全球经济似乎难以抵御人工智能泡沫破裂The Global Economy Looks Vulnerable to an AI Bust

TOKYO—A message from the violent swings in stock markets over the past week: The fallout from an artificial-intelligence bust wouldn’t stop at the U.S.’s shores.

The AI build-out has been an engine of growth not just for the U.S. this year but for the whole global economy. Squeezed by the energy crunch caused by the Iran war, the world has relied heavily on U.S. data-center investment to power otherwise lackluster growth.

Global trade has thrived as Asian economies such as China, Taiwan and South Korea have enjoyed booming exports of computer chips, server racks and electronic equipment to feed ravenous demand for computing power in the U.S.

The Silicon Valley bounty is spreading as other economies join the party. Assembly hubs such as Vietnam and Mexico are seeing exports take off. Malaysia, Thailand and Australia are building data centers of their own. Europe’s industrial giants are supplying the power and cooling systems that make the AI revolution possible.

Yet the anxiety that spread through global markets in the aftermath of the U.S. midterm elections shows this growing dependency on Silicon Valley has a downside.

A slowdown in AI investment would punish exports in economies that in many cases show few other signs of growth. A sustained stock slide would crush mom-and-pop investors all over the world who have piled into technology stocks in anticipation of bumper returns. Overstretched budgets mean governments have limited room to respond with more spending.

Economists calculating what might happen if the pace of investment slows or the AI boom turns to full-blown bust say the only real question is how severe the global consequences will be.

“There is very little that can be done to prevent a global recession should the AI boom collapse,” said Frederic Neumann, chief Asia economist at HSBC in Hong Kong.

Stock markets have been on a wild ride since the Democrats’ victory in the U.S. midterm elections sparked concerns that Congress will hit the brakes on data-center development.

The latest upset in the industry has been the alleged spying scandal engulfing Nvidia, the linchpin of the AI ecosystem because of its deep pockets and advanced chips.

Stocks have clawed back some losses in recent days, moves that suggest many investors think the selloff is overdone. Even so, key stocks and indexes remain well below recent peaks. The tech-heavy Nasdaq index closed Tuesday 10% below its year-to-date high. Japan’s Nikkei index is down 16%, and South Korea’s Kospi, home to major memory-chip maker Samsung Electronics and SK Hynix, is down 42%.

For now, the stock-market gyrations haven’t spilled over into real economic activity, with global data pointing to the AI boom still going strong.

Asian economies are still seeing bumper export growth, with many of the gains driven by surging prices for chips and other hardware. China’s exports were up 35% year-over-year in October, according to figures from data provider CEIC. South Korea’s were up 93%, with Japan, Taiwan, Singapore and others also posting double-digit gains.

Economists at Capital Economics said the export bonanza means global trade is on track for its best spell of annual growth this year since the days of hyperglobalization in the early 1990s, excluding the big rebounds after Covid-19 and the 2008-09 financial crisis.

But the consulting firm warned that as much as 80% of recent export growth has been concentrated in AI-related goods, leaving global trade vulnerable to any change in U.S. spending on data centers. China, South Korea and other major economies are beset by weak consumption, meaning a drop-off in trade would have an outsize hit on growth.

Political and regulatory hurdles could drag on data-center investment. So, too, could rising interest rates or disappointments on earnings if AI firms’ revenues fail to keep up with outlays. The AI splurge already eclipses past infrastructure booms such as the railways, and even without external pressure it might struggle to maintain that breakneck rate of growth.

As data-center investment slows, “AI capex will turn from a major prop to a drag on world trade,” Marcel Thieliant, head of Asia-Pacific at Capital Economics, said in a report.

Some economists have looked back to the dot-com bubble of the early 2000s as a yardstick to measure the possible costs to the global economy of an AI blowup.

In that episode, U.S. business investment sank and stocks crashed, hurting U.S. consumer spending and crushing imports.

As a share of gross domestic product, U.S. imports of IT-related equipment fell 55% in two years from their mid-2000 peak, according to Oxford Economics—a drop that hit Asian and other economies that were suppliers of the goods. Global trade plummeted, and world growth in 2001 collapsed to just 2%, consistent with a global recession.

The globalized financing of the AI boom would in a bust spread financial pain around the world, adding to the damage from a steep fall in trade. Gaurav Ganguly, head of international economics at Moody’s Analytics in London, said the world would struggle to absorb a sharp AI stock slide on top of an oil shock, tariffs and rising interest rates.

“The result could be a severe economic shock with global recessionary consequences,” he said.