If AI Fails To Deliver On Its Vast Productivity Promises, The Result Will Be A Devastating Depression

Moments after the Fed's surprising decision to not only cut rates but to launch Reserve Management Purchases, QE Lite, Not QE, or whatever one wants to call it, we said that, in response to Powell's comment that the implication of the Fed's rather bizarre forecasts is higher productivity spawned by AI, the US might as well "do $10 trillion fiscal stimulus for AI/data center capex? Fed joins Musk in betting on AI revolution." After all, if you are going all in and still printing the world's reserve currency, at least make sure you will emerge victorious from the civilizational AI war with China... as anything else would represent the way of the Western way of life. 

*POWELL: IMPLICATION OF FED FORECASTS IS HIGHER PRODUCTIVITY

In that case why not do $10 trillion fiscal stimulus for AI/data center capex? Fed joins Musk in betting on AI revolution.

— zerohedge (@zerohedge) December 10, 2025

And now, One River's CIO, Eric Peters, agree with us: if going all in, make sure it is really all in. Excerpted from his latest weekly note to clients:

“Let me tell you a familiar story,” I said, discussing market risks, opportunities, in Stockholm this week.

Early in my career, Sweden hit the wall. Decades of economic math that hadn’t quite added up had compounded to a point where a market unwind was unavoidable.

As the crisis climaxed in late 1992, the central bank hiked overnight interest rates to 500% to punish speculative short-sellers of kronor. We made a small fortune in that market chaos.

That was my first of many crises. Every one looked different. In each, investors, high and low, were brought to their knees. Some combination of overleverage, overreliance on parallels to the past, intellectual overconfidence, inadequate risk management, and lack of a sufficiently expansive imagination, is to blame.

With each successive crisis since my start in 1989, central bankers and politicians intervened ever more aggressively to forestall economic cleansing, borrowing from the future to repay the past and secure the present.

Such interventions have transferred risks and excesses from individuals and corporations onto the government balance sheet.

The math of our federal debts and entitlement commitments no longer adds up. The consequences are compounding. I suspect that before my career is over, a sovereign crisis like Sweden’s in 1992 will erupt in the US.

The only practical way to forestall this is through a productivity boom that materially lifts non-inflationary growth and repays our policy sins.

This is why the US government is all in on the AI buildout. If AI fails to deliver on its vast promises, the risk is far wider than the tech sector. It will lead to a deep recession and massive budget deficits, which will catalyze a debt sustainability crisis that could make 2008 seem trivial.

Sovereign debt crises are the most devastating of all financial calamities, because the buyer of last resort is the one in trouble. But I suspect that the AI boom doesn’t end neatly with Cisco retaking its 2000 highs, as it did this week.

The parallels with the dot com boom/bust are all too evident and well publicized at this point.

Which means we should still be long this bull market, while also running substantial downside hedges.

Positioned for a world with two big fat tails.