OpenAI's Ledger Spooks Wall Street: $278B Burn by 2030, Yet the Valuation Asks for $1.2 Trillion
The most expensive wager in business history: a $278 billion stake, with the win condition being revenue growing tenfold in four years.
Three key facts
On September 18–19, the Financial Times disclosed OpenAI's internal financial projections: cumulative cash burn of roughly $278 billion from 2026 through 2030, total compute and infrastructure spending near $856 billion over the same stretch, and no profit before 2030. Multiple financial outlets (PYMNTS, Seeking Alpha, Investing.com) cross-confirmed the figures.
The revenue target is equally startling: from a projected $36 billion in 2026 to $350 billion in 2030 — tenfold in four years, with cumulative 2026–2030 revenue around $840 billion. In plain terms, OpenAI's own model says the hole gets filled only after it grows into a revenue scale never before seen in commercial history.
The fundraising is already in motion: OpenAI is seeking a new round targeting a valuation of roughly $1.2 trillion; the March round had just lifted it to $852 billion. That's another 40% in six months — propped up by a projection sheet showing sustained, massive losses through 2030.
Wang Dou's Take
Put the two numbers side by side and the story comes into focus: $852 billion valuation in March, $1.2 trillion now; disclosed burn of $278 billion in cash by 2030. This isn't a "growth story" — it's an open-handed bet. Investors aren't buying today's OpenAI; they're buying the fact that it can flash its worst ledger and still get a price bump. The market logic has mutated: the heavier the burn, the deeper the moat looks, the higher the valuation climbs.
What does tenfold revenue in four years require? $36 billion to $350 billion means ChatGPT must graduate from the most profitable consumer subscription into utility-grade enterprise infrastructure — every large company, every government, every device on Earth paying tolls to OpenAI. That script isn't impossible, but it presumes every competitor falls behind. The reality: Anthropic is grappling it in the enterprise market, Google stuffs models free into a billion devices, and the open-weight camp builds comparable models for $20 million. "Monopoly-grade growth" is precisely the assumption that should never appear on a projection sheet.
The comparison to remember sits next door: Nscale lost $1 billion in half a year and got pinned under the spotlight by its own S-1; OpenAI burns $278 billion over four years and still lifts its valuation 40%. The difference isn't the size of the number — it's narrative power. Whoever owns the story of "the only road to AGI" gets to rename losses as investment. 2030 will grade both ledgers on the same day; until then, everyone can only watch the table and keep raising.
Source: PYMNTS · Seeking Alpha · Business Times (via FT)