Artificial Intelligence Spending Shifts to Business Users as Subsidies End

Investors have poured record sums into artificial intelligence, with global corporate AI investment more than doubling in 2025 and the four biggest U.S. tech firms projected to spend close to US$700 billion on AI infrastructure this year. OpenAI alone reportedly spends around US$60 billion a year on compute against roughly US$13 billion of revenue, creating a gap that was previously absorbed by investors.

In recent months, that gap is closing for downstream users. Companies are now shouldering the full cost of AI usage: Tesla capped staff spending on external AI tools at US$200 a week after large token charges, Uber exhausted its entire 2026 AI coding budget by April, and Microsoft revoked developers’ Claude Code licences months after issuing them. Meta is trimming thousands of roles partly to fund an AI capital programme exceeding US$115 billion. The shift from “token‑maxxing” to efficiency reflects a broader move to usage‑based pricing, where every prompt and autonomous agent incurs a real, metered cost that businesses must absorb.