NEW YORK (Realist English). Bloomberg has published an analysis warning of systemic risks accumulating in the US economy around investments in artificial intelligence. The publication calls the current structure a “wobbly house of cards” capable of collapsing markets and dragging down the real economy.
The Key Risk: From “Bubble Theory” to “Systemic Interconnection”
Unlike previous warnings about technology bubbles, the distinctive feature of the current risk lies in the deepening cross-investment relationships between AI companies and major cloud providers. This mechanism of mutual share ownership and circular movement of funds means that a revaluation of one node can simultaneously hit the entire network.
Concretely, this works as follows: an AI startup raises funding at an extremely high private valuation and uses the money to buy computing power, while the cloud provider supplying that computing power is its shareholder; in the same quarter, the value of the shares held by the cloud provider — based on the next funding round’s valuation — is reflected in its financial statements as income.
When the buyer and seller of “AI revenue” are the same few giants, a revaluation of any one party simultaneously affects all participants.
Key Data: Concentration and Leverage
Concentration risk. Bank of England Governor Andrew Bailey, who also chairs the Financial Stability Board, warned ahead of the G20 meeting that the market is “vulnerable to a disorderly correction capable of spreading across borders.”
He noted that the problem is not only that investors are borrowing more, but that “leverage interacts with high valuations and market concentration, especially the growing cross-investment between AI companies and large cloud providers, which could amplify a future market correction.”
Index fund exposure. A single chip company accounts for more than 7% of the S&P 500’s weight, while the combined weight of the index’s ten largest stocks approaches 40%, exceeding the dot-com bubble peak of around 29%. This means that in the “diversified” index fund of an ordinary investor, roughly 7 cents of every dollar comes from a single semiconductor stock.
Valuation divergence. Nvidia’s forward P/E is around 59, while its main customers (Alphabet at around 17, Meta at around 21, Microsoft at around 28) have significantly lower valuations. This gap itself represents a “controlled experiment”: if a genuine AI shock arrives, it should hit the still-unprofitable labs and 59-times chip stocks hardest, while companies with diversified cash flows suffer less.
Macroeconomic Transmission Mechanisms
Rating agency Fitch warned in its July report that vulnerability to an AI-related market correction has become one of the main sources of short-term risk for the global credit environment.
The agency estimates that information technology fixed investment accounts for 5% of US GDP, one percentage point above the pre-pandemic level; AI investor optimism and the wealth effect from rising stock markets have also been an important support for US consumer spending growth.
The Federal Reserve’s May “Financial Stability Report” shows that 50% of surveyed market participants named AI as a potential shock over the next 12–18 months, significantly higher than 30% six months earlier.
Respondents are concerned about the following channels: equity valuation risk, rising systemic leverage due to growing dependence of capital expenditure on debt financing, and the possible impact of widespread AI adoption on the labour market.
Stress Scenario: If the Bubble Bursts
In a stress scenario published in September, Fitch provides quantitative consequences: if a significant correction in AI stocks occurs, US stocks could fall by 35% within six months and plunge the economy into recession.
The agency estimates that in a recession scenario, GDP could contract by 1.5% in the second quarter of next year, private capital expenditure could fall by more than 6%, and US investment could shrink by 4.8%.
Broader contagion effects also deserve attention. Fitch estimates that in 2027 global GDP growth could fall below 1%, which on a per capita basis is equivalent to stagnation or recession in the world economy; stock markets in countries such as Japan and the UK could also fall by around 15%.
Regulator Warnings and Political Pressure
IMF Managing Director Kristalina Georgieva said in April that the global financial system is not prepared for the growing security threats associated with AI technology.
The IMF report notes that advanced AI models can identify and exploit system vulnerabilities at lower cost and faster speed, and that cyber risk increasingly manifests as “correlated failures” capable of disrupting financial intermediation, payment systems, and market confidence at a systemic level.
US Senator Elizabeth Warren said directly in an April speech: “I know what a bubble looks like. The resemblance to the 2008 crisis is striking.” She noted that AI companies are investing trillions of dollars through debt, and the industry needs to generate around $2 trillion** in annual revenue by 2030 to justify these investments, whereas in 2025 the entire industry’s revenue was only **$20 billion — about 1% of the required volume.
Warren also warned that some AI companies are already preparing the ground for possible taxpayer-funded assistance: OpenAI publicly promotes the idea of federal government “intervention” and “de-risking AI expansion” through expanded tax breaks and loan guarantees.
Key Conclusion: Technological Success Does Not Equal Financial Success
The Bank for International Settlements, in a July working paper, described the current AI buildout as “one of the largest technology-driven investment booms in US history” and warned that its scale and debt dependence could repeat the fate of historical bubbles in canals, railways, and the internet.
The bank’s model shows that the AI race leads to excessive investment roughly 1.5 times above the socially optimal level, and in a scenario with low demand elasticity — up to 3 times.
As The Edge Malaysia weekly noted: “AI tools can be widely adopted, data centres can be intensively used — and still fail to generate enough cash to repay owners’ debts. Concern about financial stability stems from the mismatch between speculative future revenues and current contractual obligations.”
