NEW YORK — Global AI investment has surged to an estimated $1.6 trillion since 2013, as companies and governments worldwide race to deploy artificial intelligence at scale. Economists now expect 2025 spending on AI-focused infrastructure to exceed the inflation-adjusted cost of NASA’s Apollo moon program, reflecting how the scramble for computing power has turned data centers and chips into the new strategic industrial base, Dec. 11, 2025.
The $1.6 trillion total comes from a new Reuters analysis of Stanford University’s AI Index data, which tracks AI investment in startups, public markets, deals and corporate capital expenditure since 2013. The same work estimates another $375 billion of AI investment will be deployed in 2025, meaning a single year of spending on AI infrastructure and companies will surpass the roughly $298 billion cost of Apollo in today’s dollars.
So far, most of that AI investment is not going into apps but into the physical backbone of the AI era: data centers, power generation, chips and networking gear. Stanford’s 2025 AI Index report estimates companies spent about $37 billion globally on AI infrastructure in 2024 alone, while U.S. private AI investment climbed to $109.1 billion that year, nearly 12 times China’s total.
In the United States, that torrent of AI investment is already reshaping the macroeconomy. Research from S&P Global finds that data center and related AI spending accounted for 80 percent of the growth in private domestic demand in the first half of 2025, a shift that has made server farms and transmission lines as important to growth as factories were in earlier eras.
From early AI investment bets to a global wave
A decade ago, AI investment looked modest by today’s standards. In 2017, Wired chronicled Google’s launch of a dedicated AI venture fund as an early bet that machine-learning “plumbing” would pay off.
By 2020, a Brookings Institution analysis of private-market AI investment found U.S. firms attracting about $25 billion in disclosed deals in 2019 and concluded that AI was a “global wave, not a bipolar contest,” as capital spread rapidly beyond the U.S.–China rivalry.
Bubble risk or new infrastructure age?
The magnitude and speed of AI investment are reviving old questions about bubbles. Economists note that earlier waves — from 19th-century railways to telecoms in the dot-com era — left behind both lasting infrastructure and painful busts, and recent reporting has highlighted how much of today’s buildout is concentrated in a handful of tech giants whose circular funding deals can blur the line between genuine demand and financial engineering.
Yet many forecasters argue the buildout is justified. One recent analysis of corporate capital spending estimates that the biggest U.S. cloud providers alone will plow around $300 billion into AI-centric infrastructure in 2025, rising toward nearly $500 billion in 2026 — roughly equivalent, the author argues, to rebuilding the Apollo program about once a year.
At the same time, skeptics point out that revenues have not yet caught up with the scale of AI investment. Recent research from economists and management consultancies suggests only a small share of firms deploying generative AI report clear productivity gains so far, even as AI-driven capital expenditure is responsible for an outsized share of recent U.S. growth.
Whether 2025 marks the peak of an AI investment bubble or the midpoint of a decades-long infrastructure cycle, the comparison with Apollo captures the stakes. In the 1960s, a moonshot reshaped science, industry and American identity; this time the bet is that pouring trillions into data centers and algorithms will rewire the global economy itself. The question is no longer whether AI investment is big — it is whether the returns will match the bill.
