Live global tracker · 8 companies · updated June 2026
Combined AI spending by the world’s biggest technology companies since 2023
OpenAI, Anthropic, xAI, Meta, Google, Microsoft, Amazon and Nvidia are pouring cash into compute at a pace never seen before. Cold capital in, gigawatts of heat out — counting live.
The spenders
Ranked by total AI spend since 2023. Together they now burn a combined $1.1T every year. Each figure is sourced — tap a link to check the receipts.
spent on AI since 2023
Q1 FY2027 data-center revenue hit $75.2B, up 92% YoY — a ~$301B annualized run-rate of AI hardware.
spent on AI since 2023
Amazon guided 2026 capex to ~$200B — "predominantly AWS" and "most of it is AI," per Andy Jassy.
spent on AI since 2023
Alphabet raised 2026 capex guidance to $180–190B — roughly double the $91.4B spent in 2025.
spent on AI since 2023
Record $37.5B capex in Q2 FY2026 put Microsoft on a ~$150B annualized run-rate.
spent on AI since 2023
Meta raised 2026 capex guidance to $125–145B — nearly double the ~$72B spent in 2025.
spent on AI since 2023
2025 costs hit ~$34B against ~$13B revenue — a ~$21B operating loss, ramping toward ~$27B cash burn in 2026.
spent on AI since 2023
AI-segment capex hit a $30.8B annualized rate in 2026 — $7.7B in Q1 alone (per SpaceX’s S-1).
spent on AI since 2023
Revenue run-rate exploded from ~$9B (end 2025) to ~$47B by May 2026; 2026 compute spend ~$19B.
Sense of scale
The numbers are too large to feel. Here is the same spending, translated.
Roughly a brand-new billion-dollar data center every eight hours, weekends included.
Split the bill across all 8.1 billion people on Earth and everyone chips in this much annually.
The AI build-out now spends more each year than dozens of national space programs combined.
Every single year, in inflation-adjusted terms — the cost of putting humans on the Moon, several times over.
Methodology
Estimating private-company spend is inherently approximate. We show our work, cite every figure, and separate confirmed spending from future commitments.
Last updated · June 2026
Questions
The eight largest players tracked here — OpenAI, Anthropic, xAI, Meta, Google, Microsoft, Amazon and Nvidia — are on a combined pace of roughly $1.06 trillion per year, or about $34,000 every second. Most of that is data-center capital expenditure: GPUs, custom chips and gigawatt-scale buildings.
Among the buyers, Google (Alphabet) leads with ~$185B of 2026 AI capex, followed by Amazon (~$170B), Microsoft (~$150B) and Meta (~$135B). Nvidia sits on top of the chart at ~$301B, but as the chip supplier that figure represents AI hardware spend flowing through it, not its own outlay.
OpenAI’s 2025 costs reached about $34 billion against ~$13 billion of revenue, and its cash burn is projected near $27 billion in 2026. Separately, Sam Altman has disclosed roughly $1.4 trillion of multi-year compute commitments — about 30 gigawatts of future data-center capacity.
Every figure is an annualized run-rate drawn from company earnings guidance, SEC filings and reputable reporting (Reuters, CNBC, The Information, Fortune and company press releases). Public companies use their AI/data-center capex; private labs use estimated annual cash spend. Each company card links to its sources.
No. The counters take each company’s verified annual spending rate and advance it second by second, the same way a world-population clock works. They are a live projection of a known annual pace — not a feed of individual bank transfers.