Peter Diamandis opened the panel with a single figure. NVIDIA had just booked "$96.2 billion in revenue in a single quarter" — "more than the GDP of the majority of the countries on the planet," he said — up 106% from a year earlier, with guidance for another roughly 10% in the current quarter, about $108 billion. "For context, that is more than a billion dollars per day."
The company's results announcement on August 26 covers the quarter that ended on July 26, 2026. Revenue rose 18% from the previous quarter and 106% from a year earlier. Data Center — the chips and systems sold to companies building and running AI models — accounted for $89.0 billion of that, up 117% year over year. GAAP gross margin, the share of revenue left after the direct cost of making what was sold, was 75%. Net income came to roughly $59.7 billion. The forecast of $108 billion for the current quarter, plus or minus 2%, excludes any assumed Data Center compute revenue from China. During the quarter NVIDIA returned about $26 billion to shareholders through dividends and share repurchases.
Diamandis added the longer view: Jensen Huang's guidance for 70% growth in 2028, against a Wall Street consensus of 44%, and Huang's description of the frontier AI labs as "the first generation of startups that needed tens of billions of dollars of compute to get to their product."
The case that nothing here is fragile
Dave Blundin took the sustainability question first and answered it in two parts. NVIDIA has vulnerabilities, he said, "but at the same time, they're going to be sold out for as far as the eye can see no matter what." Everyone training neural networks has to use NVIDIA; by his estimate that work supplies 90% of the revenue and 95% of the profits. He put the gross margins on those revenues at 80 to 85% — higher than the 75% NVIDIA reported — and concluded that "the profitability of this company has never been seen before on the face of the earth."
The vulnerability he named is manufacturing. NVIDIA designs its chips but does not make them: TSMC is, in his words, "their one and only manufacturer of everything they sell," with roughly a third of TSMC's output going to NVIDIA, another third to Apple and the rest of the world sharing what is left. Building its own fabrication plants, Blundin argued, would have to be done "very quietly and sneakily," because that is not a boat you rock casually.
He also described how the money changes behavior around him. Meeting portfolio companies weekly, he hears about appointments with a manufacturer in Vermont, or with J.P. Morgan. His reaction: why are you meeting with anyone other than Jensen, one of the other giants or one of their satellites? "The amount of money pouring through just that funnel dwarfs the entire rest of the economy." If you are not in that flow of conversations, he said, "you're not in the most relevant conversation in the history of the world."
The elephant Alex wants measured
Alex, a regular on the panel, began by disclaiming the strongest version of the worry: "I'd be the last person to suggest that this is one big wash trade" — a trade in which the same parties move goods or money in a loop to manufacture the appearance of activity. What he wants instead is a measurement. He wonders "the extent to which the financial markets have fully priced in NVIDIA's financing of their customers," and would like a clearer distinction between demand NVIDIA has funded and demand it has not.
The distinction matters because a supplier that lends a customer the money to buy its products is recording sales it paid for itself. If a disproportionate share of the demand turns out to be "directly or indirectly backstopped or financed or credited by NVIDIA through one or more intermediaries," Alex said, "then I think that starts to look a little bit too bubbly, which I'm not thrilled with at all." What he does not want is to discover in two or three years that a large part of the demand at the infrastructure layer had been "artificially propped up through financial engineering." As he put it: "I'd like to see no hiccups in this singularity."
Even so, his forecast was not a crash. If some quantum of the demand does turn out to be inflated, he expects "at most a mini-winter," because in his view compute is substituting for real estate, human labor and other raw inputs as the basic substrate of civilization. He offered the doubt as a feeling rather than a finding, and said so: seeing growth numbers like these next to headlines about NVIDIA financing data-center deployments leaves him "a little bit of a tickle at this point to the back of my mind."
Two kinds of circle
Closing the segment, Alex separated the loops. In one, company A sells goods to company B and company B sells goods back to company A — call it an income-statement circular economy. You still do not want trades engineered to inflate revenue, he said, but from a bubble perspective that is the more healthful form. In the other, company A uses not its income statement but its balance sheet — private credit, loans and the like — to finance company B's purchase of company A's goods. An income statement records sales and costs; a balance sheet records what a company owns and owes. Lending a buyer the purchase price turns one into the other.
Blundin accepted that the AI economy looks circular — Jensen puts money into data centers, the data centers buy chips, the chips provide inference to the companies building on top — and argued it is not a trap. "Every economy is circular in nature," he said; dollars move in a circle, euros move in a circle. What has changed is that the circle is tightening around a new economy that "cares very little about the legacy economy." His illustration was a civilization on Alpha Centauri that discovers superintelligence and builds an enormous economy for itself: it would not bother trading with Earth, too far away and too slow.
His test for whether outside revenue arrives is drugs. He said a brother of his who works with Fidelity, Ameriprise and UBS had told him a year ago that AI would not really have hit until he saw it show up at Fidelity, and has since changed his mind. Blundin's argument is that Ozempic and Mounjaro together are bigger than all AI inference combined, from a single line of drug discovery; five or ten more such discoveries, which he expects soon, would send revenue back into the AI economy that dwarfs the legacy world. Diamandis offered a parallel from the early internet: Google's AdSense and AdWords looked like a modest new thing and ended up absorbing nearly the whole advertising industry's revenue.
Salim Ismail would rather it went slower
Diamandis put the investor's dilemma to Salim Ismail — Ray Dalio and others posting on X that they are out of the markets, price-to-earnings ratios off the charts, and at the same time this growth. Ismail called it a classic dilemma and said he hedges. He agreed the circular revenue is concerning, and was blunt about why: if the whole thing collapsed, "we're in a dead period for a period of time, and that just slows civilization down dramatically."
His preference is structural. "You're better off having steady, regular growth where the market has time to self-correct into new structures. And we don't have that right now. Things are kind of going vertical." He read NVIDIA's recent moves as an attempt to stop being only a chip company and become "literally an operating system for the whole of intelligence" — citing a reported acquisition of Hugging Face, which the panel noted is still a rumor, and comparing it to Microsoft buying GitHub for $7.5 billion, a deal he said finance staff nearly killed over the lack of assets until Satya Nadella argued they were buying the loyalty of 30 million developers. The bigger risk to markets, in Ismail's view, lies elsewhere: money printing, and what is happening in bond markets and Japan.
That leaves Alex's request unanswered. Speaking as a shareholder of the entire market through low-cost broad index funds, and not offering investment advice, he said he does not feel he has "tremendous visibility" into which of the two circles the current numbers describe.