• The IMF cautions that a sharp reversal in AI investment could leave global GDP 1.1% below baseline by 2028, with U.S. output 1.8% lower.
  • AI-related goods trade has surged 29% since 2023, but an earnings disappointment could trigger a leveraged selloff hitting Asia disproportionately.
  • Policymakers urged to prepare for financial stability risks as AI spending supports near-term growth but long-term returns remain uncertain.

IMF Flags AI Investment Reversal as Key Downside Risk

The International Monetary Fund is sounding the alarm on a potential AI investment bust that could ripple through the global economy, warning that a sharp pullback in spending on artificial intelligence infrastructure could leave output meaningfully below baseline by 2028. According to people familiar with the fund’s latest surveillance work, the downside scenario envisions global GDP 1.1% lower than its baseline projection, including a 1.8% shortfall in U.S. output. That’s not a forecast of recession, but a stark reminder of how much the current expansion leans on the AI buildout.

In its October 13 update, the IMF projected global growth of 3.1% in 2026 and 3.4% in 2027, slightly upgrading its 2026 view from July partly because strong AI infrastructure investment is offsetting drag from higher energy prices. The AI bust remains a separate, darker scenario—one that hinges on whether profits and productivity can eventually justify the massive capital outlays now underway.

The $10 Trillion Question

MarketWatch (NWSA) attributes a staggering $10.3 trillion estimate for AI infrastructure and specialized chip investment over 2025–2032 to a Brookings paper by economist Stijn van Nieuwerburgh. That’s equivalent to an average 3.63% of U.S. GDP annually, underscoring the scale of the wager. It’s not a commitment from any single company or government, but a rough measure of the buildout’s magnitude—and a hint of what could unwind if expected returns fail to materialize.

The IMF’s concern is not that AI lacks economic value, but that spending is boosting demand today while the productivity gains needed to service that investment may arrive with a lag. Managing Director Kristalina Georgieva warned on October 16 that an earnings disappointment could trigger a correction spreading through a leveraged, interconnected AI ecosystem and across countries holding AI-related assets. She also stressed the upside: successful deployment could lift global growth by around half a percentage point annually over the next decade.

Asia’s Export Engine at Risk

The fund’s Asia-Pacific outlook, released October 15, projected regional growth slowing from 5.0% in 2025 to 4.8% in 2026 and 4.4% in 2027. An AI investment or valuation correction would hit Asia disproportionately because of its central position in technology supply chains. South Korea and Taiwan have benefited from semiconductor demand, while several Southeast Asian economies have gained from data-center investment. A reversal would affect exports, industrial investment and employment—not just financial portfolios.

The trade dependence is already significant. IMF figures reported by Reuters (TRI) show trade in AI goods grew 29% between 2023 and 2025, versus 8.3% for total global trade. Without AI-related goods, overall trade growth would have been just 4.6%. That concentration cuts both ways: it has powered growth, but it also means a sector-specific disappointment could now have wider consequences.

Transmission Channels and Policy Response

The IMF describes several reinforcing channels through which an AI bust could spread. First, developers and infrastructure operators scale back spending when expected returns disappoint. Second, equity valuations decline, debt becomes harder to service, and anticipated revenue commitments may fail to materialize. Third, falling asset values reduce household wealth and spending. Fourth, lower demand for chips, computing equipment and related infrastructure reaches exporting economies.

Policymakers are being urged not to become complacent. The IMF called for heightened financial supervision around AI valuations, financing structures and debt-servicing risks. Reuters also reported that U.S. regulators are questioning banks about advanced AI use, while European regulators are pressing banks to strengthen recovery plans. Georgieva separately warned that cyberattacks affecting payments, trading or access to deposits could create immediate losses and undermine confidence—a distinct AI-related financial-stability risk.

A Two-Sided Outlook

The debate at the IMF–World Bank meetings in Washington is whether AI can generate sufficient productivity and revenue quickly enough to justify the investment, and how its gains and disruptions will be distributed. IMF Deputy Research Director Petya Koeva Brooks told Reuters that even the downside scenario would not likely push growth to levels consistent with a global recession. Still, the warning comes amid elevated public debt, rising bond yields and higher energy prices, which can weaken demand and raise financing costs at the same time an AI correction reduces investment.

The IMF is treating these as overlapping vulnerabilities rather than isolated threats. For now, AI spending remains a pillar of global demand. But the fund’s message is clear: the longer the buildout runs ahead of proven returns, the more painful the adjustment could be if confidence cracks.

Correction: An earlier version of this article misstated the IMF’s downside scenario as a 1.3 percentage point reduction in annual growth by 2028. The fund’s projection is a 1.1% shortfall in the level of global GDP relative to baseline, not a percentage-point change in growth.