NEW YORK — Global AI investment has climbed to nearly $1.6 trillion since 2013, and forecasters expect 2025 to add another $375 billion—enough for one year of spending to top the inflation-adjusted price tag of the Apollo moonshot, Jan. 8, 2026.
The surge is being fueled less by consumer apps than by an infrastructure race for chips, data centers and power, leaving investors to grapple with a familiar question: Is this a durable platform shift, or the early shape of a bubble?
Why AI investment is breaking historical benchmarks
A recent Reuters investment tally compares today’s AI investment wave with past mega-buildouts, arguing the current boom has already surpassed landmark government projects in scale. The comparison matters because it reframes AI as capital-intensive industrial expansion, not just software.
On the space-race side of the ledger, the U.S. spent $25.8 billion on Apollo between 1960 and 1973—about $257 billion in 2020 dollars, according to The Planetary Society’s inflation-adjusted estimate. Even allowing for differences in methodology, 2025’s projected outlays place AI investment on pace to exceed Apollo’s total cost in roughly a single year.
Where the money is going in the AI investment boom
Much of the spending is landing in physical capacity: data center shells, GPUs and networking gear, plus grid upgrades and long-term power contracts. McKinsey estimates data centers will require trillions in global capital by 2030 to keep pace with compute demand in its report on the cost of compute and the race to scale data centers.
On the demand side, the software story is also ballooning. Gartner forecasts worldwide generative AI spending will reach $644 billion in 2025, driven largely by hardware embedded in devices and servers—another signal that AI investment is increasingly a capex-and-supply-chain story.
Bubble risks escalate as financing and returns diverge
The risk isn’t simply “too much money,” but how it’s being deployed. With buildouts measured in years and payback dependent on enterprise adoption, some analysts worry that incentives can turn into a prisoners’ dilemma—spend because competitors spend. Harvard Business School’s Andy Wu told the Harvard Gazette the danger hinges on how much leverage and vendor risk piles up before revenue catches up.
That tension is visible in the industry’s circular deal-making. In the Reuters analysis, University of Minnesota researcher Andrew Odlyzko warned that “the circular deals … are like a warning sign” for whether underlying customer demand is strong enough to justify the buildout.
Continuity: this didn’t start in 2025
Earlier data points show how fast the curve steepened. In 2023, funding for generative AI surged even as overall AI investment cooled, according to Our World in Data’s analysis of generative AI funding. In the broader startup market, U.S. venture funding fell sharply in 2023 despite the hype, Reuters reported in January 2024. By 2024, the rebound was increasingly AI-led, with AI startups capturing a record share of U.S. venture dollars, Reuters reported in January 2025.
For 2026, the watch list is getting clearer: power and permitting bottlenecks, rising debt used to fund capacity, and—most critically—whether companies can translate AI investment into measurable productivity and revenue before investors lose patience.
