SAN FRANCISCO — Big Tech is preparing to pour more than $300 billion into artificial-intelligence data centers, chips and power infrastructure in 2025, a spending wave that now rivals or exceeds the inflation-adjusted cost of the Apollo moon program and is reshaping global capital markets. Investors and regulators warn that the same firehose of AI investment driving the boom could also inflate a fragile bubble, Dec. 8, 2025.
AI investment passes Apollo-era scale and keeps climbing
The scale of AI investment has accelerated sharply. Consultancy McKinsey estimates hyperscale cloud providers alone expect to spend about $300 billion on capital expenditures in 2025, largely on AI chips and data centers, while IoT Analytics puts 2024 data center infrastructure spending at roughly $290 billion and says Big Tech is set to lift that by more than 40 percent this year — pushing AI-related outlays comfortably above the $300 billion mark.
An FT analysis says Amazon, Microsoft, Alphabet and Meta are lining up more than $300 billion of AI-focused spending for 2025 after a 63 percent jump in 2024 capex, with Amazon alone expected to top $100 billion. Taken together with commitments from Oracle, Nvidia and leading AI labs, some forecasts now see total Big Tech AI investment for the year approaching $400 billion.
A single year of corporate AI buildout is therefore on track to overtake the entire Apollo moon program, which cost the U.S. government about $25.8 billion between 1960 and 1973 — roughly $260 billion to $320 billion in today’s dollars, depending on how related lunar projects are counted. The cost of the Apollo moon program once set the benchmark for national ambition; in 2025, private AI investment may surpass it.
Only a few years ago, these numbers looked unimaginable. In 2018, Google, Microsoft and Amazon together spent nearly $53 billion on capital expenses for data centers, a sum Reuters cast as evidence that cloud computing had entered a new phase of scale. Data compiled by McKinsey show private AI funding crossing about $70 billion in 2019, and a 2023 analysis from Goldman Sachs projected global private AI investment could approach $200 billion by 2025.
Reality is already blowing past that forecast. Newer surveys suggest total corporate AI investment topped roughly $250 billion in 2024 and is still climbing, much of it financed with cheap debt. Reuters graphics describe an “unusually tight financial loop” in which firms such as Nvidia, Oracle and OpenAI buy from, sell to and fund one another’s AI projects, while a Goldman “AI: In a Bubble” report and IBM chief executive Arvind Krishna warn that hundreds of billions in annual outlays and up to $8 trillion in data center spending may never fully earn their cost of capital.
Supporters counter that past technology waves also looked reckless before productivity gains arrived. Goldman economists say sustained AI investment could eventually lift U.S. labor productivity by roughly one percentage point a year and, at its peak, amount to 2 percent to 4 percent of GDP in leading economies. Meanwhile, Reuters Breakingviews says AI-linked loans and bond issuance tied to data centers have nearly doubled this year to around $700 billion, while S&P Global notes that data center and power-plant construction is now big enough to move growth figures — leaving markets to decide whether this historic AI investment boom becomes the next internet or the next dot-com bust.
