OpenAI Spent $34 Billion in a Single Year

OpenAI spent roughly $34 billion over the past year, a sharp jump from the year before, even as the company pulled in only about $13 billion in revenue. Around $19 billion of that spending went into research and development and another $6 billion or so into sales and marketing. The figures come from financial analysis by independent journalist Ed Zitron and were independently confirmed by the Financial Times on June 16, 2026 — rare hard numbers for a private company that discloses very little.

The headline loss looks brutal: a net loss of about $39 billion. But roughly $30 billion of that is a one-time, non-cash accounting charge tied to earlier changes in the company's corporate structure, not money actually going out the door. Strip it out and the adjusted operating loss is closer to $8 billion — still enormous, but a very different picture from "$39 billion up in smoke." Revenue, meanwhile, is climbing fast: OpenAI is now making about $2 billion a month, up from roughly $1 billion a quarter at the end of 2024.

Spending far more than it earns

Even with the accounting noise removed, the shape of the business is clear: OpenAI is spending well over twice what it brings in, and pouring the difference into compute and research in a race to stay ahead. This is the classic land-grab strategy — burn investor money now to lock in scale and users, profit later. It works only as long as investors keep writing checks, which is why the company is reportedly preparing an IPO that could value it at more than $1 trillion. That IPO is not a victory lap; it is the next funding round the spending requires.

Why it matters to you

These numbers explain a lot about the product you use. The generous free tier, the heavily discounted plans, the constant new features — much of it is subsidized by investors, not paid for by what users actually spend. That is great while it lasts, but it also means the terms can shift the moment the math has to work: usage caps tighten, prices rise, free features move behind a paywall. You have already seen smaller versions of this across the industry. A company losing billions cannot subsidize you forever.

What strikes me most is the gap between the panic-inducing $39 billion loss and the more mundane $8 billion operating one. Both are true, and which one gets quoted tells you who is trying to scare you and who is trying to reassure you. The sober takeaway is simpler: AI at this scale is still a money-losing business held up by the expectation of future dominance, and you should plan your own dependence on it accordingly.