1 week ago
45
George Noble, managing partner at Noble Capital Advisors, is making a claim that should make every tech investor pause mid-scroll: the current AI investment wave could become the single largest misallocation of capital in history. Not just bigger than the dot-com bust. Roughly 17 times bigger.
Big Tech’s AI infrastructure spending has ballooned from $155 billion in 2022 to a projected $755 billion in 2026, with estimates suggesting it will blow past $1 trillion by 2027. The question Noble is posing isn’t whether AI matters. It’s whether the returns can possibly justify the receipts.
The productivity gap nobody wants to talk about
Research from Wharton suggests that for AI investments to break even at current trajectories, productivity gains would need to roughly triple. That hasn’t happened.
Noble also flags what he calls circular vendor financing and “token maxing” as practices that obscure the real financial picture. In plain terms, companies are using accounting structures that make AI investments look more productive than they actually are.
NVIDIA’s receivables and the China problem
One data point Noble highlights is particularly striking: NVIDIA has seen a 770% growth in receivables. For a company that has become the poster child of the AI boom, that number raises questions about whether demand is as organic as it appears, or whether it’s being propped up by aggressive financing arrangements between vendors and their biggest customers.
Meanwhile, Jefferies has warned that cheaper Chinese open-source AI models could inflict serious damage on the US AI market. If companies can achieve comparable results at a fraction of the cost using open-source alternatives developed overseas, the moat around trillion-dollar infrastructure investments starts looking more like a decorative pond.
Tech executives caught between two messages
That tension was on full display in mid-September 2026, when a researcher at Anthropic publicly warned about potential extinction-level risks from AI technologies before resigning.
What investors should be watching
During the dot-com era, telecommunications companies spent roughly $65 billion laying fiber optic cable that went largely unused for years. Noble is essentially arguing that AI could follow the same script, just at a much larger scale.
The Wharton research on required productivity gains provides a concrete benchmark for investors. Rather than relying on management guidance about future AI revenue, tracking whether actual productivity improvements are accelerating or plateauing will offer a more honest read on whether these investments are paying off.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) ·