Jeff Schmid, president of the Federal Reserve Bank of Kansas City, said in a speech this week that investments in artificial intelligence (AI) have grown large and interconnected enough to require macroprudential attention. The use of the phrase "too big to fail" in connection with the AI sector signals a noticeable shift in the tone of official communications about the technology.
What worries regulators?
Schmid focused on two principal concerns: the scale of the investments and the close financial links among industry participants. He argued that the current wave of AI investment resembles past boom cycles in its potential to concentrate activity: such concentration raises the risk that problems in a single sector could spill over to the broader economy.
The numbers cited are significant. Large technology companies have collectively committed nearly $2,400 billion to AI-related investments. That sum far exceeds typical corporate investment cycles and leaves less room to adjust if demand falls short of expectations.
Financing chains and credit risk
Regulators are also worried about how these investments are financed. The Bank for International Settlements (BIS) has warned that a collapse in AI markets could reverberate through credit markets in a way comparable to the 2008 financial crisis, because much development is backed by debt and layered financing structures. Schmid highlighted that when chipmakers, cloud providers and AI-model developers finance each other, the failure of any one party could trigger a chain reaction, and external observers may find it hard to map actual exposures.
Market signals
Investors are beginning to price in these risks. For example, the wave of AI deals linked to Nvidia has pushed up the spreads on the company’s credit-default swaps (CDS), suggesting that lenders to even the sector’s strongest firms perceive greater risk.
That said, the present situation is not identical to earlier asset bubbles. Some indicators — valuations and market concentration — recall the dot-com era, yet today's leading firms generally generate real profits, unlike many internet-era favourites around 2000. This presents a dilemma for the Federal Reserve: while profitability points to more solid fundamentals than pure speculation, the sheer volume of investment and reliance on debt mean that even a fundamentally healthy sector could transmit financial shocks if market sentiment shifts suddenly.
Implications for monetary policy and supervision
Schmid warned that rising AI-driven demand for energy, semiconductors and construction capacity could complicate assessment of inflationary pressures and interest-rate decisions. At the same time, spending on AI development is approaching some firms’ free cash-flow capacity, prompting a greater reliance on borrowing and external investors. As financing shifts from corporate balance sheets toward credit markets and private lenders, risks may migrate to financial institutions that the Fed supervises.
A warning, not a prediction
Schmid's remarks are not a prediction of collapse. A regional Fed president’s role includes drawing early attention to potential systemic risks before they materialize. Nevertheless, the application of “too big to fail” to AI indicates that policymakers now view the sector not merely as a market or technology story but as an issue with implications for the stability of the financial system.
Taken together with prior BIS warnings, Schmid’s speech may spur regulators and policymakers to map AI financing chains more thoroughly and identify financial intermediaries whose exposures could pose systemic hazards in the event of a market shock.
Conclusion
According to Jeff Schmid, the size and financing structure of current AI investments have reached a level that warrants macroprudential scrutiny. Concentration and intertwined financing arrangements remain the clearest channels through which problems in the AI sector could spread to the wider economy.
