David Einhorn on the AI Mania
In his Q3 2025 letter, superinvestor David Einhorn talks about the current state of AI.
David Einhorn devotes much of his Q3 2025 letter to dissecting what he calls the “AI spending delusion.” His tone blends humor and alarm, as he highlights the disconnect between the staggering sums being promised for AI investment and the basic arithmetic that underpins them.
He opens by mocking the arms race among tech CEOs. Tim Cook, Mark Zuckerberg, and Sam Altman have all pledged to spend hundreds of billions—if not trillions—of dollars on AI infrastructure, often without credible explanations of how such sums will be financed or recouped. Drawing from McKinsey data, Einhorn points out that global AI-related CapEx are expected to reach $6.7T by 2030, with $5.2T dedicated to data centers alone. Yet, he asks a simple question: “Where will the nearly $7T come from?”
There is a good chance that 25 years from now, AI will turn out to be even more important than we currently imagine. But, the path from here to there is likely to be very bumpy for investors.
The math, he argues, simply doesn’t add up. Even the Mag 7 tech giants collectively generate only about $500B in annual FCF, half of which already goes to buybacks and dividends. The total book equity of these firms is roughly $1T — a fraction of the planned AI outlay. Even if Wall Street, private equity, and VC emptied their coffers, there would still be a multi-trillion-dollar funding gap that could only be filled with massive new debt issuance. To justify this spending, AI would need to generate around $2T in annual revenue by 2030, or roughly the size of _the entire current global advertising and software subscription markets combined.
Rather than external enterprises or consumers spending new money, much of the AI revenue comes simply from AI companies buying products and services from each other.
David concedes that AI will likely become more transformative than anyone currently imagines — just as the internet turned out to exceed even the wildest expectations of 2000. But he warns that the path from here to there will be littered with capital destruction. Investors are mistaking the importance of the technology for the profitability of the investments tied to it, much as they did during the dot-com bubble. He calls the current moment “the most expensive market we’ve ever experienced,” where excitement has replaced analysis.
The capital spending numbers being thrown around today are so extreme that it’s really, really hard to understand them. There is a reasonable chance that a tremendous amount of capital destruction is going to come through this cycle, even if AI ultimately turns out to be everything it’s cracked up to be… and more.
Do you like this post? I think you’re gonna enjoy WorldlyInvest Pro:
Stock research in microcaps and special situations.
Join now and get a 40% discount for the annual plan.
He then digs into the mechanics of the AI revenue illusion. In one example, he shows how a single dollar spent on a ChatGPT subscription cascades into $8 of reported “AI revenue” across the supply chain — from OpenAI to Microsoft, to CoreWeave, to NVIDIA — even though the underlying activity is loss-making. In reality, much of the AI “revenue boom” is circular: AI firms are buying hardware and cloud services from each other rather than serving end customers. This, Einhorn warns, creates the illusion of profitability while masking massive value leakage and overcapacity risk.
He also questions the faith in Artificial General Intelligence (AGI) — the belief that computers will soon surpass human reasoning. Current LLMs, he notes, are essentially statistical pattern machines that mimic reasoning without understanding. They don’t “think” in any human sense; they only generate correlations that give the illusion of inference. While these systems will improve, David doubts they are on the verge of true reasoning or “self-improvement.” Drawing a parallel to self-driving cars — which were perpetually “one year away” for a decade — he argues that expectations of near-term AGI are wishful thinking amplified by financial incentives.
His conclusion is both clear and caustic: AI might change the world, but investors are unlikely to profit from it in its current phase. The capital required to build it dwarfs the plausible returns, and Wall Street’s role in promoting the narrative only fuels the excess.
When the tide turns, it does so quickly and without warning. Even 25 years later, it’s still not clear why the internet bubble popped when it did. Our view is that it was due to the last buyer buying and the last short seller covering – a phenomenon that is very difficult to time. This remains the most expensive market we have experienced, and we don’t see a better option than continuing to be cautious.