Spending on AI infrastructure is now contributing extra to U.S. gross home product (GDP) progress than the whole shopper economic system, in keeping with new information from the Bureau of Financial Evaluation (BEA). The comparability, which was posted to the social platform X by economist Heather Lengthy on Monday, means that hype might not be the one factor propping up the excessive inventory costs and valuations of synthetic intelligence firms resembling Nvidia and OpenAI.
Right here, “consumption” means shopper spending on items and companies for private use, which historically contributes about 70% of U.S. gross home product. “AI spending” means enterprise funding in software program and data processing tools, together with information heart building, chip purchases, and computing infrastructure.
The numbers present that AI spending contributed 1.05% of whole financial progress within the first half of 2025, after contributing solely between .02% and .03% from 2022 by way of 2024—a four- to five-time progress ratio.
The unhealthy information (for the general economic system) is that shopper spending has fallen dramatically, from contributing 2.6% of GDP progress to only 1.05% in mid-2025. The decline may stem from flagging shopper confidence because the Trump tariffs take maintain and inflation fatigue and financial uncertainty proceed.
Heady instances for AI firms
This has all been excellent news for AI firms and their suppliers.
Michael Cembalest, chairman of market and funding technique at J.P. Morgan, writes in a recent brief that because the look of ChatGPT in November 2022, AI-related shares have accounted for 75% of S&P 500 returns, 80% of earnings progress, and 90% of capital spending. OpenAI is now valued at $300 billion. Anthropic is valued at $183 billion.
The “AI spending” is especially centered on the brand new information facilities that the AI business says are essential to ship next-generation companies to each companies and shoppers. Now, Cembalest says, new information heart spending is eclipsing new workplace building for the primary time. However he additionally factors out that new information facilities are coming below elevated regulatory scrutiny due to their inordinate calls for on the facility grid.
However, the BEA numbers counsel that the economic system is more and more depending on one slender section, whereas a wholesome economic system would present broad company investments throughout sectors. That is likely to be high-quality if AI infrastructure funding lasts, and if shopper confidence recovers.
However buyers sometimes abhor extended capital expenditures, particularly if there’s no signal of that infrastructure resulting in measurable enterprise efficiencies. And to this point, the efficiencies introduced by generative AI and new automation have been spotty.
