Too big to pause: Could an AI slowdown crash the economy?

Calls for a slowdown in AI development are clashing with a $800 billion investment boom in the United States. Could an AI slowdown devastate the US economy? Artificial intelligence is a powerful industry in the US, with leaders and officials having completely opposite views. While top AI industry leaders are urging a slowdown, President Donald Trump aims to outdo China, announcing a "Super Intelligence Force" led by former SEC boss Jay Clayton.

The five major US-based AI hyperscalers plan to invest $800 billion in the AI buildout this year, according to Goldman Sachs.

US stock market and GDP growth increasingly depend on the AI industry's health. A slowdown could tank the economy, causing companies to cancel infrastructure plans, investors to reprice AI assets, and lenders to withdraw financing. An IMF estimate suggests a 20% decline in US equity markets, a 1.5% drop in US GDP, and a 1.2% reduction in global output. Fitch, however, is even more pessimistic, projecting a US recession due to a 35% equity shock and capital expenditure retrenchment.

On the other hand, some argue that a halt in AI development might not lead to economic catastrophe. Existing models could still bring productivity gains despite a complete halt in training new models. AI adoption is lagging in many industries, and even within software engineering, it is lagging across different types of businesses and organizations. Even if AI development slows down, there are still opportunities for growth in inference demand.

The IMF also warned that weaker AI-productivity expectations could lead to reduced investment, a financial market correction, and a decline in household wealth. Thus, while a slowdown could have an impact on the economy, it might not necessarily result in a disaster.