IMF: AI Infrastructure Spending Is Holding Global Growth at 3%

IMF chief Kristalina Georgieva said on Monday in Bangkok, at the start of the IMF and World Bank annual meetings, that spending on AI infrastructure is helping the global economy grow by about 3% a year. She noted that this is happening amid high energy prices, Reuters reports. For an institution that usually talks about interest rates, debt and trade barriers, it is a telling shift in emphasis: data centers and chips now appear in the global growth conversation alongside the classic factors.

What exactly Georgieva said

The thesis is short. The global economy is growing at about 3% a year, and the investment cycle around AI provides a significant part of that support. Energy remains expensive, which means growth is not running on cheap fuel but in spite of costly fuel. In the excerpt available at the time of publication, Georgieva also talks about Asian economies. The full context of that part of her speech has not yet been disclosed, so we do not speculate about the details.

The 3% figure is modest. By IMF estimates, average global growth in the two decades before the pandemic was noticeably higher, at about 3.8%. In other words, AI investment is not accelerating the economy; it is helping keep it from sliding lower.

Why AI infrastructure works as an economic stimulus

Building compute capacity is not just a matter of buying accelerators. It takes land, concrete, networking, cooling systems, electrical equipment, contractors and installation. That spending creates demand in dozens of adjacent industries, and the effect shows up before the models themselves start generating revenue.

Asia plays a special role here, and it is unlikely to be a coincidence that the meetings are being held in Bangkok. A large part of the supply chain for AI hardware is in the region: production of chips, memory, servers and components. The growth in capital expenditure by the tech giants turns into export orders and investment for Asian economies. This is our own analytical interpretation, not a direct quote from Georgieva.

Energy as the weak spot

High energy prices hit AI infrastructure directly. Data centers consume a lot of electricity, and in some regions grid connections are constrained by lead times and capacity. The more a kilowatt-hour costs, the higher the cost of compute and the greater the pressure on the margins of operators and cloud providers.

There is a second risk. If a noticeable share of growth rests on a single investment cycle, any cooling in it is felt immediately at the macro level. The IMF has warned before that AI company valuations may turn out to be inflated. The current statement highlights the flip side: the economy has become dependent on whether companies keep spending.

Key parameters of the statement:

• Global growth: about 3% a year
• Driver: spending on AI infrastructure
• Backdrop: high energy prices
• Venue: IMF and World Bank annual meetings, Bangkok
• Regional focus: Asian economies
• Source: Reuters, speech by Kristalina Georgieva

What to watch next

For investors and operators, the nearest benchmark is the IMF's updated forecast and how the fund assesses the risks of growth being concentrated around AI. If its reports include separate calculations of the contribution of AI capex, that will be a signal: macroeconomists have recognized the industry as a standalone factor.

For companies planning their own compute capacity, the main takeaway is practical. The cost of energy and access to the grid are becoming a risk item on par with the price of accelerators. Electricity contracts, choice of region and energy efficiency should be built into the calculations from the very start of a project.

For developers and startups, the effect is indirect but tangible. As long as capital spending continues, compute stays affordable and cloud providers compete for customers. If the cycle turns, inference prices and GPU rental rates will have to be revisited. It is worth budgeting for both scenarios.