post. Sorry, that's not the way it works. Yeah, but, but that said, everyone loves coming here because it is the same reason I love going there. It's shooting fish in a barrel. It's just like speed dating with everyone you want to see and is just extraordinarily convenient. And then going back to my point about the Pacific Northwest in the summer, there are worse places to be than amid the quaking aspens, you know, in Ketchum. In July.

Yeah, that's great. Well said. So get us up to speed on what's going on with the Warner Brothers Discovery acquisition. Maybe a little bit of the prehistory just to refresh on where the deal stands and then the latest back and forth that emerged last week trickling into this week.

Yeah, look, David Ellison has had to jump hurdle after hurdle to try and get this deal done. Obviously, $111 billion deal. Bring Warner Bros. Discovery into the fold with Paramount. And the last hurdle— he might face some challenges in the UK, but really the last hurdle here in the States, now that he's cleared everything with the Trump administration, are these Democratic state attorneys general led by California AG Rob Bonta. Finally, Bonta came out this week, sort of very cinematic. He was like on a trail in front of the Hollywood sign. And, and effectively said like, yes, we are suing. We are trying to block this deal. Here's why. We can get into it. It is very hard for me to see that he has a, a strong case for blocking the deal, that he and the other AGs have a strong case for blocking the deal. Because if you think about it, like, David Ellison wants Warner Bros. Discovery to have some semblance of the scale that would be required to compete in a world that is dominated by, you know, like, pick your brand— Netflix, YouTube, TikTok, Apple, Amazon, like the list goes on. And so what, what Bonta and co are doing is sort of presenting this like very strict definition that is rather archaic, right? It has to do with like theatrical distribution, cable distribution. Issues like that. And I get from a political perspective why he feels the need to do that. And I think there are probably political ramifications for him if he didn't do that. But the notion— I mean, like, just think about it on its face. We all live in the real world in 2026. The notion that somehow a combined Paramount Warner Bros. Discovery, a combined CBS, CNN, combined— what are the streaming services? HBO Max, Paramount Plus. Is going to be this like a massive dominant business that is going to somehow like, like, you know, corner every aspect of the market. That's just not real. That's not the world we're living in. And so I think Bonta has an uphill climb in that regard, but he can certainly create headaches and certainly legal fees for Ellison for, for many months.

And so zooming out, the, the like Hollywood is under attack. Loosely or, or facing competition from tech platforms, YouTube, Instagram. And so time, time and attention is more broad than just Hollywood. And so with that backdrop, while consolidation might not be good within the view of Hollywood, there's less buyers for movies. It might be a defensive move in a world where you're facing extra competition for attention across the internet. Yeah, that's right.

I mean, it's just how do you define the competitive marketplace? Yeah. And if you're talking about, you know, the fact that, look, Hollywood has— Hollywood is in decline. I don't think anyone can dispute that. And frankly, like, there are comparisons to Detroit, you know, of decades ago. And I don't know, you guys are in Hollywood. I live in Hollywood. Sometimes it certainly feels that way. Yeah. So there's this notion that like, okay, we used to have 6 studios and then we have 5 studios, and if we have 4 studios, that's going to be a problem. And so all of this sort of, you know, there are all these negotiations over, okay, well, we'll release 30 films a year, we'll keep both studios open, we'll have a 45-day theatrical window. And I'm sorry, but that's just— that is not where the— that is not where consumers are. That is not what consumers care about. That's just not what the market cares about. Like, you're not people aren't— people, people don't think that the one form of entertainment that they have in the world exists in a theater. And they care about a 45-theatrical-day window like we live. I mean, to even say that we live in an era of like Netflix and YouTube almost seems passé when you think about everything coming down the pipeline in terms of AI platforms and everything else. So it just seems like remarkably outdated, which I suppose is an evergreen commentary any time you have Washington going up against business, entertainment, media, tech. And again, I just don't— I don't see how Bonta wins that case. Jordi, do you have a question?

Yeah, it was just the— how did the threat to leave California?

Yeah, because it feels like there's a short-term incentive to stop consolidation in Hollywood. But if you wind up with, you know, studios leaving Hollywood, well, you've lost the war even if you won the battle. But How, how—

Yeah, it's just, it's just reminding, it's just reminding me of the, of the whole Getty Images Shutterstock deal, which is, was announced in January of last year. It was a $2 billion merger at the time. Now both companies are worth roughly like $300 million in the public market.

So two companies that are under attack and I'm trying to get to the idea that that this does create like a stock photo monopoly. But in the world of AI image generation, like a monopoly in the real world, stock photos actually taken with a camera is probably the only thing that can go up against just random image gen. And so it's a very logical move.

But yes. Well, look, let me, let me just say first about the threat to leave California. Yeah. One, everything that every headline we see needs to be viewed through the lens of brinksmanship, right? Everything is basically trying to win the hearts and minds of anyone who's paying attention to this deal. And the threat to leave California all of a sudden would seem like, if you take it at face value, would seem to reverse the narrative. It's like, okay, Bonta's, you know, ostensibly coming out here to try and fight on behalf of Hollywood. And like, lo and behold, his measures are going to drive Paramount Warner Bros. Discovery out of Hollywood. A lot of people, I think, looked at that and thought that that was a bluff. And there's no world in which, you know, Paramount WBD can leave Hollywood. Yeah, I don't know. I mean, look, I do think it is a bit of a bluff. I do think it's brinksmanship. On the other hand, as folks I spoke to reminded me, like David Ellison did not get into this business either to just have like an old-fashioned Hollywood business that couldn't really compete with the big tech players, nor did he get into it in a David Zaslav kind of way to just like flip the business in 3 years. He's young. He's into like— he has ambitions of grandeur. He has ambitions of empire building. He really does believe that he can position this company to compete with all those other companies we've already mentioned. And if you think about it through that lens, Is he nostalgic about Hollywood itself and like the actual location of Los Angeles? No. If you look at what his father did with Oracle, he moved Oracle from Redwood City to Austin. I believe now it's moving from Austin to Tennessee. And there's other precedent, right?

You have Tesla, you have Chevron, you have Hewlett-Packard.

Disney, too, moved a lot of people to Florida. During COVID and there was a lot of backlash and not everyone moved and they still have a huge presence.

And now is, now is it like, you know, 30 years ago if you said, yeah, I'm going to have one of the biggest entertainment companies in the world, it's not going to be based in Hollywood, you would have been laughed at. But now you talk to anybody here that's like actively making movies, TV shows, etc., and they're already traveling a bunch. So it truly doesn't need to be here.

No, you're traveling a lot. Different states have different incentives for filmmaking. And by the way, like, go back to the Oracle Tesla examples. Like, you still have businesses here. You can— you could— you could theoretically have a Paramount Warner Bros. Discovery based in Austin or Nashville or anywhere else that still had studio lots in California. And to the, the, the, you know, the folks in the writers room, the showrunners, the talent, they wouldn't know the difference. Yeah. So like, and the only person who would feel the difference is Rob Bonta, who suddenly would not be on the end of what David Ellison is touting as a $30 billion investment. Yeah. And at a certain point, Bonta is walking a very fine line because he's trying to fight for Hollywood, but he is running the risk of alienating one of Hollywood's most valuable players right now. And so it's, it's just It's, it's politics. It's pure politics. And another point I brought up this week in my writing is like, just to underscore how much this is politics, I want you to imagine that David Ellison, who, by the way, is not some like right-wing MAGA guy, like he's not— he's, he's donated to Biden, whatever. He's really just a pragmatist who's trying to get a deal done. He would— he took Trump to ringside at the UFC. He went to the State of the Union address. He held a He held a private dinner for Trump and his administration before the White House Correspondents' Dinner. All of that to like the folks in the newsroom at CBS News looks like, you know, he's like the new Rupert Murdoch. Sure. The truth is, is if Kamala Harris or Joe Biden were in power, he would have done the same thing. It just would have been at like, you know, the US Open or Wimbledon instead of a UFC match. Sure. And, and like, and the fact of the matter is, is that in that scenario, Rob Bonta would probably not be bringing this case, and it would probably be Republican AGs who would be, who would be trying to bring this case. And I think we have to be honest about the politics there.

Yeah, I like the idea of tennis at the White House. Maybe that'll happen soon.

Um, right, yeah, US Open on the White House lawn.

Yeah, maybe just all sports venues, the hockey, and then basketball, and then baseball. Was there adequate sponsorship?

I didn't watch the UFC match on the White House. Was there adequate sponsorship? Like, did they milk that for our—

not nearly enough. We could have gone way further.

Yeah, it's unclear.

There were a lot of sponsorships. Yeah, yeah, yeah.

I remember a specific image of Justin Gaethje, the, you know, the American champion who won his fight, doing a backflip with a Polymarket logo. And you can see the White House in the background. And there he comes. It was very iconic image. Yeah.

Can you talk about the debate that sprung up, I think a little bit earlier once the merger was confirmed around the duplicative studios, Paramount, Warner Brothers, they have two. When we say studios, there's the abstract concept, but there's also like the physical historical locations that are very meaningful. There was this discussion over is one going to be turned into apartment buildings and that feels like sad from an I Love Hollywood perspective. Also, if you're a housing person, it feels like maybe it'd be good to bring down housing prices. But. Right. But, but at least the reporting that I saw from, from Puck was that there was a lot of pushback towards like, don't ruin this amazing financial history, this historical institution. Is that chip still on the table? Has that been resolved? Because there's been a lot of like Okay, we're making this pledge, we won't do this, we won't cut. And I'm just wondering if any of those chips come back on the table.

Well, here's what I would say to the one thing that Bonta has gotten right. Forgive me because I forget verbatim what he said, but effectively, like in spirit, it's I can't regulate pledges, right? And that's true. And like, there is so much that is going to be said, which you should effectively take with the same grain of salt that you take from like a a political candidate who tells you everything they will and won't do in office. Like, yes. Is he committed right now to 30 films a year? Is he committed to a 45-day theatrical window? Is he committed— again, I don't have the latest on this, but to keeping both studios open? Sure. Once you've got the deal and once all of this is behind you, are there ways in which you can sort of finesse one of those lots into something else for your business and then maybe finesse that real estate? Out of the portfolio to save some money? Yes. And that will 100%, 100% happen. I think about this, you know, I covered this deal a lot at the news level and the sort of like tie-up of CBS and CNN. The minute this deal happens, folks at Paramount are going to start looking how to offload the real estate that is currently used by CBS News and move CBS News operations into Hudson Yards at CNN. And why you do that? Because you have $80 billion of debt and because you have promised synergies. And there's no world in which you're sitting there with two lots that are duplicative, that do the exact same thing, and think like, in order to honor the history of Hollywood, yeah, we will continue. We will continue spending twice as much money as we need to.

It's just not going to happen, especially not in the news business. Like, no, definitely not. Like You can sort of make the claim around specific film studios being historically important to Hollywood. Hollywood means films. But when you think about the, the what it actually takes to run CNN and CBS News, like there are tons of overlapping, I mean, just pieces of equipment that you're like, oh, that teleprompter there and that backdrop and that lighting is like all the same.

And sure. And look, like, I am sure there are some Scott Peli types who have very warm feelings about the labyrinthine halls of the CBS News headquarters up on, up, up in uptown Manhattan. But there was a group of folks from CBS News who went to Hudson Yards, who went to Atlanta to tour it, and they're like, holy shit, these facilities are a lot nicer than ours. These are great. This is— it would actually be a lot easier to create TV here. So that, that will be really easy. And look, I think there is— as someone who covers change in the media space, there's always so much panic. And nostalgia anytime anything changes because you think that thing that's changing is sacrosanct. And the truth is, is that it usually takes a matter of weeks or months before everyone sort of comes around to the new status quo and gets very comfortable with it. And that inevitably is what is going to happen here.

Speaking of change in the media industry, have you noticed— I mean, you, you report and operate at what I see as like a very, very high level, the executives, the the mega dealmakers. But have you seen any knock-on effects from the summer of YouTube blockbusters that we've been talking about here from the creator economy perspective? You have Iron Lung from Markiplier, Obsession, Backrooms. There's a number of YouTubers who sort of, you know, labored away in obscurity and then just put up incredible numbers at the box office. It feels like Hollywood might be looking for more scripts, more interesting IP, less sequels, less bought-in things. But is there any individual studio that's excited about that trend or anything that you're seeing that's actually rising to the level of your reporting? I—

here's what I'll say. I do not have as good of an insight into this as I do into other things we've been discussing. But here's what I will say just from a distance. Every studio is thinking about how do we— like all the way to A24, right? Which is sort of like the belle of the ball when it comes to quality. Content in Hollywood is thinking about like, how do we incorporate AI into what we are doing? And I think the signal that those creators sent with films like Obsession and others is like, this can be done and this can be successful. Does it mean that, you know, like every human being is going to go away and we're going to stop? Like, no, it does not. I think that most folks will learn— will find ways to incorporate this in ways that does not significantly, like, degrade the value of the content that they're making. But I think the sky's the limit. Like, I really think that those examples have shown that this can be done and that there is more opportunity here than the folks who are sort of, like, clutching to Hollywood's past have realized. And by the way, that doesn't need to be disruptive in a bad way. That can represent, like, extraordinary opportunities in the same way. Like, you guys talk about this day in and day out. AI creates extraordinary opportunities in every aspect of business, industry, commerce, everything else. And like, that can happen here too in ways where the human element remains. Mm-hmm.

How much are you tracking these AI short-form vertical drama apps? There's— I'm counting 3 in the top 25 of the US iPhone App Store right now. There's Vibe Shorts, AI comic dramas, Net Short Exclusive Short Drama Video Space Exclamation Point and then Story Reel Exclusive Drama.

I'm not following any of this. It sounds just hearing you read it, it sounds like— is this just Quibi AI? Yeah, it's AI.

AI Quibi. Katzenberg entirely vindicated on the format. Yeah, but But yeah, I've just been— I don't know anyone that uses these apps. I think we need to have Tyler on our team just spend like an entire weekend, watch every single one of these and review them for us. But it's funny to me that, that these are— these apps are seemingly being created not within the tech industry. Yeah, they're tech, they're technology products and not within Hollywood. So it's like entirely just outsider groups that are ranking— The hoverboard content. Like NetShort and VibeShort are both outranking Kalshi, Paramount, Polymarket, like during these massive sporting events, right?

Crazy. I mean, I guess, yeah, I get it.

Like VibeShort is outranking Gemini and Anthropic by a very meaningful margin.

Yeah, I guess here's what I say, and I'm granted, like, I'm a very poor authority on this because my entire content has been like World Cup and Wimbledon or just staring at the quaking aspens in Sun Valley. Here's what I would say. As in every other industry, like, the technology has democratized the ability to do things, right? And so, like, to bring this conversation full circle, there's no no real reason anymore that this has to happen in Hollywood. Now, I'm not suggesting that like, you know, every institution just goes away, obviously. But what I am saying is like, there is— you cannot look at something like the success of the YouTube creators or the engagement that these, these short-form platforms you're saying, like the engagement that they're seeing, and just dismiss that that as a fad. That is a total reorientation of the possibilities that exist for people to come along and just create something that is going to be a massive hit. And again, to bring this back to Ellison and to Bonta, you cannot argue that people who can put movies in cinemas on Hollywood Boulevard are somehow in a different business than the people who are making AI-generated Shorts. Yeah, it is solely a question of where people are spending their time and then whether or not that time, that time and engagement can be monetized. Yeah. And that— and it does not matter what category that falls into. It's just a matter of a question of eyes on screens.

Do you have a view on how US-China relations might evolve with regard to Hollywood, because it feels like some of the video models coming out of China are a little too good. And they might be training on some Hollywood intellectual property across different studios. And it feels like in order to actually take a lawsuit to the finish line or have any sort of settlement or retribution or some sort of even just truce to not allow you to put Mickey Mouse or Spider-Man or Batman in an AI-generated open-source video model or something like that. It might require Hollywood sort of teaming up, going to Washington, asking for some sort of concession or, you know, deal within the construct of like the larger trade negotiations. Around rare earth, and it could be one of many bullet points in a new deal.

Yeah, I mean, I do say, like, and we're here and we've been here for a long time, like China somehow seems to be the great unifier when it comes to businesses saying, like, basically as the antithesis, right? Like the— it has been years that I have been hearing tech executives invoke China as a way to try and push back on regulation of US companies. Companies. Oh, I can certainly see a world in which China becomes like the, you know, what's the term, the bugbear, like, oh, that gets Hollywood to basically come together, work with, uh, regulators, and say like, we need to protect this because this is American. And that is a very palatable argument for the folks in Washington or the folks in Sacramento. Yeah, yeah, I can totally, I, I can totally see that. I guess the, the thing I would say though is I think what businesses really want is to be able to play in that market. And I don't— I think at a certain point, your ideal outcome is a way to protect your own IP, your own technology, while also being able to open up what is obviously like an incredibly robust and large market and work with them as well. And that, you know, that's a negotiation that obviously like Tim Cook has been involved in for a long time. A lot of this when it comes to the US-China stuff is above my pay grade, but that is sort of my sense of what will happen.

Last question. Quaking aspens. You've mentioned them twice. Are you— I love them. Arbor— what is it? Arborist. Arborist. Are you into trees or is this just something everyone that goes to Sun Valley recognizes about the trees? Here's what— here's the—

thank you for asking. I've been wanting to talk about this for a long time. No, here's what I think. I wish I could convey to people, I'm very privileged in my position as covering the world that I cover to get to go, like it is sort of a treat of the media, tech entertainment space that these guys all sort of choose to convene in the Mountain West. Yep. So depending on where you are in the sort of hierarchy, like you're either going to Sun Valley or Jackson. There are many retreats sort of in Colorado. A lot of these guys own ranches from like Montana down to Utah, There's of course like the Aspen Ideas Festival. It is just like a real pleasure that these guys chose to convene there as opposed to say Cleveland, right? Or like just do everything in New York City.

Or Vegas. CES is in Vegas and that would be, you know, where you'd be if you were on the electronics beat probably.

Thank God I'm not. Thank God I'm no longer the politics beat and have to go to like Iowa and New Hampshire every year. Yeah. If there's any way to just sort of convey the atmospherics of that world, I know of no better way than, than like the lull of the quaking aspens and the occasional sound of like a Gulfstream jet taking off on the tarmac.

Beautiful. Are you writing a book right now?

I don't know yet. Should I?

You should be. You're tan, you're hanging out in the mountains. It seems like you have an incredible energy.

Maybe an audio book.

It seems like since the last time you were on here, through it.

Okay, maybe an audiobook and it's just a recording of the quaking aspens and the sound of Gulf Streams passing by. There you go. To fall to sleep, to fall asleep to. There you go. Uh, thank you so much for coming on the show.

This was fantastic. Great to see you.

Have a great rest of your summer. Hope to see you soon, and we'll talk to you later. Have a good one. Cheers. Goodbye. Let me tell you about Figma Agents. Meet the canvas. Your AI agents can now create and modify your Figma files with design system context. The quaking aspen, it's a tree. They're defined by the natural feature of Sun Valley, Idaho landscape, famous for their pale white bark and shimmering autumn foliage. The quake is a flat leaf stems that cause the leaves to flutter and rustle at the slightest breeze.

So when you look at the tree— Dan says Gulfstream ASMR. Gulfstream ASMR might be good. Be a market. Tyler, I need you to watch a few Vibe Shorts, a few Net Shorts, and a few Story Reels by tomorrow. Deal? Deal. And we're gonna— I don't know if we can watch them on the show. We can probably play a trailer on the show, maybe. But yeah, pick your favorite on the 3 apps and, uh, we'll circle back tomorrow.

Well, our next guest is With us in the waiting room, we have Noah from Terra Firma. He's the co-founder and CEO. Welcome to the show. How are you doing? Doing well.

How are you guys doing?

In the safety vest, looking great with the flag in the back too. I love it. Please introduce yourself and the company. Tell us a little bit about Terra Firma.

Yeah, so Terra Firma was founded by my co-founder and I after we were working at SpaceX and we were trying to build Starship rockets down in Boca Chica and the construction was just too slow. We're like, why are we able to build rockets the size of skyscrapers at like one a month? But like building a road or a factory is taking years and years and years. So we quit our jobs to start a company to change all that. We're basically becoming a new type of construction company, like a full-stack construction company that builds their own robotics software. And we actually operate as a construction company ourselves.

Okay. Robotics software. So are you— how deep in the supply chain are you going to buy equipment? Are you buying from Caterpillar and retrofitting them? Are you buying from the people that make the tracks and the treads and the motors and then building your own bulldozers and cranes? Are you using custom software and robotic automation just as a harness for the project management layer? At what level of the stack are you operating now and where do you want to go?

Yeah, on the hardware stack, you can probably see behind me, we take existing Caterpillar machines and we retrofit them. Caterpillar, John Deere, Komatsu. Those machines, they're pretty good. They're robust. They've been around for hundreds of years, 100 years. They're pretty good. Everything else we build ourselves. So we're building all of our own software from scratch, kind of the manufacturing engineering software that we built at SpaceX in order to operate large-scale factories. We're building that for construction. Mm-hmm.

And how did you actually go from zero to one? At this point, you've raised $115 million. What was the first project? To find the customer, just sort of do it all manually, or did you actually build robotic tools first and then go and prototype them? What was the flow?

This is a great question. Um, so, uh, 12 months ago we were only 4 people, so for the first year of the company it was just my co-founder and I. We wanted to make sure we had product-market fit before we started like raising a bunch of money and spending it. Yeah. So we actually built, uh, a bunch of robots using Tupperware containers that our lunchbox came in, and like Raspberry Pi and Arduino from our college like kits, and we got robots working. And then we went around asking like, who is willing to let us make some money with these robots? And our landlord's like, hey, I got a building for you to demolish. I think you guys can do it. Right? I mean, it's not rocket science. So my co-founder and I literally operated the robots ourselves, like in the brain was a little Tupperware container. And we knocked down a couple buildings and got paid our first check. And we're like, whoa, you could just become a robotic construction company there. It's Texas, like you can do whatever you want down here.

That's awesome. What I feel like you'll quickly get to the point where permitting is like the gating factor where like you get to the point where hypothetically we could build this, you know, we could build this whole, you know, shell in like the shell that you're in right now. We could build something like that in 3 weeks, but then you're limited by sort of local government and approvals and things like that. Is that, Is that right? And if so, how are you thinking about approaching that problem? That's a great question.

People bring it up a lot. Um, for starters, I just want to say like permitting is an issue, but it's not the issue. Like at SpaceX, once we had the permits, theoretically it should go super fast, but construction still took longer than it should have. Like we're operating at maybe 25% of a work week, 9 to 5, 5 days a week. We could operate at 168 hours a week. We could be at 100% utilization rate and we're not. So you can make construction 4 times faster once you get the permits before permits are the issue. Now, our approach to this is very similar to like how Elon approached building rockets. Like, okay, we have to solve manufacturing, we have to solve supply chain, we have to solve testing the rockets and designing them. And now the rocket's on the launch pad, fix the permits. And like, I believe in the bureaucrats of this country wanting to make the country better. If you have everything else ready to go and you make their jobs easy, you chew the food for them, they'll start approving permits faster and faster. They're not currently the main bottleneck most of the time.

I imagine that we are firmly in the Centaur era at best. I imagine that every robotic piece of equipment has a human in the loop somewhere at this point. But walk me through your vision for where this goes. I imagine that in the long term, it's sort of, you know, You create a spec for what you want to build or destroy or the construction project that you're working on, and all of the robots go off and do everything. But before that happens, it'll probably be one human overseeing two machines. Like, how does all that play out?

Another way to think about it is like, you probably don't want to be like taking the approach to like same approach as like vibe coding with like building a building, because it's like, hey, go build this ditch. And then it builds it like slightly in the wrong place and then you have to like, you know, fill it back in and, you know, it feels like not as forgiving as the software domain.

This is a great question. Do you want the economics answer? Do you want the sci-fi answer? Both.

Let's do economics first, sort of like short-term. I mean, we're familiar with Waymo. They have, you know, Overwatch drivers that might see 4 screens. They can sort of beam into one at a time? How— what are the economics of that? And then where do we go from there? And then let's talk sci-fi.

So in terms of economics, you have two curves that cross. You have the cost of autonomy, which becomes exponentially more expensive as you go from 90% to 95% to 100%. Waymo spent $30 billion or whatever it was over 20 years. And you have the value of automation drops off a cliff. When you have one person controlling one machine to one person controlling two machines, wow, 50% reduction in labor. When you're at 1 to 4 changing to 1 to 5, The difference between 25% of your labor and 20% of your labor. So I think those two curves cross in construction at about one person controlling 3 or 4 machines, about 75% autonomy. That's the goal for us, 75% autonomy. And then you add a new machine and then you add a new workflow and then you go from dirt works to subsurface utilities to concrete to steel. Oh, that's interesting.

We're building up the tech tree. Yeah, and that sort of makes sense because I imagine that when you're actually just one, like old school piloting a construction machine, there's probably about 60% of your time is like downtime. Like, I know I need to get the crane from here to here and I've now, you know, push— put my finger on the button and I just have to hold it there while I get there. So while that's happening, you can sort of beam into a different machine, do something else and sort of cue them all back and forth.

Yeah. It's a little bit more video game than that. You ever play like StarCraft or Age of Empires? Oh, yeah. Yeah, yeah, yeah. So we're able to say excavator, dig here. Yeah. And you design a task for a minute and it will run for 20 minutes. Then you say bulldozer, run here. Sure. Roller run here. Yeah. And this is actually our teleoperation center. Sure. We have like 20 desks behind me, and these 20 desks allow us to operate, um, like 80 machines at a time.

That's very cool. Uh, so what are your—

what are— yeah, on the sci-fi front, timelines to, you know, you have a piece of dirt and, uh, robots arrive to the piece of dirt autonomously, and you have zero humans in the loop, and you just get a structure and a very short period of time.

You don't want zero humans in the loop. It's not cost effective. There's so many edge cases in construction.

Yeah, but I'm saying like, but, but if you continue to execute in the way that you're doing right now, wouldn't we be able to get to that point in 10 years, 15 years?

I'm not sure if construction will ever fully be without a human in the loop. I think you'll have like a very small team of people that's very efficient building larger and larger projects, and it doesn't ever make sense to get rid of the human. That's my personal belief. Even on Mars, I think we're going to have a small team of astronauts conducting like a large fleet of robots. There's just a lot of things that edge cases and things that the robots have never seen before. You can't collect enough training data to simulate every possible thing in the world. But if you're asking me the SimCity moment when I can say I want a road here, I want a restaurant here, I think we're like 5 to 10 years away from that. I think that coincides with our Mars ambitions.

Mm-hmm. Coincides with Mars. 5 years to Mars. What are you thinking?

We are technologically capable of going to Mars today if we had the political and economic desire. I mean, we've actually sent robots to Mars.

Like, America has sent robots to Mars, so it's possible.

He's like, "Send me, send me." Yeah, yeah, yeah.

It's not a technology problem. It's more so like an economics problem. Yeah. But I think we will solve both within the next 10 years. How?

But I imagine that the machines that you build for the moon and Mars will be much more highly specialized. When I think about the helicopter we deployed to Mars or the rover, like those are science experiments meant to take samples and test specific factors of what it means to move around on Mars. But I imagine even in terms of building, you wouldn't just plop down a Caterpillar, would you?

No, diesel engines won't work in vacuum. So you gotta redesign the hardware. But at the end of the day, it's like a smart shovel. Like you're moving dirt. It's not the most complicated thing in the world.

Sure, sure, sure. Very cool. Well, let's hit the gong. Yeah.

Last question, just because the crowd is loving it and I wish we had more time, but how do you— do you expect any changes to like the materials side, like building materials side as robotics become much more a part of of building various types of structures, or is it like, let's just change one thing at a time? Because I feel like—

This is an awesome question. Are you familiar with the term DFM, like design for manufacturing? Yeah. Like at SpaceX, we literally had to, the materials were pretty simple, it was like stainless steel, like the most common type, but you had to change the inputs to your system to be assembled by a robot. Like Starlink couldn't take V1 Starlink and make it at the speed that they're building V4 Starlink. They had to change the piece parts. I think we're gonna change the way we connect pipes. I think we're gonna change the chemical formulations of concrete. We're gonna change how we weld steel structures. We're going to redesign the building codes to match something that can be built quickly by a robot. That's amazing.

I figure— tell us about the round. How much did you raise?

So we've raised $115 million. Who'd you raise it from? So our Series A was $100 million led by Kleiner Perkins. Kleiner Perkins.

Oh, we got Bain Capital, Banner VC, Saga Ventures, Trust Space Ventures, Magnetar. Wow. Saga. Murdoch's Crow. Clay Burke Capital. You got a bunch of great— Legends. Folks around the table. Congratulations. The American industrialists. Glad you have the capital to build. We need it more than ever right now. So thank you for everything you're doing and have a great week.

Do us a favor. Raise another massive round. Yes. Come back soon. Maybe take it up from 9 figures to the next zero. I like it. Come back on this year. You deserve it. Awesome. Great to meet you now.

We'll talk soon and we'll host our next one from the Mars base.

Fantastic. We'll talk to you soon. Have a good one. See you. Let me tell you about CrowdStrike. Your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. Our next guest is returning to the show. We have Sam from Greylock Partners. He is a partner there. Sam, welcome back to the show. How are you doing?

I'm doing great. Great to see you guys.

Thanks for having me back. Thanks for hopping on. Let's kick it off with another gong hit since it's warmed up. Tell us what happened. Give us the news. What's new in Greylock world?

We've, we've got some big news today. We're announcing Greylock 18, our 18th venture fund. It's $1.5 billion. Yeah.

Yeah. Early-stage founders. Early-stage founders specifically. I mean, I want to know opportunities created in AI. I'm interested in the next generation of AI companies. Obviously, there's so many huge private— now SpaceX is public, so like there's There's a number of DeepMind, there's a number of like first-gen AI companies that are huge, but there still is a lot of opportunity. Where are you seeing the opportunity? There's early-stage companies that are going deeper in the stack selling to the big labs. There's also niche players. Like, how are you viewing the market? How are you mapping things out?

Yeah, I'd say overall is we're very early in this wave and we think there's gonna be winners up and down stack. And I think the only mistake you can make in AI is underestimate the size of the wave and think it's binary in outcome. So, you know, we've invested up and down the stack at the foundation model layer. You know, we're investors in Anthropic and OpenAI. Yeah. In infrastructure, you know, we invested in Base10 back in 2019 before generative AI was a thing. Today they're the leaders in inference, right? Yeah. Braintrust, I remember in 2023 no one was using the word evals, right? Notion was using their product. And today, fast forward, Eval's are the core IP of AI and most of the Valley is using Braintrust for it.

Right.

So we've done a lot in infra and then of course a lot in applications. We're going to see new opportunities up and down the stack. So on the infra side, I think, you know, you guys are talking about agents every single day. There's going to be a new set of abstractions around infrastructure to actually power agents. Of course, inference and observability is going to play a big part of that. But there's more to do there. There's new things in data structures, databases, run times, etc. On the application side, I think we're just getting started. It wasn't until, you know, the model set of model releases in December of '25 that you actually had robust agents. And now, like, and I see in our portfolio, like, we led the initial round in Resolve, which is an agentic software reliability engineer. And today, companies like Coinbase, it completely automates on-call. So engineers are no longer getting woken up because there's an incident, because the Resolve agent fully handles it autonomously or We're in SevenAI, which does the same thing for security teams. Those products are like coming to life because the underlying models are finally there to really support these long-running agent workflows. And so you're going to go into different application categories and be able to build new agent offerings. So I think that's going to be really big. And then, of course, things outside of software like we, you know, we were some of the early investors in autonomy like Aurora, which is, I think, the largest standalone public company in self-driving. We led the initial round in neuro, others. We're now going to see a whole new wave of things happening in robotics. And I think the underlying model capabilities and the datasets that we're building are finally going to get us there where we're going to see a proliferation in the real-world economy as well. And we've made some new investments, are still in stealth. We're excited to talk about soon.

Yeah, we talked about Aurora yesterday because they bought ATG from Uber. They advanced the autonomy group. I'm, I'm interested in hearing, are you noticing any categories of company that have shifted from venture-backable to more likely to be successful as a lifestyle business because they are so capital light in the AI era that you, if you have a friend that comes to you and says like, I wanna start a video game company or I wanna start a niche SaaS company, company, you're like, yeah, 20 years ago you'd need to raise $10 million to rack a bunch of servers. A couple of years ago you'd have to go through YC or raise a million bucks to hire some software engineers. But this, you can just build it this weekend and go get customers and be profitable tomorrow. And so are you noticing that trend? How is that— how is that changing your investing philosophy or is that just a ruse?

I think the lens on whether or not something should raise venture capital is more about the ceiling, not the floor. So I think it's absolutely correct that you can get a lot more done with less capital now. Yeah, but what's also correct is that the size of the prizes that are available and our whole brand around the new fund is the map is blank again because we think AI just resets the entire map. Yeah, the size of what's possible now, the size of these categories is so enormous. Yeah, that venture capital dramatically, even if you can be capital efficient and get a lot done on a given dollar, if you can increase the size of your capital base, the aggression and speed with which you can move is just much faster. And I think like— I think it's an interesting kind of observation. You take the last 3 years, there's a lot of discussion around what you can get done with AI, etc. At the same time, the capital velocity in companies is faster and at a larger order of magnitude than it's ever been. Like, we look at this in our portfolio. You know, we're primarily doing the first rounds and the speed at which companies we raise their Series A, Series B, Series C, and, you know, the size of these rounds, which often are in the hundreds of millions, you know, very, very quickly. Like, that's an interesting observation. Yes. Why is that? And I think it's because we're operating in such big— like, in such big new categories and opportunities that the prize to be the number one player is so significant that you need to do everything you can as a company. And so if you can be more efficient, The real lens is like, what else can you go do and how quickly? And that's why we're seeing companies have much faster product velocity, have much faster revenue growth. They're spending, they're getting more out of that spend, but they're spending. And in fact, I would say they're spending more in this last 3 years than in the prior decade.

The map is blank. I love that as a rallying cry. At the same time, I see a new market map in every category and it's stuffed full with AI. Startups and it feels like it's more competitive than ever. I'm interested to know, are there categories that are now more oligopolistic, more like, like less monopolistic? It feels like the Software 2.0 era was very much, oh, there's a runaway leader in this particular niche and it's winner take all and they're just compounding. But now I've been so surprised when I check in with the, the third biggest or third most well-funded company in a particular area and they're accelerating revenue and they like tripled revenue last year. And it's a great company by historical standards, even though it's like a laggard in the modern era. Yeah, it's a great question.

And yeah, it's interesting, by the way, like we— there's all these market maps, people crown these companies as winners. Yeah. And I think we forget that like the game's just beginning. I mean, you rewind the clock just 18 months and you look at what the foundation model landscape landscape looked at at that time. And then you fast forward to today and we fast forward another 18 months and there's going to be more evolution. So I think we all live into it on Twitter where we're trying to clear everything over like every single minute. Like we, we should zoom out and remember these things are very nebulous. But yeah, yeah, I think, you know, the difference with AI is the size of the prize, especially when you're automating labor and you're shifting these labor dollars to software and technology dollars. Is so dwarfed, like, relative— or so dwarfs relative prior software categories. They think you can have more winners that can get larger much faster. And I'll give you just two examples. In the customer support category, we led the initial round in Cresta. Cresta is north of $100 million in ARR, you know, growing very quickly, doing customer support AI. There are likely— there's at least two other significant companies in that category, maybe more, you know, at similar revenue value scale. And I think multiple companies in that category will get to a billion of ARR in the coming years. And the reason for that is just AT&T alone spends a billion dollars a year in contact center spend. And you extrapolate that across all enterprises, you think about what's addressable.

Yeah, it's way different than IT spend, right? Uh, like, because there was a war for Zendesk, and, and, and Salesforce had a solution, and there's a number of previous era, you know, software products, but it's a very different pool of spending to pull from. That's right. That's right.

What are, what are your first checks? What have your first checks been looking like? It's a $1.5 billion fund. You're investing. You want to be doing the first— like, 2 early-stage deals. Yeah, just $750 million. No, but yeah. So, so how are you thinking about it? What, what percentage of the new fund is going to be reserved? Follow-on capital versus like, you know, net new investments?

So, you know, for us, the North Star is we want to back the most important companies of this era. And we think some of those companies require much more significant checks when they get started because of the people who are starting them, the ambitions, the opportunity. So as one example, you know, we— 21 years ago, we backed Palo Alto Networks when it got started, got started in our offices. I think that company today is almost at $300 billion in market cap. When we wrote the initial check, uh, uh, I think it was a $250,000 initial check. Wow. Fast forward. Yeah. The founder of that company, Nir Zook, left Palo Alto last year. Yep. He started a new company with Greylock. Yep. Right. That company's called CyLake. Right. It's building a really new, important cybersecurity platform. Sure. Our initial check there is $36 million.

That's what I think.

That's— yeah. That's what I think.

You want to back—

Nir built the largest company in cybersecurity in history, a $300 billion company. We want to back him to go build something of consequence, and that has a different capital requirement. And so we have a larger fund, not because we can—

we just got a goal. Sorry, World Cup's on. Who scored?

Who scored? Spain just scored. There we go. There we go. There we go.

Sorry, we had to cut to the fan cam. I respect it.

Anyway, who you got? Who you got in the game?

Who's playing? So Spain's up 1-0.

I thought France was going to win this easily.

Oh, well, Spain's on its hair, I guess. But you are too. And thank you so much for coming on the show, breaking it down for us. And congratulations on the new—

I'm stoked for you guys. So for you. And thanks for having fun talking to all the new founders out of the new fund. Yeah, we'll talk to you.

Thanks for having us. Have a good one. Goodbye. Let me tell you about Cisco, critical infrastructure for the AI era, unlocks seamless real-time experiences and new value with Cisco. So yeah, what actually happened there? I have not been watching the game, but yeah, there was a Spain goal, 22 minutes left.

Okay. Okay. 1-0 Spain. Was, you thought the PK was fair? Yeah, yeah.

Very nice shot. I don't know.

He like kicked Lamíñamón on his like side.

It's kind of like a bang, bang, bang. Spain play.

Okay, I want Spain to win. Bang bang.

Okay, well, our next guest is in the waiting room. We'll bring in, uh, Yanis from Reflexion. He's a co-founder, president, and CTO. It's his first time on the show. Welcome to the show, Yanis. How are you doing? I'm doing all right.

Thank you so much for the invitation.

Thank you so much. Hopefully we didn't pull you away from the World Cup. I'm not sure if you've been following, but, uh, it's 1-0.

Hasn't even seen a minute. No, he's locked in. He's locked in.

But since it is your first time on the show, I would love a brief introduction on yourself, and then I'd love to hear about the news and the deal with Nebbius.

Absolutely. So first of all, again, thank you so much for having me. I'm Giannis, I'm the co-founder of RefleXion. Before that, I was one of the founding engineers of DeepMind, joined really early, and I spent most of my career there working on deep reinforcement learning research. So anything from, from DQN, which is the first deep reinforcement learning agent to ever exist, to AlphaGo, AlphaZero, MuZero. And I was doing RLHF, I was leading RLHF for Gemini before I left. And left with Misha to start Reflection where we're building frontier open models and we want to ensure that open intelligence remains open and accessible to everyone.

So I imagine demand for the business is through the roof because you just signed this big new deal. What's actually driving it? What are the enterprise use cases that you're seeing? Who are the key customers? What's the shape of the customer base? And then why now with this particular deal and why Nebbius as a partner?

Yeah, so I mean, let me just start by saying that building Frontier Open Models requires two things, right? It requires a lot of incredible talent. And we've been extremely privileged with the fact that many of the best people in the industry have actually joined us to work on Frontier Open Models and work closely closely with me. So just really like for that. At the same time, the other thing that you need is compute, right? This is what fuels AI research. That's kind of the thing that keeps the researchers busy. So to this end, we actually signed this billion-dollar deal with Nibius to just get the compute that we need. And that's— Congratulations.

Thank you. Amazing. I'm interested in if we can shift to history for a second, just your reflections on, no pun intended, sorry, just the progress in computer use. Recently I saw a demo of Codex and 5.6 SOL playing Slay the Spire this card game that I played a lot of and I know how hard it is and it sat there for 5 hours and played the daily challenge. And I'm wondering if, are we ahead of your timelines in terms of generalization? What is your overall thesis on progress? Where have you been surprised based on what you obviously had a front seat to a preview of years ago?

Yeah, that's a really good question. I think that, It's actually like, you know, I've been in the industry for like 15 years and I've actually like seen the progress we've actually made in the past 15 years. And, you know, every time that we felt that there was a wall or there was like something stopping us, we just kind of like overcame it almost immediately. So it's kind of like really, you know, this is like the best time for anyone to be doing like AI research. And, you know, I think that I'm not surprised anymore. I think I've seen so many things in the past 50 years, so it's really hard to be surprised. Sometimes, this exponential growth or exponential curve, it's really hard for the human mind to just fully understand it. But things just change extremely fast, and we should just be really adaptable and really understand that AI is still on this exponential curve.

Incredible things are ahead of us. But once you're on, once you're living on the exponential, you sort of internally, like emotionally, operate on the second derivative. And so you're sort of like, yes, this is as expected, once you've internalized it fully. But, uh, yes, I, I agree that it's shocking.

Can you, can you talk about the advantages and disadvantages between, uh, Chinese, uh, labs that are making open models versus American labs like Reflection that are making models? Because I imagine it's like, you know, you might have better access to compute, maybe more capital, but potentially more restrictions around, you know, things like distillation that could provide advantages in some way. But how do you think about that competition? Yeah.

So, you know, one thing is that like it's really opaque exactly like what the Chinese labs are doing. Like, on the ground. Some things we know is that maybe there hasn't been as much of a respect to IP and restrictions in terms of use of data that we know. As an American company, of course, we are fully compliant. At the same time, there have been accusations. I don't know if they're true or not, but from US labs that the Chinese labs are actually doing distillation at an industrial scale. From their models. And that's not of course something that like any Western lab would ever do. And there are definitely gains in terms of access to computer, things that we can get access to like the latest compute. At the same time, I've also heard that like the Chinese labs have found ways to bypass maybe the restrictions in terms of compute, like via getting compute from like other places. I think that we have definitely the advance of having the talent here in the sense that the Frontier Labs, the Close Frontier Labs are based in the United States. And many of these people have actually been to the frontier. They've seen the frontier and many of them have actually chosen to join us now and just work on our models. So that's definitely a benefit. Yeah, I think this is kind of like the world as it is. But I don't think that the fact that we are a Western lab and we cannot cheat means that we won't just build the most powerful open models. I think it's actually the opposite. The fact that the closed labs also never cheated and they are at the frontier means that we have the talent, we have the know-how, and we now have the compute of all these deals to ensure that we catch up to the Chinese open frontier. And after that, our ambition is to just close the gap between closed and open and ensure that frontier intelligence is accessible to everyone.

Awesome. In keeping with bringing access to frontier intelligence to everyone, your former colleague, Demis Hassabis from DeepMind, just today is advocating for a US frontier AI standards body. And he specifically says that he wants to apply rules and, and tests, submit models for testing that would even include open source models. Have you grappled with that? I mean, you're a large company, very successful, $8 billion as of the last round. And yet, as a, as someone who can be a little wary of overregulation, I I don't want to slow an exciting company down that might not have the resources to staff a huge lobbying group in Washington, D.C. to get open-source models approved. So how are you grappling with the idea of, as models get more powerful, deepening your relationship with the United States government? Yeah, so I think it's important—

as a reflection, we also always want to engage and to ensure that the models are safe, they're well received and they actually meet standards. So we just do whatever is required from us. I want to ensure, and I think many people in the community feel the same, that there is a frontier open model in the US market. This is kind of the bedrock upon which research institutions around the country are building their solutions. This is the tools that developers and early-stage startups use to just like build new products and kind of like drive innovations. So we need to just like be— we need, we need to just ensure that there is like room for frontier open intelligence in the United States and in the Western world. And I feel like this would be just a net positive for the society as a whole. So, you know, always here to engage in any way we can, just like from— and also like express the voice of people who believe in open— in the open ecosystem.

Yeah, I think my nightmare scenario would be some situation where you're slowed down, but international firms are not slowed down by some weird quirk. And so I hope that whatever happens, we at least get a level playing field for everyone, both closed source, open source, but also international. You don't want to advantage a geopolitical rival by accident, but that makes a ton of sense and I'm sure you'll navigate You did it all flawlessly. Thank you so much for coming on the show. Congratulations on all the progress. And— Yeah, great to meet you.

Really great to meet you.

Have a great rest of your day.

Great meeting you too.

And thanks so much. We'll talk to you soon.

Let me tell you about Railway. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy web apps, servers, databases, and more while Railway automatically takes care of scaling, monitoring, and security. Our next guest is Jack Dent from Chai Discovery. And he's here with us now in the TVP and Altarum. Welcome to the show, Jack.

How are you? Welcome. Thank you. Great to be here.

Thank you so much. Could you quickly introduce yourself and the company? And then I want to go through the news, but I'll let you start with an introduction. Sure.

So I'm Jack. I'm one of the co-founders here at Chai. At Chai, we're building what we like to call the computer-aided design suite for molecules. So think like Photoshop or Blender or SolidWorks, except Instead of designing a car, you're designing molecules. These are small collections of atoms which are often used for things like medicines and that actually go inside our bodies and do things like cure us. And so we operate at all levels of the stack. We're both a frontier research lab, also a product company, also a bit of a science company, and tie all of those things together to put that piece of software in the hands of some of the largest pharmaceutical companies in the world.

So specifically narrow to pharma, but I imagine that you will have applications in material sciences all, all over the place. But is, is pharma just the best landing zone or is that such a deep market that you'll stay there in perpetuity?

Yeah, I think pharma is right now where we're really focused. Our models, they work at the level of atoms and molecules and that you also require different categories of models when you're trying to do different tasks. So it's not like a specialized model that will design an antibody will necessarily like out of the box work for material science as well. That's still some additional work. But, but pharma is also, you know, this is a multi-trillion dollar industry. It's actually larger than the chip industries in revenue. In and of itself. And so it's this massive market that obviously does huge amounts for human health. And there's just so much runway to build into that.

So what actually is different about the models that you're developing? Do you have the same sort of massive CapEx around pre-training runs and scale is all you need and you need to marshal all this compute? It doesn't have a different shape. Is there a different— data puzzle that you're solving? How important is your feedback loop with your customers? How does your lab look different than sort of a lab that's focused on coding, for example?

Yeah, it's a great question. It's really all of the above. So when I say we're a research company, that means that we actually train our own models from scratch. These aren't like LLMs that we take off the shelf. Fine-tune some open-source architecture. We're starting at, you know, PyTorch, and then below that, down even at the kernel level sometimes, and creating models which actually look quite unlike the sorts of models that you build elsewhere. It's sort of like how if you're training a video diffusion model, that's a pretty different architecture to what you train for language. These, these just different model categories require fundamentally different avenues of research. You know, one piece of intuition here is in the same way that video models sort of think in 2D and they put pixels on a grid. And so, you know, autoregressive token prediction with a language model maybe isn't the best fit. You know, our models think in 3D, really. They put individual atoms in 3D space to construct these molecules. And so that means we really need to have a command of the entire stack and then the data is as well. This is not natural language data. It's very domain-specific data. It's these vast troves of protein sequencing data sets and structure data sets where people have used essentially really powerful microscopes to go and look atom by atom at the structures that exist in nature. And then we also do a lot of data generation in-house as well. We throw that off as well. So— and that's just at— the research layer, then there's a whole, you know, product layer on top of that as well. It's almost like, you know, we're training Claude as we're also building, you know, Cursor or Cognition. Sure. And so we need to do those things hand in hand. So there's also this, this feedback loop with the scientists and our— at both at the pharma companies, but our users, many of whom sit inside the company and dogfood and try and push the frontier of these models. So it requires us to be just A+ across both science, AI research, and product engineering as well.

Please. Have you seen any scientists get quote unquote one shot in the way that certain Vibe coders have? Like staying up all night? Where they're just like staying up all night.

Laptop open with Chai Discovery running at all times.

Yeah, because I mean, it's such an interesting challenge that the labs have faced where you're trying to balance like, like making products that are enjoyable to use, effective. In a weird way, you know, you are optimizing for engagement if you're optimizing for revenue, just because someone's more engaged, they're using more tokens. But I'm curious if you've seen any signs of that, because so far the labs that have experienced the most extreme product market fit have gotten to the point where there's this idea of like one-shotting. And then you have to like, you know, align the model better and improve that.

Yeah, totally. And actually, you know, maybe it's worth just providing some historical context here on just how quickly the research has moved in this field. You know, a year, a year and a half ago, basically none of this stuff worked. You know, it was— or the success rates were extremely low, you know, 0.1%. And over the course of 2025, these models went from being these research curiosities in that they were maybe interesting things to study in some academic labs, but they weren't really being used in real drug discovery workflows. In July— June, July of last year, we put out a paper called Chai-2 where the title of that paper was Zero-Shot Antibody Design in a 24-Well Plate, literally getting at this exact point where you could start to design these molecules without needing to fine-tune them on a lot of data. You could really just prompt them in the way that you'd prompt an LLM with an input, a target, maybe some disease you're going after, a target that's implicated in that disease. And it would design a number of candidates that would help do that, essentially. And so that created some really interesting situations where we would go into a pharma company company, and we'd be presenting some data. And there's one that comes to mind in particular where somebody pointed to one of our slides and says, oh my God, you solved that one? Did you choose that on purpose? And we're like, no, we didn't. We didn't choose that one on purpose. Why? And they said, you know, I spent 5 years of my life working on that target, trying to find a binder to it. So this technology has really moved beyond— I think AI and drug discovery has been in the realm of promise for a really long time, where there's been a lot of hype and attention around it. But in 2025, we really crossed the Rubicon in a way. And now these technologies are actively being deployed in companies like Eli Lilly, Novartis, Pfizer. These are some of the most scientific, highest-tech companies in the world. And they're putting them into real drug programs and making them part of their core discovery engine. And I think that's what might get missed here. So much has changed in the last year on the technology side. And it's created this wave of new applications. And what's amazing is the scientists themselves are the most happy about it. They don't love the fact that they fail 90% of the time or that their work—

Yeah, or they could dedicate— half of their career to one disease and make, you know, modest progress, or maybe their entire career. Do you think there will be something— the music industry is funny right now because, you know, every artist is using AI, or not every artist, but many of them, both casual artists, non-professionals, you know, amateurs, but also professional musicians are using it. But no one wants to talk about it. I expect that pharma will be much happier to talk about it. But is that the right intuition?

Yeah, I think that's exactly right. I think 2026 is the year where these technologies are going from being fringe technologies which people are adopting on the side to actually technologies which, you know, entire discovery programs are built around. And so The pharma companies are not necessarily the loudest. You know, they care a lot about their IP. And they, you know, there's obviously some, some alpha in figuring out some, you know, an advantage in getting a drug to market faster than someone else. So I think that will be lagging. And then I think there will also be some time for some of these de novo drugs to get into clinical trials in humans and they'll be validation there still. But what's important is, you know, that adoption curve is well underway now. It's kicking off, you know, it is the most sophisticated companies and, you know, some of the stories we're hearing already are just remarkable.

On the business model side, you're very full stack already. You're building the, you know, the harness, the model, and then the research that goes into the model. At some point, I imagine you'll have a customer that develops a new drug using CHAI, and that drug goes on to generate tens of billions or hundreds of billions of dollars of revenue. To me, that's like what success— Blockbuster. A blockbuster drug created with CHAI. Depending on how you charge for that, it's possible that you end up capturing— $200 a month. Yeah. Just some very minimal amount of value, which is good. Every company, you know, huge positive, want to capture less value than, than they create. Um, but at some point, would you consider, um, or, or have you considered a model where you're a true partner, uh, to these companies and, and you're doing sort of joint, joint ventures where you're basically giving them access to the product in exchange for long-term upside?

Yeah, you know, that traditionally has been the model if you look how biotech deals tend to work, a lot of them includes things like royalties and upside sharing in the drugs. You know, this is really just all partner-driven for us. You know, what does— you know, we live to serve our partners, really. And what do they, they want to do? What they want to do right now is a lot of them want to take our technology and deploy it fairly broadly across the portfolio. And that, that means that, you know, we get We get rewarded financially for that. We also can reinvest some of the money they're paying us into building even better models for them. And so that way of doing business is working well for us right now. There may come a point where some pharma companies turn around and say, hey, we'd love to do a joint venture or go 50/50 on this thing with you. And I think certainly open to that. But really, it's partner-driven for us. And we are trying to— What we see ourselves as, as this core platform layer which enables their discovery engines. And if you look at the top pharma companies, they might spend $5, $10, $15 billion a year in R&D. And so that's not even, you know, on a— what they're spending on royalties or anything. That's just their R&D costs. So it's a massive market to be building into. In fact, you know, it's even larger than some of the money that's spent on, you know, semis R&D every year. Pharma is one of the most enduring and durable categories of applied R&D spend in the world.

Good point. We have some new partners on the financial side. Tell us about the round. I want to hit the gong. How much did you raise?

Yeah, so we raised $400 million.

Who came in? Everybody.

Everybody came in. Yeah, we raised $400 million at a $3.8 billion valuation. We're super lucky to be partnered with Index, Kleiner Perkins, Sequoia, and Dimension, who are the largest checks in the round. And, you know, it's an all-star cast. Some people we've known for a very long time— Nina from Index, Ilya from Kleiner, Pat Grady from Sequoia, and then Zohab Many. But really the whole Dimension team, you know, they're all— yeah, it's rare to get this caliber of investors on the cap table and especially rare in one round. So we feel, we feel incredibly fortunate.

Well, congratulations and thank you so much for coming on the show.

Great to have you on.

This is incredible progress. We'll talk to you soon. Have a good rest of your day. Goodbye. Let me tell you about the New York Stock Exchange. Want to change the world? Raise capital at the New York Stock Exchange. I think Chai Discovery might be there any day now. You're putting up multibillion-dollar rounds. Anything's possible. Our next guests are Evan Burns, the CEO and co-founder of State Affairs, and Jamie Seltzer, co-founder of State Affairs. Very excited to be joined by both of them. How are you guys doing? We haven't seen each other in, what, a year and a half? Hereticon. I think that was the last time we hung out. How you doing? Welcome to the show. Quick introduction yourself and— so great to see you guys.

Good to see you guys.

Happy to be here, guys. Jim, you want to go first?

You got it. Yeah, I'm one of the founders of State Affairs. I'm also a GP at LightShed Ventures. I fashion myself as a poor man DeLions or a poor man Josh Kushner. Doing kind of both. But yeah, founded State Affairs with Evan 4 years ago.

Very cool.

And you're an absolute dog, a very, very humble one at that, and a dear friend. So it's great to see you here at long last. Evan, over to you.

Hard to follow that. Entrepreneur, most recently built and sold Finnish Long Drink. I should have come on here when we had that exit a couple of months ago.

The boys were so ready to crack them open in the studio today.

They were. They were. They wanted us to be here.

They wanted you guys to join and we go, yeah, they were like, this is huge. Anyway, sorry, State Affairs. Yeah.

And so also the co-founder and CEO of State Affairs.

Yeah. And, uh, just set the table on like the shape of State Affairs business, why it's unique, how like the footprint, the strategy, because it's fascinating. I think, uh, underdiscussed.

Yeah. Jamie, why don't you tell me about how you didn't even mention or tell me to invest until about a couple years had passed and you guys were at tens of millions of revenue. Now I'm just kidding. Evan, go for it.

I think Jimmy was worried about that when we were coming on here. He was like, I don't know, we might be, uh, might be asking for this.

Um, we had to do something different than everybody else in finance.

So of course we picked this, but, um, but the, the problem that we saw and that we're focused on every day is, and it's probably true for most of your guests, these policy and regulatory markets affects so much of what companies do, but everybody just acts like they're a thing that have to happen to us. We're all reactive to them. So Jimmy and I had this idea, like, what if we could build a platform kind of like Bloomberg? Instead of focusing on financial markets, we're just focused on building a knowledge graph around the day-to-day movement federally in all states and eventually cities around these policies that change around regulations. And in the day, we think they affect most businesses more than what the financial markets do. But if you're a leader, the business, how do you pay attention to them? How do you hear what the discussions are about SNAP benefits access or about AI data center building or about public safety tracking? Are we going to have tracking or not? Like, these things are massive. And I think most people in America have just decided, oh, they are— these things are what they are. So we said, let's try to build something around that.

And so how much does this look like a media business? How much does this look like newsrooms across the country roll-ups? Are you hiring people? Are you acquiring whole businesses? Walk through how you actually get a broad footprint out there to track all those topics you mentioned.

Yeah, I think, I think what we didn't realize getting into this was if you look at the legislative process of making policy, there's so many pieces to it and so much data to it. So you have all the hearings that happen in every state. You have all the bills that are introduced. And every bill, if it even makes it, is going to have hundreds, if not thousands, of data points that change along the way, whether it's the wording or the votes or the committees or the sponsors. And of course, we did actually build out journalists, a team of journalism. And it's very important that they're nonpartisan going and doing coverage what's happening in the state houses also, asking questions around this policy and regulation. So it's that whole mix together that becomes this knowledge graph that enables the users to plug in and say, okay, what's happening today with SNAP access benefits? What's going to happen next year around GLP-1 access with Medicaid? Like these things are changing kind of by the day, but you might not see it in a law for another 6 months, a year, 2 years even.

And so for the pre-AI, the last few years, the experience of the customer was a subscription and then sort of like a fire hose of facts that were not previously reported about things that are happening in areas that they care about or all over the nation. How has that changing in the AI era? What are customers asking for and then what are you building?

Well, the, the first thing that we're able to do is map the relationships of all of these different datasets using AI. And so sometimes something will happen in one state and it can affect, you know, states that are tangentially down. And so without basically having built out, you know, through journalism, we have on-the-ground people that input data that sometimes doesn't make it to the government websites. Budgets, bills, hearings, fiscal notes, executive orders. We're mapping the relationships of all of these things through our models and we're finding net new insights for our customers.

Yeah. Talk about how legislation and ideas flow between different, different, you know, state and local governments. We were covering, you know, New York's data center ban or moratorium earlier today. We were talking also about how Paramount and the Ellisons are up against a number of different AGs. And so it just feels like you guys could get to a point where you can almost forecast the way that policy is kind of shifting and moving around the country.

We do do that to a certain extent right now. I don't say we do predictive, but we do model the momentum of how bills are likely to get picked up or that the likelihood of a certain policy picking up more and more steam in a particular state. So, you know, we basically have our models have 150 kind of unique data points that we source from on a daily basis that are kind of constantly resetting the momentum on these.

Historically, big companies, let's say Fortune 500s, like how many, how many, how many individual— what would the team teams look like, you know, the government affairs teams look like to try to get the level of coverage that you guys are now offering? I imagine they're buying data, working with maybe more local companies, but the pitch now is like, you know, one platform, access to intelligence across tons of different, you know, counties and states and things like that. But how are you pitching it?

I think the main thing you have to when we talk to all these folks and we've sat down with a ton of Fortune 100 brands and have, you know, brands like Mastercard and McDonald's and Walmart that are already subscribers, but like the problem of what everybody is facing before is they've got like lobbyists or like old school data platforms that are often like much slower than real-time information. So you're waiting for a firm to send you a PDF deck from like something that happened 2 weeks ago and you've got 50 different states with 3 different formats.. And so the ability to plug in and understand real-time movement on what's happening with discussion around, you know, data center licensing or power buildouts, uh, it's very tough and it's late. So we want to switch people from being kind of late to the game to understand what's even happening to proactively involving themselves in the democratic process and ideally trying to get their voice heard, involved before, before the bills are even introduced, let alone waiting for them to get all the way to the governor's desk, which happens a lot. Like, we were talking to— the CEO of a publicly traded company that I think the— I think the CEO's been on here. I won't say his name, but their company business model was basically banned in a top 15 state. Got to the governor's desk, and this CEO didn't even know about it until it got to the governor's desk until their business model would've been banned. And then they had to pull all this very painful, very painful strings and favors to kind of get it fixed at the very last second. But if they'd just been involved a year before, they'd have seen it coming.

How— what does your guys' team look like? I imagine it's extremely distributed.

Well, we have about 187 employees. It's about half editorial, half tech and sales. Um, the editorial newsrooms are on the ground in each state, so yes, they are distributed. Like, we do believe that being in the room is the difference maker on a lot of this Intelligence from the editorial side and then engineering. We're in D.C., San Francisco, and Miami.

Are you seeing acceleration or value in sort of like Vibe Coding integrations to government websites to scrape and organize data?

I think it's very difficult to do that. Really?

It's the final— that'll be AGI. We need super intelligence to get filings into a database.

However unsophisticated federal government websites are, state government websites are that much harder to navigate. I think like 2 of them have APIs. Okay. Just to kind of level set on what we're dealing with. And, you know, a lot of this, the syntax on how they label their data is very different. And so there's like a huge cleaning process in Ohio, for example, if a bill has an amendment to it, sometimes it never goes onto the government website. You have to go into an office in the state, go and get it. There's this example we love to give in North Carolina that if you don't watch a Senate floor committee hearing, you have the only way to go and get a recording of that is to go into the Congressional Library, pay $1.25. Get it burned onto a CD-ROM, and then you get custody of it. So like, yeah, it's, it's, it's, uh, pretty antiquated. Yeah.

And you have information every day, right? So we, we've made the joke before, but like, we all think about the Roman Empire a lot, of course, but you only have to label the Roman Empire once. Like, it's done, it's over. And like, you gotta get different, the information and data around what's happening in government legislation every day. Yeah. Because what the process was last week has changed and whatever today, uh, happening today is the most What do your customers want you to do next in terms of coverage?

Is there demand for international coverage or even more local, like, you know, like HOA or something? I don't know how deep you could go. Like, how low, how low level can you go? And then how high—

the mayor in this town of 5,000 people sneezes. I want to know about it.

I want to know.

For the right price, we can definitely figure that out. You know, I don't know about HOAs, though. I don't think sanity, sanity quotient match. But look, if you, if you look at these major— we have international customers, um, Fortune 50 brands, you name it. The thing that they're trying to figure out again is this real-time movement on things that have billion-dollar or tens of millions of dollars of effect in their business. Yeah. So for sure, like we do federal now, we do state, we'll get into city. Um, there's a huge ask for international because if you've already got your executive team and government affairs leadership on us and you've got US, Why not Canada? Why not Australia? Why not UK? So the goal to do this, this daily, build this kind of daily knowledge graph and platform for sort of all Western democracies. But we're definitely just started here in the US.

That's very exciting. Well, you raised some more money. Tell us about the round. How much did you raise?

$70 million. A bunch of nobodies, right?

Who, who participated?

How many, how many times a day do you get to do that, by the way? Uh, Today's like 5 or 6. It's great. Yeah. You gotta warm it up too.

Sometimes we'll hit it twice. Yeah. I'll hit it for you guys too. Who participated in the round?

Who participated? It's been primarily Coastline and Founders Fund. Big shout out to Keith and Keith.

Those guys, they needed a warm up. It's good to see them.

Never heard of them. Never heard of them. Never heard of them.

Glad you guys are putting them on the map.

Yeah. Yeah. It's good. Thank you for what you're doing for them.

So great to have you guys on and incredible progress. I'm glad you're talking more about what you're doing because I've been getting the updates for, for years now. And yeah, I know you guys would only start talking about what you're doing once you were confident you were going to eat the entire market. So great. Congrats.

We'll talk to you soon. Have a good rest of your day. Goodbye. Cheers. Let me tell you about Console.com. Console builds AI agents that automate 70% of IT IT, HR, and finance support, giving employees instant resolution for access requests and password resets. Our next guest is Tyler Page, here with us in the TVP Newsroom live. Thank you so much for taking the time to come on down and chat with us. Of course. Thanks for having me. Welcome to the show. Please, why don't you kick it off with a little introduction on yourself for everyone who's watching?

Sure. I mean, look, longtime listener, first-time guest. So thanks for having me. Excited to be in studio for that. Yeah, look, I'm the founder and CEO of Cipher Digital. Yeah, we're a developer of AI data centers, a popular trend these days. Over the last year, we've really completely transformed our business. Once upon a time, we were a Bitcoin miner. Yeah. Also a data center building business, actually a lot closer to building AI data centers than I think is in the popular imagination. Mm-hmm. But we've really had just a complete transformation over the last 12 months. We've signed a series of leases with hyperscalers or with companies affiliated and backed by hyperscalers. We're currently building 700 megawatts of, uh, AI data centers at 3 sites in West Texas. There we go. Love that. That's huge. That was, uh, Bonamask. That was awesome. Um, yeah, so look, we're really busy with that, and then we have a massive development development portfolio beyond that. So we've at this point sort of proven that we can sign these leases, we can construct these data centers, we can finance them. Also a really big part of it. And then we have a really big development portfolio. So the total portfolio at Cypher is 4.2 gigawatts. Wow. Wow. And he's going back.

It's even louder.

It's even louder in person. Yeah, that's just It's just really awesome. Yeah. So, I mean, look, we're on the path to sort of rinse and repeat on that business. Demand remains really high. Of course, everyone in non-podcast media will point out everything's a bubble or there's challenges or whatever. And, you know, we can run through those things. But, you know, right now, listen, the demand remains really high. Sure. I don't see any signs of that pulling back. If you've got power available in the next couple of years. Yeah. Uh, and it's a really exciting business. I mean, honestly, with all the things going on in the space, Cypher lives in a little bit of, um, you know, it's got a unique structure that we're really doing colocation. So building like a full turnkey AI data center for a multi-decade lease and handing over the keys. And on a spreadsheet, truthfully, it's, it's a pretty simple business. If you can execute in the real world, it's real estate development. It's a real estate deal. It really is. And, and, And there are so many interesting things going on in the space, and there's lots of people trying to be more vertical up the stack and build a software layer, and that's all super cool, and there are big prizes to win there, probably. But I love our simple rinse and repeat business.

You're in the warm shell business.

How are you thinking about site selection across the development portfolio and new opportunities? We had the New York State data center moratorium get announced this morning. I would expect other states to follow suit. You guys don't want to invest, you know, tons of time and resources into a state that could eventually say like, hey, you just got to stop what you're doing.

Yeah, I mean, listen, so I'm actually based— our corporate headquarters is in Midtown Manhattan. Oh no way. None of our data centers are anywhere in the state of New York. I think that's a little bit of a hangover from our days starting as a Bitcoin miner when not only did people not want to give you the power in New York, but they also decided they didn't like Bitcoin.

Oh, so, uh, and I think the biggest data center in New York right now is still a Bitcoin miner, or they were, and they've been pivoting. I think it's pivoting over.

I think it's one of our good friends and rivals. But, um, you know, look, I think there are going to be layers of challenges from the federal level to the state level, to the local level, anywhere in the country. Of our 10 sites, 9 are in Texas and 1 is in Ohio. Texas is generally known as a very kind of low-regulation, business-friendly state, very entrepreneur-friendly. That said, the exact same concerns about data centers exist there. Sure. And we spend a lot of our time trying to be good neighbors and good partners and deliver the message of how it's going to be a wonderful addition to the community when we come join. But quite understandably, people show up. I mean, it is outside of the universe of the savvy listeners to your podcast. Yeah, there's a lot of like naked dislike of data centers.

You see it in the polling data too.

But, you know, it's interesting, even as recently as yesterday, we had folks down at a local hearing and I completely understand why people show up and they're not happy. Because they're like, I've heard this stuff's terrible. Totally. And it's going to end the world. And, you know, you know, it's going to end things that are creative and everything's going to be an automated chatbot. And like, I don't like anything about that. And then you get all the way to the spectrum of, frankly, just people don't know what happens in the building. So you get questions like, is this going to harm my cows that are next door? Or like, is my— I mean, someone asked like, is my pacemaker going to have a problem? Because of the building and it's interesting. It's, it's not for the record.

Yeah, but, you know, well, yeah, there's no, there's no upside, no, no immediate upside to them often. And then there's potential downside like noise and energy prices and things like that. What have you thought about certain proposals or ideas around doing direct payments to people, you know, in towns or counties that are going to have a data center developed and they're thinking, Well, I don't, what's the benefit to me other than some construction jobs and a handful of jobs and local tax revenue, but it's not as like direct as like, you know, cash coming into your account.

So that may be where it goes, Jordi. I mean, I think there's, this needs to be answered on a couple levels. There's like a really, really high up public policy question about do we have the right tax regimes and ownership regimes and large language models language models and government ownership and things like that. And that's what I mean, like, it goes through all the layers down to the state. I think what I'd start with is just number one, back to the big point about this scary new neighbor you might have. I mean, I try to focus very much on starting with what happens in these buildings. And, you know, I'd say, look, we're a pretty entrepreneurial-driven culture at Cipher, but some people, especially a lot of younger people, a little bit more mission-driven.. And I point out, like, you guys are racing on this construction project and that literally might be the building that has the breakthrough in research to cure cancer or dementia or something like that. Or we've got, you know, 1 in every 5 employees at our firm is a veteran. And I point out, like, we may have work that goes on in there that saves our soldiers' lives. So part of it is like just starting there to get to your actual question, though, about like money sharing. I'd point out there's already a fair amount of framework in place that envisions this, not exactly the giant AI buildout that's happening, but like, for example, in Texas, we will pay tens of millions of dollars a year to the Texas public school system. Yeah, yeah. We are the largest taxpayer in the counties typically of our data centers. And then we try to go above and beyond that to do a fair amount of community outreach. So in one of our locations, We're refurbishing the small hospital there. We're buying them a new fire truck. We're rebuilding the playgrounds and the pools, little things like that. And having—

Jordi Hays, Virginia, as an example of a state that benefited from tax revenue as the AWS campus built out there. And a lot of local people that live there have lived through the data center boom and seen, you know, lower taxes on their balance sheets because of that.

Yeah, I mean, listen, affordability is a really big political topic right now. And so people say, well, wow, my electricity bill's a lot higher. Yeah, must be the data centers. Sure. And it's sort of like, well, we printed an awful lot of dollars starting 5 years ago and we've had a couple of wars and there's a lot of inflation. And so it's not really the data center. But I mean, I think, look, we try to be very front-footed on that. We do not want to raise local ratepayers' bills. By the way, all the regulators in Texas feel the same way. So So like that problem will get addressed. I really think it's the big picture thing of like, this is something you should want in your community. Sure. You know, it's a big warehouse that's air conditioned and full of computers. Yeah. It's not, you know what I mean? And where we are building them, and it's really, it's interesting, it's, we've gotten to be a bit of a disruptor to the traditional data centers, like in Northern Virginia. Sure. When we started our pivot, we almost had an unfair advantage. From being a Bitcoin miner, because when you're a Bitcoin miner, the business doesn't work on the spreadsheet unless your power costs are really low. Yeah. And where are they really low? In really remote places with a lot of generation. Yep. And typically not a lot of demand, like not a lot of houses. So if you look at just the map of Texas, most of the people live in the east. There is a ton of generation in the west. You can't ship the electricity that far. And so it's really cheap historically in West Texas. Yeah. When we started having this idea of like, wow, look how fast, you know, ChatGPT is taking off and look at all the implications for how many megawatts you're gonna need at a data center. We said, well, surely they're gonna have to draw the conclusion that everyone's gonna have to move to places like West Texas. And the entire incumbent data center industry was like, no hyperscaler will ever sign a lease at those locations. This is as recent as like a year and a half ago. A year ago, uh, and the floodgates—

because it was a hassle and they'd want to just co— you know, locate around there.

I think they were just using pattern recognition where they're like, I've been in this business 20 years and no one's ever signed a lease out there. And I think— and then there was a second question about like, well, is latency really bad out to West Texas because it's a long way away from the city? But the truth is there's a big fiber line running out there and it's fast enough. Yeah, you might not do traditional cloud work out there. But even Northern Virginia exists because everyone went there 20 years ago and then built out all that infrastructure. Yeah, I can tell you because we're the tip of the spear, like all these guys are coming to West Texas. Our portfolio alone is almost as big as all of Northern Virginia. So, uh, I mean, I, I just, I think this is kind of the next thing that's going to happen when people— the sort of next skeptical thing about locations like ours is Well, we're at this really funny time where there's no power available. And so there's a ton of demand. And so people will go to these places, but that's going to be like a dinosaur relic at the end of this lease in 15 years. And we are 180 degrees opposite of that. I actually think the cap rates will be lower. That will be a more valuable, like, terminal value for a data center at the end of a lease in West Texas because of what's of what West Texas will become, I think, over the next decade or two.

It feels like when new data centers come to town, there's like a suite of questions that local residents have, everything from aesthetics to power rates to water usage to sound to air quality. What of those five have, you know, solid regulatory frameworks in in place where you can go and say, okay, we did the EPA test and we passed and so we're good? And which ones are more subjective?

So it, it really depends, which is an unsatisfactory answer. But yeah, I think we always start on the power side. Okay. And so if you look at ERCOT and the way they interconnect large sites, that's generally a low regulation zip code state. And so what I'd say is you need to address all those issues. There are some local permitting requirements around them, but I'd say it's more about the, like, what issues could be raised. There's no real framework of we'll just hold up the noise permit. Sure. Because there's a lot of oil and gas and like industrial things in places we're building. But I would say probably in our experience, Of course, people care a lot about affordability, so we need to discuss that. But I think water at our locations tends to be the biggest issue. Sure. You know, they say, I think, whiskey is for drinking and water's for fighting in some of the places out there. Just it's dry. But what's interesting is, and I know you guys have talked data centers a lot on here before, but, you know, with modern closed-loop systems, it's much less— really don't use that much water. And I think the other thing is you can innovate. So there's actually a lot of like brackish, non-potable water underneath the dirt. Sure. Texas. And so you can build facilities to filter it. Sure. And use it and actually help the aquifer. So, you know, I generally think all these kind of growth issues are challenges, but they're all solvable.

Yeah, it feels like a meet-in-the-middle here is instead of jumping straight to like data center ban, you're focused more on what is the decibel level that is acceptable in this city for any building. Right. What is the power consumption for any building? What is the impact on the water supply for any building? Right. And then that framework. So then if somebody shows up and they're just like, yeah, I have a house where I'm gonna be blaring music, it's like, yeah, that can't happen in this neighborhood either because we've determined this. And then that's applied broadly as opposed to just narrow targeting that.

Well, what I'd also point out is like all of those same objections could be raised about almost any industrial build. Yeah, for sure. So of course, I mean, if you go to the places where we build our data centers, we own hundreds of acres and in some places you can see to the horizon in all directions. Sure, sure. So I'm really not too worried about, you know, we'll be cognizant of measuring the decibel level at the edge of our property, but like, yeah, I'm not worried about anyone hearing it. Sure. Um, I, you know, I think it's, it's going to vary by jurisdiction. If you try to plop one down in a, in a neighborhood or something, that's not what we're trying to do.

It seems like a lot of them have sort of like grown over time where there was, it was, it made a lot of sense to co-locate some data center for delivering Netflix to a local neighborhood, and then the, the, the place just got upgraded and upgraded and upgraded until it's like running diesel generators 24/7, and then people are annoyed. Yeah, so I mean, and I understand creep up on you is more of a problem than like the new project coming in with a dedicated—

I think this is like any industrial build though, where you've just got to balance, you've got to strike the right balance. It gets back to what you said, like maybe the end policy solution is a different way of taxing, regulating, sharing the income, whatever. And we will certainly be—

Yeah, it's not necessarily, it's not even necessarily like more dollars that are being, shared, but it's like how you share the dollars. Because again, like money going to someone's, you know, city feels a lot different than money going directly into someone's pocket, right? That they can use to pay their bills.

So walk me through your capital market strategy when you're working on a new site, new project. What does your roadshow look like? We're familiar with like a Series A pitch, Series B pitch, even an IPO occasionally. But you're talking to private credit. Are you talking to real estate investors, banks, debt? Walk me through the whole picture of like who you're calling, who, who you're pitching, who's around the table, just in broad strokes. Sure.

So, I mean, high level, we're— the sites we're developing, and, and again, this is some of the most lucrative real estate development in the history of real estate development. We have purchased these sites typically really cheap. Yeah. Like couple million bucks. For 100 acres? For the land is almost secondary. It's getting the interconnect for power. So, okay, the price has gone up like per megawatt. But for example, we're building a 300-megawatt site right now for AWS. Yep. And we paid like $7 million for the site 3 or 4 years ago. Wow. I mean, it's been a while. Today that would be hundreds of millions of dollars to set sell it. And we own lots of acreage around that. Um, so, uh, the— back to your question about the model. So typically we will try to find a good deal, harder to find these days, but we have a pretty big portfolio. Like I mentioned, it's closing in on the size of Northern Virginia, so it's pretty big to still be developed. Um, next phase is we want to shorten the development timeline as much as we can, and that typically will be an early equity investment. So the longest lead time item to build one of these typically is the, the like high voltage to mid voltage conversion infrastructure, the substation. Okay. Okay. And that might be like 18 months to get one and call it $100,000 a megawatt. So 300 megawatts, $30 million. Okay. We'll get that done and we'll do some early civil work to get the site like the timeline to build shortened. Okay. That's currently at least, you know, who knows if we evolve over time. But currently, at least, that's where we stop generally our equity investment until we find a tenant that's interested. Mm-hmm. That's very hard. It takes forever. You negotiate for months and months, hundreds of pages of documents. But if you get, let's say, a hyperscaler long-term lease, at that point, they will be very opinionated about the particulars that they want for that site. Sure. That's a very iterative process with our engineering. Team, you figure out exactly this is the style they want, ticks all these boxes. At that point, if we have a signed lease, we will try to debt finance as much as possible.

Um, you know, are you buying the actual GPUs or is the hyperscaler— warm— saying like, we're just going to—

to this point, that business has been less interesting to us. Yeah, it's not very interesting. But the— yeah, it's like you don't want to— yeah, the piece we see the certainly risk-adjusted return, and for the last year or two, frankly, just the best absolute return, uh, is building everything down to the GPU.

Yeah, basically.

So, so yeah, walk me through exactly how far you're taking it. So the building goes up. Yep. The polished floors are in.

Yep. We're bringing— yeah, we're bringing cooling, everything down to the rack.

Okay. They're gonna— but then they bring in the racks. Got it. Okay.

Yeah. Connect the various connections. Okay.

And then water lines. Yep. Power lines.

All of that. Chillers, everything down to the rack. Yeah, it's a very like everything ready to go. And the reason why, by the way, on a long-term lease, right, figure they're going to recycle those racks. Yeah, probably twice. Sure. Yeah. So like a lot of the modern design and iteration is how do we build this site to make it as easy as possible to not be outdated when we want to change the racks? And the racks are going to be like, there's going to be way more rack density. I don't know, they might replace them in 5, 6 years with 1-megawatt racks. You know, you might get 120 kilowatts or something now and it's going to be way more dense. And so, you know, high level, you have to bring basically, you know, electricity and water and air into the building. Yep. You have a very big open building, kind of like a lot of these types, like an Altered House.

Yeah.

Fantastic. And then the ability to dial down those various valves and so forth as you get towards the rack. Yeah. And probably the floor space shrinks over time and you have cool ping pong. You open the podcast studio in the other end of the— at the other end of the data center in a few years.

So just finish this.

Sorry, just because I didn't finish it. Like, so then we'll go debt finance it. We have gone to the, the high yield markets. Yep. We've raised a couple billion dollars at about 6% and change. So amazing. That's really— yeah. Wow. LTC. So yeah, the market's wide open. That's great.

That does your team look like a private equity fund with a bunch of people sitting at desks in front of spreadsheets and then you call people to actually go and build the building and they're off your staff and you're hiring construction firms? Or do you have sort of, you know, guys who can work a shelf I think it's a really weird company.

Like I tell people, so I started the company in my attic during COVID So we have had to hire, like every person knew how to do something better than me at some point. And then that's cascaded down. Yeah, yeah. We only have about 100 employees at this point. It's a very unique firm in that it's, it's a little bit like a Venn diagram of, yes, a finance sort of structuring spreadsheet person. Project management. Yeah. That type of stuff. Let's call it a technology firm to figure out a lot of what needs to be done, like networking and stuff like that. It's not exactly the industrial side of construction. Yeah. And then the last part really, and like the most populated part, is the more industrial piece. We do in-house project management and most of the procurement activities. We externalize the like trade work. Sure. So if there's—

you might pick the transformer to buy, schedule it for shipment, but you're calling a trucking company to bring it out. You're calling other people to unpack it off the truck, plant it, bolt it to the ground, do whatever else they need to do.

We have an external EPC that really brings all those trades to the sites. We have 1,800 people.

What does EPC stand for?

EPC. Oh, EPC. Yeah, sorry.

Engineering Procurement Construction. Got it. Right.

So sorry, jargony on you. Yeah. So, so if there's 1,800 people on the site, maybe 5 of them are Cipher employees. Yeah. The others are like the staffed up folks from the outsourced provider of a lot of that trade work.

Do you like railroad analogies?

A lot. Actually, it's the best analogy. More than oil? Yes. And I'll tell you why. Because it's this— the buildout of this industry is so tied to capital markets. Markets. It's one of the reasons why we have such a New York nexus. Sure. Like, honestly, even in Bitcoin mining, that was a big part. We went public as a way to raise a bunch of money in a business that's hard to like get the CapEx for. All the hyperscalers are now doing the like varsity version of that with big debt offerings and so forth. But I like the railroad analogy because the modern capital markets were kind of created because of the railroads. Like there wasn't enough bank lending in in the world. Yeah, to get them built. So the idea of like buying equity shares in a project is kind of the railroads. So that's why I like that one. None of them are exactly perfect, but no, that's closer than any of the other ones I could think of.

Yeah, Jordi, do you have something else?

Sorry, uh, we like the railroad analogy minus the, you know, massive bus.

So true. I actually haven't studied the railroad boom and bust as much as we should study. I mean, it was a very long, long period of history. Yeah, uh, fascinating. Well, thank you so much for coming.

Yeah, great to have you here in person. Really interesting.

Great to see you. I'll see you again. Going a level deeper like this, this is fascinating. Uh, yeah, please send anyone to us, uh, send anyone on your team our way when you have news. Uh, and congratulations on everything that's going on. We will.

Thank you so much. Great to have you on.

Have a great day. Congrats on the podcast. Thank you. Hang out, we're gonna close out. Um, let me, uh, on—

before we jump, I gotta talk about, uh, my dear friend Brandon Jacoby, who I saw in the chat earlier, launched his, uh, new studio, a multidisciplinary design practice for those who challenge the boundaries of technology. He combined a Star Star Wars intro style video with a barrel, a wave. A barrel. Oh, cool. Okay, I'm visualizing that. I think he made this for us. Okay, you want to pull it up? There we go. There we go. Look at this.

Wait, motion design. Oh, interesting. Yeah, this is both of us. Our intro. Yeah, this is perfect.

He made the launch video For an audience of two.

For some reason, I was imagining the text curling up like a wave and it being sort of hard to read, but this is much better. I love it. It's a good statement. This is a mission statement.

This is an essay.

Worked together for a few years. And he's been doing this—

He was one of the first personnel news we did on the show. We tracked his move to X, the everything app.

But anyways, he's been doing this kind of work forever. He was a design lead at X as well as Cash App, as well as my last company. And he's incredibly talented. So he's open for business. Fantastic. Well, that's our show, folks.

Let me tell you about Public. Investing for those who take it seriously. They got stocks, options, bonds, crypto, treasuries, and more with great customer service. And that is our show. We'll see you tomorrow at 11 AM sharp. Leave us 5 stars on Apple Podcasts and Spotify. Sign up for our newsletter at tbn.com, and we will see you tomorrow. Goodbye.