Arthur Hayes argues that AI’s biggest threat may not be machines taking over the world, but large numbers of salaried workers losing paychecks—and with them, the ability to service loans.
In an essay titled “This Is Fine” published February 18, the BitMEX co-founder and Maelstrom CIO lays out a scenario in which rapid AI adoption produces a wave of white-collar unemployment severe enough to ripple through the financial system. His basic chain of events: mass job losses among knowledge workers reduce consumer income, driving spikes in credit-card and mortgage defaults, which then stress banks that are ill-prepared for that magnitude of consumer deterioration.
The math behind the thesis is stark. Hayes uses a working estimate that 20% of the roughly 72.1 million U.S. knowledge workers could be displaced by AI—about 14.4 million people suddenly facing income loss. From that shock, he projects roughly $330 billion in consumer-credit losses plus $227 billion in mortgage losses, a combined total of around $557 billion in defaults.
Those losses wouldn’t hit all banks equally. Hayes argues the shock could wipe out about 13% of U.S. commercial bank equity, with regional banks taking the biggest hit. Regional lenders tend to have heavier exposure to consumer lending and residential mortgages—exactly the categories that deteriorate when middle-class professionals stop making payments.
Hayes also connects this risk to crypto markets. He points to Bitcoin’s recent divergence from the Nasdaq 100 and the coin’s drop from a reported high near $126,000 down toward $60,000 as evidence the market is already pricing in tighter fiat credit conditions and potential stress ahead. In his scenario, failing regional banks would force a Federal Reserve response of very aggressive monetary easing; that stimulus, Hayes suggests, would ultimately send Bitcoin and selected crypto tokens well beyond previous highs.
Why this argument merits scrutiny
Hayes has a history of bold macro calls, and he’s explicit that his numbers are projections, not certainties. The timing of AI-driven job losses is crucial: 20% displacement over a short period (months to a couple of years) looks like a systemic shock; the same displacement spread over a decade looks more like a gradual transition that markets and lenders can absorb. AI adoption rates, sectoral impacts, and how quickly firms cut roles are all highly uncertain.
History shows regional banks can fail quickly when conditions shift faster than risk models anticipate. The 2023 collapse of Silicon Valley Bank wasn’t caused by exotic derivatives but by a straightforward duration mismatch and poor hedging—an example of how traditional banking vulnerabilities can surface rapidly.
Bottom line
Hayes presents a clear, numerically grounded worst-case scenario in which AI-driven white-collar unemployment triggers large consumer and mortgage losses, disproportionately harming regional banks and ultimately reshaping policy and asset-price responses. The scenario is internally consistent, but it depends heavily on adoption speed, displacement timing, and how quickly lenders, regulators, and policymakers react.
Disclosure: This article was edited by the Editorial Team. For more information on how content is created and reviewed, see our Editorial Policy.