Multiple Fed Officials Flag Financial Stability Risks from AI Investment Boom
Alina Collins
Several Federal Reserve officials have raised concerns about leverage and financing structures behind the AI investment boom — not calling it a bubble yet, but placing it squarely on the central bank's stability watchlist.
Is this a bubble? What does the Fed actually think?
New York Fed President John Williams told Reuters plainly: "I don't think this is a bubble situation." He acknowledged that borrowing tied to AI investment is rising.
His reasoning: investors are pricing an "almost impossible question" in real time — how large AI returns will actually be. This means → volatility is a byproduct of the pricing process, not a sign of systemic breakdown.
Williams added that the companies involved have strong earnings power, and said he is "not too worried about leverage-related financial stability risks right now."
Where does the "too big to fail" worry come from?
Kansas City Fed President Jeffrey Schmid struck a more cautious tone, asking directly whether the AI sector is becoming another "too big to fail."
His focus is not any single company but the entire financing chain — data centers, energy suppliers, and the communities they serve are locked in "circular commitments." In plain terms = A guarantees B, B guarantees C, and C depends on A's orders. One break can cascade.
Schmid pressed further: are these circular commitments already over-leveraged, and how would a "spark" spread?
Investments still at the "announcement stage" — good or bad?
San Francisco Fed President Mary Daly noted that many AI commitments remain announcements, not yet converted into physical assets.
This reflects a counterintuitive logic: money not yet spent actually means "stranded asset" risk is lower for now — there are no large physical assets locked in yet.
But Daly warned that rising borrowing is the real concern, and called for a forward-looking risk dashboard — "not watching what happened during the financial crisis, but watching what could make all of this go wrong."
Compared with the 2005 housing boom, where does AI investment stand?
Apollo chief economist Torsten Slok offered a benchmark: current data-center construction is less than half the 2005 housing boom peak, which hit 6.6% of GDP.
The critical number, though, is the growth rate: AI investment relative to GDP is already expanding faster than pre-crisis real estate did.
This means → the absolute scale is still manageable, but the pace itself warrants caution — the 2005 housing market did not turn into a crisis overnight either.
What does this mean for markets?
Multiple officials speaking up at the same time signals that AI financial risk is moving from a fringe topic to the center of policy discussion.
The Fed has no unified position yet: one official says "not a bubble," another is already debating "too big to fail." In plain terms = the central bank hasn't reached a verdict, but it has started paying serious attention.
The next key test: whether AI financing structures can hold up in a high-rate environment.
Content is for reference only, not financial advice.