Arthur Hayes warns AI could trigger new subprime crisis

Arthur Hayes has a theory about AI, and it doesn’t involve robots taking over the world. It involves something arguably scarier: robots taking your mortgage payment.

The BitMEX co-founder and CIO of Maelstrom published an essay titled “This Is Fine” on February 18, laying out a scenario where rapid AI adoption triggers a wave of white-collar unemployment so severe it could rival the 2008 financial crisis. His thesis: when knowledge workers lose their jobs en masse, they stop paying their credit cards and mortgages, and the banking system absorbs a hit it isn’t remotely prepared for.

The math behind the meltdown thesis

Hayes estimates that 20% of the 72.1 million US knowledge workers could lose their jobs to AI. That’s roughly 14.4 million people suddenly without income.

The downstream effects, in his model, are brutal. He projects $330 billion in consumer credit losses and $227 billion in mortgage losses, totaling $557 billion in defaults.

The damage wouldn’t be distributed evenly across the banking sector. Hayes estimates the shock could erase 13% of US commercial bank equity, with regional banks absorbing the worst of it. These are the institutions most exposed to consumer lending and residential mortgages, the exact categories that blow up when middle-class professionals suddenly can’t make payments.

What this means for Bitcoin

Hayes points to Bitcoin’s recent divergence from the Nasdaq 100 as evidence that fiat credit conditions are already tightening.

Bitcoin fell from a high of $126,000 to near $60,000, a decline Hayes attributes to the market pricing in AI-induced credit contraction before most people even recognize the problem.

When regional banks start failing, he argues, the Federal Reserve will have no choice but to respond with unprecedented monetary easing. In Hayes’ scenario, Bitcoin and select crypto tokens would surge beyond previous all-time highs once the Fed’s response kicks in.

Why this argument deserves scrutiny

Hayes has a track record of making bold macro calls. His earlier essay “Double Happiness” from the 2024-2025 cycle explored similar themes around AI’s intersection with crypto markets.

The 20% displacement figure for knowledge workers is a projection, not a certainty. AI adoption timelines are notoriously difficult to predict. The difference between “20% of knowledge workers lose their jobs over 18 months” and “20% of knowledge workers lose their jobs over a decade” is the difference between a crisis and a transition.

Regional banks proved in 2023 that they can still fail spectacularly when conditions shift faster than their risk models anticipate. Silicon Valley Bank didn’t collapse because of exotic derivatives. It collapsed because of a straightforward duration mismatch that nobody in management bothered to hedge.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.