How Top Investors See the Future of AI | Chase Coleman, Brad Gerstner, Josh Wolfe at Robin Hood 2025

Here to moderate the session, please welcome founder of Tiger Global, Chase Coleman.

Thank you. It's a pleasure to be here with all of you today supporting Robinhood and the incredible work they do to combat poverty in New York City. I'm looking forward to a good discussion between three incredibly thoughtful investors on a topic we all care about— AI. So just to frame it, AI CapEx is contributing approximately 1% to US GDP growth, which is half of US GDP growth. ChatGPT is the fastest growing consumer application of all time and is growing 3x the rate of Meta, Google, and AWS when they were at a similar scale. Anthropic might add, or added more than half of the net new ARR of the, of the, of the SaaS industry in the first half of this year. We're seeing multi-tens and hundreds of billions of dollars infrastructure investments on almost a weekly basis. And we're sort of at the outset now of reasoning models, video models with Sora and Nano Banana and Vio3 from Google and stuff like that, and agents, which should drive pretty strong token growth for a while. So as we like to say in this industry, it's on. So excited to hear from these guys on their, on their thoughts. So Brad, let's start with you. I just recapped what everyone's thinking about now. This is Q4 2025, but help us extend the time horizon. Tell us a little bit about applications and infrastructure and what we might expect over the next several years. Right.

So first, it's great to be here. An incredible cause. And as I said to Andrew Ross Sorkin, who was hyperventilating about bubbles this morning on CNBC, rarely do you have bubbles when the when the hit book of the day is about the stock market crash of '29 and when everybody's talking about bubbles and when NVIDIA is trading at 27 times earnings. So if you extend the runway, I had Jensen Huang on, on my podcast last week and he said, you know, we're going to have $3 to $4 trillion of compute buildout over the course of the next 4 to 5 years. On the one hand, you have to hold two truths simultaneously. On the one hand, to contextualize that, that's about 10x the Manhattan Project. So the Manhattan Project was $4 billion, inflation-adjusted $40 billion, GDP-adjusted about $400 billion. We're going to do $4 trillion, all funded privately. So this is a massive tailwind for the economy, as you just suggested. Why are we having that much buildout? It's not just because of this generative AI stuff. Right, we're building this out because all businesses in America are moving from general purpose compute to accelerated compute. You would have no TikTok, you would have no Reels, you would have no Google Answers immediately on your phone, but for the build out toward, you know, TPUs or NVIDIA's GPU stack. So when I look at the $3 to $4 trillion, I asked Jensen, what is the chance we're gonna have a glut in the next 5 years? He said, I don't know about the next 5 years, but no chance in the next 2 to 3 years. Because in the next 2 to 3 years, it's just the hyperscalers that are paying all this money. When you hear all of these headlines that have been talked about, I mean, Sam Altman must have announced 10 deals in the last 2 weeks. Remember, these are frameworks. He's laying out a framework of what could get built over the next 5 to 10 years if all the demand shows up, if the chips are delivered, etc. Nobody is in handcuffs and has to go over the, over the falls here. So I think that, you know, there's a lot of excitement. I would not underestimate this. You and I got involved in the internet at the start in 2001, 2002. Everybody said it's a fad, right? We had just come down from the crash. The reality is we hadn't even gotten started. I think we're in, you know, inning 1 or 2 of a decade-plus-long build-out that will transform compute, will transform all of our lives, transform our businesses. But certainly just this year, we had 2 25% drawdowns in NVIDIA. There's gonna continue to be a lot of volatility.

Andrew, you've talked about, uh, the AI layer cake. Um, tell us about that analogy you use and, and what layer of the cake should we be devouring?

Yeah, it's a, a great, uh, concept. So the, the AI layer cake, uh, people have different views of their own layer cake and different recipes, but this is how we think about it. So bottom layer, chips, so semiconductors, the middle layer, the large language models. So OpenAI, Anthropic, top layer applications and tools. And I think when we think about prospectively, like where value is going to get captured, we think that semiconductor layer is going to capture a very, very significant amount of economic rent over this next 10 years, which is, I think, quite different than the playbook that worked from the global financial crisis through, call it, 2021, where you could invest in software, you had two massive tailwinds with the shift to the cloud, the rise of the SaaS business model, which provided, you know, tremendous growth for that industry. And I think there's just a fundamental shift here in terms of the cost of doing business for those companies where delivering their solutions is just going to require significantly more compute. And so that's why we're excited about that, that chip layer.

And Josh Wolfe, I've read some of your letters. They're very thoughtful. You talk about a 5-year psychological bias. Will you explain what that means to the audience and then apply it to AI for all of us?

Yeah, first of all, 800-pound gorilla in the room, which is that Brad is the most well-dressed guy on stage. And he is normally like me in all black t-shirts. And so he's broken ranks. There is now a contraindicator, I'm convinced, which is, you know, he went from black shirt to dapper suit. If he's wearing a tux next year, we have hit the top. Uh, 5-year psychological bias is, uh, everybody wants to be invested today where they should have been 5 years ago. Our job as venture capitalists, anybody that's got exposure on the private equity side, uh, doing private deals, uh, you're trying to anticipate, uh, where do people want to be in 3, 4, 5 years hence. And, uh, you know, that's inherent speculation. You're betting on people, you're betting on technology. I'll give you some contra views on AI and where we're putting money today. I was very lucky 10 or 11 years ago, 2015, to be an investor in a company called Zoox. It was a self-driving autonomous company. We sold that to Amazon. It's the basis for all their self-driving. And there were two people inside of a room after we had put about $25 million in their seed round that looked like they were playing video games. I said, what are these guys doing? They're messing around. They said, no, no, you don't understand. The cars outside, vibrational sensors, thermal sensors, optics, LiDAR, radar, they're all processing it at the rate of 1 second per second, what we call reality. Inside, these guys are doing thousands of simulations on these chips that aren't even on the market yet. Oh hey, who they from? NVIDIA. Pitched that at a public charity event, 2016, $15 billion market cap at the time. Intel was $150 billion. I was like, this is the pair trade of a century. Today, $4.5 trillion. I feel the same way today about the memory players. And the logic here is that the consensus view is that you're gonna see a massive continuous scaling law push for data centers and hundreds of thousands of H100 clusters, all to the benefit of the chip players and NVIDIA and AMD and others, the data centers, the hyperscalers. And we'll talk a little bit about that on why some people I think are pulling back. But I'm convinced that 50% of your inference, which Jensen would like you to believe 100% of your inference is gonna be done in these data centers. I think you need it for training. I think the large-scale, large language model players today, I'll also talk about, are overspending. But I think that 50% of your inference will be on device. And there was a paper about a year and change ago that said you could do large language models on device using flash, NAND, and other kinds of memories. The winners here are SK Hynix I think they're probably the most nimble on the Korean side. They also bought Intel's NAND flash business for a few billion dollars, have 60% of the market share, benefiting today from attach rates on high-bandwidth memory for GPUs. Next is Samsung, which is probably the Intel of the space, a little bit more bureaucratic and slow-moving. And then you've got Micron in the US, which, you know, will probably face restrictions just in US export and geopolitical issues. But I do believe that the memory players have that same dynamic that 11 years ago people were like, oh, GPUs are just tightly coupled to the gaming. PS5 console wars, Xbox, et cetera. And these are commodity memory chips, but I think they're gonna be ascendant. And I think it's gonna crack the narrative that we need to invest these, you know, tens of billions of dollars into data centers. All the data center spend, it is true that it is a framework, but at the same time, most of them are quoted in gigawatts. It's an obfuscation of the actual price. $40 billion for a 1-gigawatt data center, $400 billion for a 10-gigawatt data center. If you assume 50% software margins, that implies $80 billion of revenue you need on a 1 gigawatt data center and $800 billion on a large scale. So the history of most CapEx in tech benefits all of us. And I agree 100% with everything that Brad has said. Yeah, we are going to get massive productivity booms. Industries are going to be disrupted. But it also doesn't mean that you couldn't have mass destruction because of CapEx in spending here. Last thing I'll say, the way that we apply this in the 5-year psychological bias, anything in 2-dimensional AI is done. Meaning the, uh, entrance and the amount of money and the amount of capital and the runaway success that people have had, uh, for voice, video, text, code, et cetera. We all have exposure there across companies from Applied Intuition, Cognition, Runway, et cetera, and OpenAI and others. Um, three-dimensional AI is the next wave, and it's where people have been investing for the past two years. Biology and robotics, in particular, those two domains don't have a density of, uh, repository of data to train upon. So it's scarce, and therefore it's valuable, and the people that are going into robotics and biology I think are gonna be winners. Last thing, when I do searches, I do, you know, 5 prompts across 5 different browsers or on my phone in 5 different apps. And I've got ChatGPT and I've got Anthropic and I've got Gemini and I've got Perplexity and I've got Grok and I test them all. And you know, sometimes they converge on a common truth. Each one of those guys is trying to get 100% of my market share of use. They roughly have 20%, but they're all spending 500% on CapEx. So it's just something to sort of watch in this space.

Okay, so, so the investors in the room want actionable insights. We're going to do a lightning round here. I'm going to list companies and we'll talk about percentage value appreciation from today. You all get to pick one. So OpenAI or Anthropic?

I'll take OpenAI.

Yeah, same.

We're making a choice between—

Yeah, you have to make a choice.

Well, OpenAI, $500 billion. I think, you know, they'll go public somewhere, probably $2 trillion, you know, maybe 2 to 4x from here. The one thing I will say here on OpenAI is they were number one in the App Store. Google was an absolute joke a year and a half ago with Bard. Today, Gemini is arguably the most performative model and ascendant. There are zero switching costs for users on these kinds of things. People are trying to habituate you into this. So, you know, they have captured the consumer. Anthropic on the enterprise. I think Anthropic ends up probably getting acquired by Amazon.

Google or Meta.

I think Meta from here. The challenge for Google is they are doing excellent in AI, but they have a profit pool that represents over 100% of the profit of the business that's fundamentally at odds with AI. It's 10 blue links in advertising. None of our kids use that anymore. They all want answers. And so I think while they'll gain a lot of revenue on the search side, by the way, they're going to beat the numbers for the next couple of quarters. Don't run out and And, you know, I own Google, but, but if you ask me fundamentally over the course of the next 5 to 7 years, I think it's a challenge for Google. And I think Meta is making the level, the type of founder-led investments that I would want to see make in terms of the people that they're hiring. They want to build the world's best foundation model. I think it'll be a closed model. It's already dramatically benefiting the organization. I think they're just a little bit more risk-seeking, which I like.

Yeah, I think I would echo a lot of those comments. I think there is risk to kind of the golden goose of search. And what does that look like over the next 5 years? You know, over the next 3 quarters, probably okay. But I think over the arc of time, there probably is more challenges to that business from AI. And so I probably would copy you again. So Meta.

I think Google wins in AI. That's sort of my dark horse. But I think that Meta is going to win on device stuff, which people are underestimating. And the reason I believe that is we had a company called Control Labs that we sold to Meta for a little under $1 billion. That was basically a wristborne device that innovated the 15,000 nerves that control your 14 muscles in your hand, and you could get rid of all your control surfaces, keyboards and knobs and remotes and all that kind of stuff, and just gesture in free space. The same way that if I would've told you 10 years ago you'd be on a glass panel, you know, swiping on an iPad, or getting rid of your CrackBerries and typing on a screen, you would've been like, that's crazy. And they did that because Zuck doesn't wanna be remembered for reinventing advertising or destroying democracy, but for basically reinventing hardware and human-computer interaction, and needs to route around Apple iOS and Google Android to be able to commercialize and monetize you and your eyeballs. So I think that they're gonna win on glasses, devices, wrist-worn devices, and this phenomenon that I call life-cording, where every part of your life is gonna be recorded throughout the day, 24/7. Younger people will be more comfortable with it, older people and our general counsels will be mortified. But the idea that you will basically be capturing 24/7 passive sound audio data, and then visual data, and all the AI insights unlock from that I think are going to be a big deal. And I think Meta wins that.

Okay, for the lightning round, we're going to go, we're going to go one sentence.

Yeah. If I speak really fast.

OpenAI, Anthropic, Google, Meta.

I'd go OpenAI and I'd go second one. I like them both, honestly. Yeah. NVIDIA or Broadcom?

NVIDIA.

NVIDIA.

Broadcom, TPUs, Google.

Interesting. Microsoft, Oracle, or Amazon?

Microsoft.

Microsoft, yeah. We gotta change the order up a little bit. Microsoft, yeah.

I'm going Amazon.

Interesting. Tesla or Waymo?

West Coast, East Coast, yeah.

Tesla or Waymo in self-driving?

I would say Tesla.

Tesla.

Hey, how do you, all right, fine.

Jinx. Waymo.

2 and 2. Okay, so those are my selections. I'll let you go. So favorite AI beneficiary, you can do public or private that you'd share with the audience. And if you're at a, if you're at a block, just think like a big position you have that you want to hype.

You want me to go first?

You go first.

Okay. I'll talk about a public company, even though my focus is now more on the private side. It dovetails. I'm so much copying everything I'm hearing right and left side here. But SK Hynix. On the memory front, like, I guess here's our view on memory. When you look at each generation of NVIDIA chip or Google TPU, memory is like the one piece of the bill of materials that continues to grow each generation. And so they are capturing more and more value. And you have a dynamic where you have a leader at SK Hynix, which dominates HBM, which is the critical memory that's used for AI models. And there's a fundamental supply limitation where you just are not able to bring capacity online as quickly as you were able to do historically. Which is going to kind of keep a lid on supply over, I think, frankly, the long term, which is going to be much different than what we've seen historically in the memory cycle where it's been incredibly volatile. Right. And the joke is always like in a Micron group meeting, there's no one with gray hair because like all those folks have gotten so obliterated investing in memory over the decades that they avoid it. But I do think, you know, I'm hesitant to say that it's different this time, but it does feel quite different. And I think the last piece of this is that, you know, book value tends to be kind of a floor for these stocks. So they'll trade at kind of 1x book value in a down cycle. And you are creating book value so quickly right now. SK Hynix will probably grow book value 50% in 2026. And so you have kind of your floor is actually kind of going up on the stock. So your downside is actually going down as we move into the future, if that makes sense.

So, well said. Well said. You know, one of the things I would say is when I come to events like this, everybody wants to look for the idea that they've never heard of. Um, but if you asked me in 2005 or 2006 what my best internet idea was, I'd say Google. And then my PM would come in and say, yeah, but it's up 50%, what's your best idea? And I'd say buy more Google, right? And what we've seen over the course of the last 10 years, the Mag 7 is up over 10x, the S&P 500 x Mag 7 is up just over 1x, 1.5x, right? We learned in business school that elephants can't dance. Technology companies get big and they get disintermediated. What's happened in the world over the course of the last 15 years is the exact opposite. Scale has led to bigger advantages in the age of AI, where compute and data are massive advantages. I think these scale advantages will continue. So if I had— if you said to me I had to put all my money in two stocks, it would be NVIDIA and OpenAI. You know, today, because I think they have massive scale advantages. I think that ChatGPT is the winner in consumer AI. I think that's going to be the biggest market in, in over the course of the next 5 to 10 years in AI. But there are lots of other things we're investing in, big investors in Hynix as well. But I do not run away from the scale players. There are massive scale advantages.

Long in the big mega-cap would be Google. And the number one reason again is going to be They're gonna Microsoft a lot of the large foundation models. What I mean by that is in the early 2000s, you had the Excel, PowerPoint, Outlook, et cetera monopoly, and then Google came in and basically gave it all for free. I think that they can drop pricing and basically give away for free, and the enterprises that are on Google are gonna, uh, price, and people are gonna say, why am I spending $20 a month or $200 a month for this? Unless you're in the enterprise or in coding. Uh, SK, deeply agree with on the private side. Um, and I think they'll be public within the next year. We're founding investors in an important new defense company called Anduril. Many people may know it. You will if you don't. AI and autonomy is the future of defense in large-scale, attritable systems with a single operator or warfighter being able to control thousands of systems. And I think they're the leader there. I think it'll end up being a $50 billion-plus public IPO on probably $3 billion of top-line revenue and high growth. That's different from the traditional cost-plus margin players. If I could offer a short, it would be the basket.

That's the next question.

Okay, I'll wait.

Yeah.

Yeah.

Okay. TSMC. I'd like to talk about the right and wrong side of change. So You know, we're talking about a lot of optimistic stuff about how AI's gonna change the world, but tell us something that's on the wrong side of change.

I personally would be short all the basket of all the quantum companies that I think are utter, you know, what I call quantum flapdoodle. And, you know, these are things with $10 to $50 million of revenue trading at basically $5 billion market caps. They make absolutely no sense. Quantum has been promised for 10+ years, you know, always 10 years away. And everything that they ever said that you could do, molecular modeling and unbreakable encryption, you can all do with modern compute and GPUs. And the second would be the basket of all the modular nuclear reactor companies. I used to be a nuclear bull. I tried to rebrand it Elemental Energy. I believe that nuclear is the future, but it's large-scale fission, and the entire narrative of endless demand for data centers and modular power, if you are building data centers, you should do it near the Permian or do it out in the Middle East. But I think that all these companies, you know, are basically gonna be disasters on the modular nuclear publicly traded side.

Andrew? I mean, I'll just do a little rebuttal on quantum. I'm not a huge like quantum bull or bear, but I think clearly the retail community has embraced these stocks like collectively as a group. And I try to like step back and try to understand like, well, like why is that the case? And I think if you look at the aggregate market cap of all those quantum companies, it's probably $100 billion something kind of in that zip code. And they do very little revenue. But I think they probably look at, okay, the semi industry today has a market cap. You know, Nvidia is $4.5 trillion, Broadcom is close to $2 trillion, right? So whatever the number is, close to $10 trillion. So you have the semi industry at $10 trillion, you have quantum at $100 billion. You know, I don't think you're going to have a situation where quantum like replaces semiconductors or something like that over the next 10 years. But like, do they get some piece of all this compute that's going to happen over the next, you know, 10 to 20 years? Like, I think that's what people are betting on, whether that's right or wrong. But I think that's like part of the mindset there. And I would say, like, if you look at like an NVIDIA, they have gotten much more serious about quantum over the last 12 months. If you talk to the senior execs there, like they are taking it more seriously. So I think like I would— I understand like when I was running the public tech book at Maverick, like, like I would like, you know, love to dive in short these things and there's no valuation kind of logic to kind of support where these stocks are. But I think there is like a piece of that when you look at like an industry leader like NVIDIA spending more time on quantum, it does make you kind of step back and pause and say, is there, is there something more here that I don't fully—

Yeah, I'm probably in the Josh camp. We don't own any, we're not short it. Um, I think it's really hard to short this meme stock stuff in a market where rates are coming down and risk appetites are moving further and further out on, on the curve. What I would say, think about the last 25 years, the fundamental architecture of all value creation in the internet was Google. Google was the tax collector, the tollkeeper, the organizer. Of the internet. Their fundamental objective was to get you to ask a question, and then they would send you to these sites. They would then charge a, a fee to ch— send you to these sites. And we had trillions of dollars of equity value that was created in the underbelly of Google. Companies like Booking.com, over $150 billion, or Amazon. All these companies were the large advertisers in Google. Now think about the business model for ChatGPT. It's the exact opposite. My objective is not to send you away to somebody else's site where you can find your answer. My objective is to give you the answer. Similarly, I expect that we will have our earbuds in and we will just say to our personal assistant, hey, book me the Mercer Hotel next Tuesday at the lowest price. I don't particularly care where you book it. I don't tell my assistant, hey, book it at this site or at this, right? I rely on her intelligence to figure out where to book it. So what is the consequence to the intermediaries, to the Expedias, to the Bookings, to all of these companies that sat in the middle that would collect a 20% or a 15% toll in the underbelly of Google when we move to a world that is agentically led. We heard an announcement this week about agent commerce with Walmart where now, you know, I can just say, you know, if I sign up with Walmart, hey, tell my assistant, sign me up with Walmart. By the way, make sure I have paper towels every month or whatever the case may be. I think it puts a lot more pressure on the underlying brands. So whoever the winners were in your app ecosystem, they're suspect. I'm not saying they can't make the transition, but I said to the CEO of Booking.com over the weekend, how are you going to get chat— how are you going to make it so that the first time I direct ChatGPT to book my hotel, I say, book me the Mercer Hotel on Booking.com. Because unless I do that right, then the 50% of traffic that you have coming to you for free today, I think that's going to be, you know, it's going to be a real question.

I would add just one more kind of point, bringing it back to actually that layer cake comment that I made earlier. And this is actually a comment from one of the large private company CEOs that I guess his layer cake would be chips, data layer, LLM layer, and then apps. But, you know, his view was you know, when you look at this LLM layer, it is going to consume or subsume this entire kind of SaaS app ecosystem. And, you know, that is a bold statement. Like, will it play out that way? Perhaps. But I think that really makes one— as an investor, you have to kind of question, like, what the future looks like and how much value, you know, these legacy software companies—

and the discount rates on those software companies have gone higher and the multiples have come down for sure because people are thinking about this.

But I think that, that in my mind is like a structural 10-year type question of like, what happens if we are in a world where the LLMs do just absorb all that?

As an anecdote in support of both of these comments, because Brad hit it on the agentic piece, there's at least one publicly traded company I know that alluded to this in their earnings, in their call, that said they were using something like 60 SaaS businesses for CRM, accounting, HR, you know, you go bit by bit, and now they're using AI internally to basically obsolesce those. So they'll go from like 60 down to like 6. So that's gonna be a big hit to some of these players.

So a few areas of focus, and we'll go in reverse order this time, you know, that I think are interesting over the next 5 years or 10 years. Josh, tell us a little bit about, you mentioned Android, but tell us a little bit about embodied AI or physical AI, you know, and just a few things that we can all think about, maybe a company or two.

Two big things that I'm interested in here, embodied intelligence is basically how do you take AI and put it into the physical world? A robot walking into here that was picking up chase this bottle, you know, if it's full, has to have an intuitive sense of how much force to use when it grasps. If it's down to 50%, we all can do that very intuitively, you look at it, you don't even think twice about it. But there is no training data for that. It isn't like they can train on Reddit and Wikipedia and Google Archive and all the books and everything. So a lot of that is physical people actually teleoperating these things. That goes into a question around humanoid robots. I personally am not a believer in humanoid robots, not robots, I'm a huge believer in robots, but there's no reason that we should have two arms and two legs. Evolution, is inferior to engineering. They should have 6 arms and 6 legs and do kinds of things where you look at it and like, holy shit, how did it just do that? And it's just like a creature out of the scene from the cantina in Star Wars. So they should look like Darwinian endless forms most beautiful. Last thing I'll give you, which is a big idea we're spending a lot of time on, and AI plays a key role here, is the most unsexy thing you can imagine, and it's sexy for that reason, which is maintenance. And why I say maintenance, we've been in a low-rate environment And there was a lot of CapEx and a lot of growth, and everybody was building planes, trains, and automobiles, and buildings, and CapEx, and HVAC systems, and industrial systems. And if you're a good activist investor or capital allocation committee or CFO, you're saying, "Enough with the spending. How do you maintain the assets you have for longer?" We're now looking at software for surveillance, early preventative maintenance, sensors for acoustic aberrations in machine that might be breaking down, robots for remote repair. I don't know what the term is going to be called. Somebody will come up with something cool like defense tech, it'll be maintenance tech. At the moment I'm calling it fixware, like software, hardware, and fixware. But this is going to be a new category where I think just rationally because of the cost of capital and the CapEx cycle, growth has been overdone, maintenance has been underdone, and you'll see new demands and technologies, and AI will play a big role in that.

Silicon Andrew.

Yeah.

So what's new and disruptive?

Yeah. So I'll keep on that theme of kind of like edge and moving intelligence out of the data center into the edge. And I think if you look at what happened in the data center over the last 3 years where, you know, 3 years ago it was an Intel CPU-first data center. And, you know, that has gotten washed away by the AI tsunami with, you know, Jensen riding his surfboard on top of that tsunami is kind of the analogy, right? And I think what you're going to have happen is a similar dynamic on the edge where you have kind of all these kind of, for lack of a better term, kind of like stupid devices that are going to get intelligence in them over the next 5 to 10 years. And I don't think the game is frankly even started yet. But even simple things like, I will go sit down with like the most sophisticated tech companies in the world in their conference room and like it takes 10 minutes to get the presentation on the screen. Right. But like stuff like that will be like instantaneous. And I think when you think about moving into whether it's robotics or autonomous driving or even frankly, your smartphone or your PC, like I think we're kind of in the top—

is that the device manufacturers benefit or are you thinking about some, some—

I think the semiconductor layer again there, right. You have a lot of Frankly, like Qualcomm, like, uh, yeah, I think Qualcomm is, you know, okayly positioned right now with kind of the opportunity. I think they'll probably have to do some M&A of some of the more kind of startups that are focused on AI.

Brad, you get, you get apps. Tell us about what's exciting on the application layer.

Well, I think the whole thing's going to be rebuilt. You know, I heard somebody say the other day, well, the apps haven't come yet. Well, the biggest app in the history of the world is already here. It's called ChatGPT. Yeah, right. And so, no, we have a few coding apps too. Right. So I think that everything is going to be derivative of that. Every internet company I know, they used to do website optimization. Now it's like, how do I build an agent that can interact friendly with the super agent? Right. We're going to have— we're going to have armies of agents that are interacting on our behalf. And so there are a lot of enabling companies, you know, Chase in that space. And then there are a lot of principal actors as well.

Brad, staying with you, one of the things that we've talked about is how— or A lot of people have talked about is how AI could exacerbate wealth disparity. You did something with Invest America that, you know, is interesting and topical in this room. So just give us a second on that.

Yeah, real quickly. First, I heard Jamie talk earlier, you know, about the challenges to capitalism, right? This city, more than any other city on the planet, knows that there's a fight for the soul of the country. Like, what's the future direction? Is it going to be a country that follows the direction of Europe and heads toward the direction of socialism, or are we doubling down on capitalism? I put forward and worked with a bunch of you guys on the Invest America Act starting 3 years ago. It got passed into law this year as part of the big beautiful bill. Trump accounts will go into effect next year. 65 million American children under the age of 18 are qualified, but starting in 2026 and forevermore, every child born in America will get a prosperity account at birth, a 401 from birth, seeded with $1,000 in the S&P 500. We're gonna have thousands and thousands of companies that are contributing to those accounts, moms and dads. And early December signups will begin, get everybody to sign up to these. You'll hear a lot more about it, but this is how we get everybody into the game. Capitalism doesn't work if 60% of people don't have assets that compound. We need them to feel like they're part of the system. And if we do that, right, I'm confident that we're going to overcome the challenges in the age of AI. Where there's no doubt going to be dislocation and displacement. The way we do that is we make everybody feel like they're on Team America, and I think it's a huge step in the right direction.

Well, we could keep going, but our time is up. I want to thank our panelists, and I want to thank Robinhood for bringing us together.

Thanks, guys. Great job.