Investor Jordi Visser returns to argue the entire US economy is now an AI trade: stocks are carrying consumption, the agentic era turned tokens into a chips-and-energy commodity, and the real risk is bottlenecks, not a credit freeze. Plus why crypto becomes the settlement layer for all of it.
↓ Jump to the video and timestamps
Jordi Visser came back on the show this week, and he came with an argument that sounds like an exaggeration until you sit with the numbers: the entire US economy is now an AI trade. Not a sector of it, not the part of it that shows up in the Nasdaq, the whole thing. I follow his weekly updates closely, and he has been ahead of almost everyone I read on this theme, so when he frames it that bluntly it is worth slowing down on.
The setup for the claim is uncomfortable for a lot of people, which is part of why Jordi keeps making it. If you spent last year insisting AI was a bubble and that the revenue would never show up, the charts since then have been a personal problem, not just a market one. His read is that the hardest thing in this market is not analysis, it is admitting you were wrong in public after you put your name on the other side of it.
This is a markets conversation, not a Bitcoin one, but it lands on Bitcoin by the end, because Jordi's view is that the same forces remaking the equity market are the ones that make crypto the settlement layer for whatever comes next. Here is the case he made.
The argument Jordi is borrowing and extending comes from Apollo's chief economist Torsten Slok, and it is genuinely radical once you follow it through. Apollo's own 2026 outlook makes the point in plain numbers: AI-related investment contributed more to US GDP growth than consumer spending in 2025, while corporate investment outside of AI has flatlined to roughly zero growth.
Jordi's extension is about the consumer side. Consumption is close to two-thirds of GDP, and historically you do not get consumption growth without job growth. Right now job creation has stalled and consumption is holding up anyway. His explanation is the stock market. A ten percent move in equities is something on the order of six trillion dollars of net worth, most of it owned by Americans directly or through pension funds, and that wealth effect is doing work that paychecks used to do. Add the share of Americans who receive transfer payments, and you get something that functions, in his words, a lot like a universal basic income running through the market. A rising market is no longer just a Wall Street story. It is the consumption engine.
That is the part economists mostly reject and Jordi mostly embraces, and it is the load-bearing claim under everything else he said.
The most common pushback on all of this is that we have seen this movie, and it ends with a crash. Jordi's answer is that the comparison gets the mechanism wrong.
In a traditional capital cycle, the thing that breaks is the debt. Monetary policy tightens or someone gets over their skis, the credit market seizes, and the deleveraging starts; then come the layoffs to defend margins. The dot-com unwind ran on that script, and so did 2008. His view is that the AI build-out does not break that way, because the constraint is not leverage. It is physical. As he puts it, the bottom of Jensen Huang's stack is energy and chips, and you cannot deleverage your way out of a shortage of gigawatts.
The demand side backs him up. Goldman Sachs Research projects token consumption rising roughly 24 times by 2030, reaching on the order of 120 quadrillion tokens a month, with a growing share coming from autonomous agents rather than humans typing prompts. Where the 2000 broadband build raced ahead of demand that was not there yet, this build is chasing demand it cannot currently supply.
I pushed back on the scarier version of the bubble case, the circular-financing theory that the hyperscalers are quietly Enron-ing the whole thing into existence. Michael Burry has been beating that drum and a lot of people are reading it as gospel, but his record since the one great 2008 call has been mixed, and every time I actually use these tools the honest answer to "is this real, is it getting better, do I want more" is yes, yes, and yes. Bottlenecks are real. A fraud is a different claim, and I have not seen it.
Jordi's caution is more interesting than either the permabull or the permabear position. He keeps a spreadsheet of the probability that specific gigawatt-scale data centers actually get built on schedule, and his honest read is that the delays run in years, not weeks. So the negative catalyst, when it comes, will not be a credit event. It will be a single data point, a quarter where DRAM prices roll over or CPU demand softens or a power timeline slips, and the market will seize on it as confirmation the whole thing was a mirage. He thinks that is a real part of the distribution of outcomes. He just does not think it is the most likely one.
The framing of Jordi's I keep coming back to is that there is no signpost marking the moment everything changed, so people miss that it already did. His marker is late 2025, the end of the pre-training era and the start of the agentic one. Before that, the consensus was that agents would arrive slowly, a handful this year, a couple hundred next year. Instead they arrived all at once.
His way of describing the demand shock is deliberately physical. Imagine the planet's population doubled in a single year. Think about what that does to traffic, to food, to oil, to every commodity humans consume. That is roughly what happened to the digital economy, except the new consumers are agents consuming tokens, and a token is just chips plus energy. That is why there are now token futures. It is being priced like the commodity it is.
This is also where his investment conclusion comes from, and it is blunt: own the build-out, not the software riding on top of it. He is skeptical of seat-based, human-priced SaaS, because the agentic world routes around the seat. The companies he finds interesting are the ones with something the agents cannot reproduce. His favorite example is pharma, which he calls "human software," companies like Eli Lilly sitting on a century of research data, including every failed compound, that AI can now re-examine for combinations nobody had the time to test. The asset is the data, especially the data on the losers.
Two ideas from the back half of the conversation stuck with me.
The first is a reframe Jordi uses for almost everything. He wants to retire the word "side effects" and replace it with "trade-offs."
Instead of saying side effects, the true word to be using is trade-offs. Everything is a trade-off.
He applies it to medicine, to careers, to portfolio construction, to whether his mother should take a vaccine she personally would not. The reason he ties it to AI is that the tools are very good at surfacing the other side of a decision, the cost you were not weighing, and most bad decisions are just trade-offs someone refused to look at squarely.
The second is organizational, and it is the part most relevant to anyone running a company. Jordi pointed to Jack Dorsey's recent conversation with Sequoia, where Dorsey lays out a model for building a company as an intelligence rather than a hierarchy. Dorsey's framing, in his piece From Hierarchy to Intelligence, is that AI can now do the information-routing that middle management existed to do, so you compress the layers and let people work at the edge, where human judgment and context actually matter.
I told Jordi this was validation of what we have been building at TFTC, because it is exactly the shape of it. There are six of us, and since January we have been building a memory system the whole team queries: every podcast transcript, every newsletter, every tweet, every ad partner, organized so anyone can ask the agent what we have said about a topic and how our take has changed, instead of asking a person. We are a media company, so almost everything we produce is text, which makes us close to an ideal case for these tools. The point is not the specific stack. It is that the org chart that has governed companies since the industrial age is not a law of nature, and the teams that internalize that first are going to have a real, durable edge.
Everything above is a markets argument. The reason it belongs on this show is where Jordi takes it.
His case for crypto does not depend on price. It depends on plumbing. The agentic economy is going to generate transaction volume the existing financial rails simply cannot carry, and machine-to-machine payments do not fit neatly into a system built for humans and card networks. To Jordi that makes crypto adoption a question of when, not if. It is the same argument the Cashu developer Calle made when he came on to talk about ecash and the future of digital cash, approached from the investor's side instead of the builder's.
He is most animated about tokenization, which he frames as this cycle's version of the ETF moment, a structural change that looks slow until it is suddenly everywhere. He ran the S&P options book and then the ETF business at Morgan Stanley back when most people thought ETFs would go nowhere, and he thinks tokenization rhymes with that: a wrapper that quietly eats the old structure. His provocative claim is what it does to the assets that have not been priced in real time. Private credit, private equity, real estate, the illiquid buckets institutions love precisely because they produce a steady return without visible volatility, get repriced once tokenization brings them price discovery and a live mark. That, in his telling, is not a credit crisis, which he is adamant is not coming. It is the slow end of an era where you could earn a bond-like return and never have to look at the volatility.
Where does that leave Bitcoin? His line is one to sit with:
Bitcoin is the S&P 500 of the digital asset world.
The point is that it is the default index of the space, the place capital consolidates when a given token's thesis breaks, the same way money flows back into the S&P when a single stock fails. He is candid that Bitcoin has been a poor trade lately, sitting near 70,000 in what he calls a bear market when a year ago he would have guessed 500,000. He is not trying to call the bottom. He uses moving averages instead, and his reasoning is the most quietly humble thing he said all hour:
I think moving averages are an admission that the market knows more than you do.
What he is waiting for is the breakout, the moment the parabolic chart that AI names have enjoyed shows up in crypto. And he made one observation that reframes Bitcoin's old reputation. The volatility that used to be the knock on Bitcoin, the 70 to 80 readings that made institutions flinch, is now the volatility of everything that is working in the AI trade. The market has spent a year getting comfortable with charts that look the way crypto charts have always looked. When the breakout comes, he thinks the audience for it is already trained. For the longer version of why the settlement question matters, I wrote separately about how the digital dollar is already arriving, just not as a CBDC.
Jordi Visser is an investor and macro strategist with three decades on Wall Street, including running the S&P options and ETF businesses at Morgan Stanley and opening the firm's Brazil office. He now publishes widely followed research on AI, markets, and crypto through his Substack and YouTube, where he puts out weekly updates.
