UK Committee Chair Warns AI Boom Echoes Dot-Com Debt Spiral

Justifying the trillions of dollars now being spent on AI infrastructure will require more than $2.5 trillion in annual revenues by 2030 — and a senior UK parliamentarian who lived through the dot-com collapse says the sector's financing structure makes that outcome look fragile.

Chi Onwurah, a Labour Party member of the UK Parliament and chair of the House of Commons Science, Innovation and Technology Committee, argues in an article distributed by Project Syndicate that the AI investment boom may repeat the debt-fueled infrastructure failures of the early 2000s. She writes from direct experience: she worked as an engineer at a telecoms infrastructure company that went bankrupt in the dot-com bust, a collapse that cost her job, her visa, and her right to live in the United States.

The stakes are macroeconomic, not just sectoral. Technology, media and telecommunications (TMT) companies now represent nearly half of the broader US large-cap equity market, according to the article — up from one-fifth during the 2008 financial crisis. A downturn in this one sector would therefore transmit through markets with a force that earlier crashes could not match. In plain words: the AI sector is now big enough that its failure mode is everyone's problem.

What happened to the last 'shovel sellers'?

Onwurah's argument rests on a historical parallel. During the late-1990s internet frenzy, the prevailing investment thesis held that infrastructure providers — the companies selling "shovels" in the Gold Rush metaphor of the era — were the safe bets, insulated from the failures of the application layer. One company she worked for used an investor presentation showing the Sierra Nevada mountains with the text: "In 1848, you didn't want to be panning for gold. You wanted to be selling shovels."

The thesis failed. Global Crossing and WorldCom filed for Chapter 11 bankruptcy, while Nortel Networks and Lucent lost approximately 99 percent of their stock value. The infrastructure layer, it turned out, was the most exposed layer, because it had relied most heavily on debt to finance large capital expenditures.

"Fundamentals can be overlooked for only so long," Onwurah writes — a warning aimed at an industry whose leaders, she notes pointedly, are almost all too young to remember the crash firsthand.

The survivors were not the leveraged infrastructure builders. Microsoft, Google, Amazon and later Facebook bought out would-be rivals and consolidated into the monopolistic giants of today. In plain words: the last boom's winners were the deep-pocketed, not the highly leveraged.

Is AI financing circular?

The structural mechanism that destroyed the telecoms sector, Onwurah argues, is present again. In the dot-com era, a significant share of infrastructure financing was circular: equipment vendors lent customers the money to buy their own products, a practice known as "vendor financing." The company she worked for made hardware designed to boost high-speed internet capacity and often gave customers the money to buy the boxes. "Everyone made money and got paid back as long as the boom continued," she writes. "But when the bust came, the virtuous circle turned vicious."

She identifies comparable arrangements in today's AI market:

• Anthropic buys compute capacity powered by Nvidia chips from a company in which Nvidia is a major investor.
• Microsoft has invested more than $13 billion in OpenAI while also selling it cloud services through its Azure division.

Onwurah poses the question without answering it: "Is this a case of financial innovation? Or is it an unsustainable boom-and-bust engine?" That she raises it at all, from her position chairing the UK parliament's science and technology scrutiny committee, signals a regulatory attention to AI's balance sheets, not just its safety properties.

Why the downside could be worse than 2001

Onwurah lays out two scenarios, both unpleasant. If the revenue materializes, it must come from paying customers using AI to improve their own revenues — "which will generally mean removing costs, primarily salaries, from their bottom lines." Removing trillions of dollars from the West's salary bill would, she argues, cause societal upheaval, and given current private debt levels, a financial crash may well follow. Alternatively, markets conclude AI firms cannot repay their debts, and the order reverses: a potentially unprecedented financial crash first, followed by societal upheaval.

She calls the potential crash unprecedented for two reasons. First, the scale of debt exposure today is considerably larger than in 2001, making the situation more comparable to the run-up to the 2008 global financial crisis. Second, wealth is far more concentrated in the TMT sector than at any prior crisis point — nearly half of US large-cap equity, versus one-fifth in 2008.

Booms end when financiers stop lending into them. Onwurah dates the beginning of the end for her own company precisely: "The first big sign that my sector was going under in 2001 was American Express's decision to suspend our company credit cards, which came some months before we filed for Chapter 11 bankruptcy protection." Small credit signals, in other words, can precede large failures by months. In plain words: the first warning may arrive not as a market crash but as a bank quietly cutting off a borrower.

What does this mean for UK policy?

The policy hook is specific. The UK holds the G20 presidency in 2027 and, Onwurah notes, is putting AI at the center of its ambitions for it. Her recommendation is that the presidency agenda include security, standards, and — the notable addition — explicit planning for a potential AI crash.

She also raises, without resolving, the question of where capital would go if financiers withdraw from AI: whether redirected funding would flow to productive local economies such as Newcastle's, or to unemployed young people, or whether investors would simply flee to traditional safe assets like gold. "I do not know," she writes — a candor unusual in policy commentary, and a marker of how unsettled the scenario remains.

The constraints on her argument deserve note. Onwurah does not claim the AI boom is certain to bust, does not estimate a timeline beyond the 2030 revenue figure, and does not allege that any named arrangement is improper — only that the circular financing pattern rhymes with a pattern that ended badly once before. Her evidence for the parallel is structural, not audited. What she establishes is exposure: heavy capital expenditure, circular money flows, concentrated equity valuations, and a debt environment larger than 2001's.

Whether AI generates the $2.5 trillion in annual customer revenue that the infrastructure buildout requires is, on current evidence, an open question — and the person asking it now chairs the committee that oversees UK science and technology policy.