OpenAI is generating roughly $25 billion in annualized revenue and is on pace to lose about $14 billion in 2026. That's the short answer. The longer answer is that the losses get much bigger before they get smaller.
OpenAI's run-rate has been essentially flat near $25 billion since February 2026 after an 18-month sprint from $6 billion, even as Q1 2026 revenue alone came in at $5.7 billion. Gross margin improved to 39%, up from 33% a year earlier, but that improvement is being swamped by compute and talent spending that's growing just as fast as revenue. The company's own internal projections, shared with investors as part of its funding process, show cumulative losses reaching $115 billion through 2029 before the business turns cash-flow positive around 2030.
OpenAI Revenue 2026: The $25B ARR Number, Explained
OpenAI's annualized revenue run-rate was approximately $25 billion as of mid-2026, calculated by taking the most recent month's bookings and multiplying by twelve. That figure grew from roughly $6 billion in early 2025 — a more than fourfold increase in under a year and a half — but the growth curve flattened out this spring. The run-rate held near $25 billion from February through the summer even as Q1 2026 revenue alone hit $5.7 billion, which on its own would annualize closer to $23 billion, suggesting the pace of new bookings has cooled from its 2025 trajectory.
Two consumer numbers explain where the revenue is coming from: ChatGPT crossed 1 billion monthly active users in June 2026, with roughly 900 million weekly actives, and enterprise and API revenue continues to scale alongside the consumer subscription base. For context on how the broader private AI market is being priced against revenue like this, see our AI Valuations dashboard.
The $14 Billion Loss: Where OpenAI's Money Is Actually Going
OpenAI is projected to lose approximately $14 billion in 2026 against total spending of roughly $22 billion — meaning the company is spending nearly a dollar for every dollar it loses on top of what it earns. Q1 2026 alone produced a $3.7 billion cash burn, or roughly 65% of that quarter's revenue consumed by costs. Two line items dominate the spend: compute and people.
| Cost category | 2025 | 2026 (est.) | Notes |
|---|---|---|---|
| Inference / compute costs | $8.4B | $14.1B | Cost of serving deployed models to users |
| Azure compute commitment | — | ~$13B | Contracted spend with Microsoft |
| Talent costs | ~$3B | ~$4B | ~4,500 employees, ~$900K fully-loaded each |
| Total spending | ~$9B | ~$22B | Includes R&D, sales, and overhead |
| Total revenue | $13B | $13–25B | Range reflects differing full-year vs run-rate measures |
| Net loss | ~$5B | ~$14B | OpenAI's own internal forecast |
Figures are 2026 estimates blended from The Information, futuresearch.ai financial forecasts, RD World, and OpenAI investor disclosures reported by multiple outlets in mid-2026. Revenue ranges reflect the gap between annualized run-rate and full-year booked revenue.
OpenAI Profitability Timeline: Why 2030 Is the Target, Not 2027
OpenAI's own internal projections, shared with investors during its most recent fundraising, show cumulative losses reaching $115 billion through 2029 — with 2028 alone expected to produce roughly $74 billion in operating losses as data center and chip commitments ramp. The company doesn't project a swing to profitability until around 2030, when revenue is forecast to scale well past $100 billion as enterprise contracts, API volume, and consumer subscriptions all mature simultaneously.
The scale of that bet is why OpenAI has been raising capital aggressively even at a run-rate most software companies would consider a massive success. The company's confidential S-1 filing on June 8, 2026 targeted a $1 trillion valuation, though reporting as of late June suggests OpenAI is now leaning toward delaying any public listing to 2027 rather than debuting in September 2026 as earlier reported. A revenue multiple north of 40x on $25 billion in ARR only makes sense if investors believe the 2030 profitability story — otherwise, the math on a trillion-dollar valuation against $14 billion in annual losses doesn't hold up on fundamentals alone.
OpenAI vs. Anthropic: The Revenue Lead Just Changed Hands
The most important competitive shift in OpenAI's 2026 revenue story isn't the flattening run-rate — it's that Anthropic overtook it. Anthropic crossed roughly $30 billion in annualized revenue in April 2026, just two months after OpenAI hit its own $25 billion mark, and has since been reported near a $47 billion run-rate by July 2026. That's the first time a frontier-AI challenger has passed the category incumbent on reported revenue.
The gap comes down to mix. Roughly 80% of Anthropic's revenue comes from business customers buying Claude through the API, while OpenAI's revenue leans more heavily on ChatGPT's consumer subscription base — a larger audience, but one that monetizes at a lower rate per user than enterprise API contracts. Anthropic has also reportedly trained its models at roughly a quarter of OpenAI's compute cost, which is starting to show up in relative burn rates even as both companies post multi-billion-dollar annual losses. For a full breakdown of how the two companies are being priced against these numbers, see our OpenAI vs. Anthropic vs. Google valuation comparison.
Where OpenAI's $25B in Revenue Actually Comes From
OpenAI doesn't break out a clean revenue split by product line, but the shape of the business is visible from the pieces that have leaked or been confirmed. ChatGPT Plus, Team, and Pro subscriptions remain the largest single bucket, riding on top of 1 billion monthly active users as of June 2026 — a scale no other consumer AI product has reached. Enterprise contracts and the developer API make up the rest, and that split matters because subscription revenue from hundreds of millions of individual consumers is inherently lower-margin and higher-churn than a handful of seven- and eight-figure enterprise agreements.
That consumer weighting is also the simplest explanation for why OpenAI's run-rate growth flattened this spring while Anthropic's kept climbing. Consumer subscription growth tends to saturate faster than enterprise land-and-expand motions, especially once a product has already crossed a billion monthly users — there are only so many net-new subscribers left to add. Enterprise AI spend, by contrast, is still in its early innings, which is part of why Anthropic's 80%-enterprise mix has been compounding even as OpenAI's consumer base plateaus in growth rate, if not in absolute size.
OpenAI has been responding by pushing harder into enterprise itself — deeper Microsoft 365 Copilot integration, expanded API tiers, and new agentic tooling aimed at replacing junior analyst and developer workflows. Whether that shift shows up in the next few quarters of ARR data, or whether OpenAI's brand strength stays concentrated in the consumer market it built ChatGPT's reputation on, is arguably the single most important open question for anyone underwriting the company's 2029-2030 profitability targets.
What OpenAI's Revenue Numbers Mean for Investors
The uncomfortable truth in OpenAI's 2026 numbers is that the company's revenue growth has decelerated at exactly the moment its spending has accelerated. A business that quadrupled its ARR in 18 months and then went flat for five straight months is sending a different signal than one still compounding — even if $25 billion in annualized revenue remains, on an absolute basis, one of the fastest scale-ups in software history. Anyone tracking this alongside broader hyperscaler AI capex should cross-reference our Big Tech Earnings dashboard, since Microsoft's Azure commitments to OpenAI are effectively a shared line item between the two companies' income statements.
For VCs and LPs modeling exposure to OpenAI through secondary markets, SPVs, or funds like RVI, the profitability timeline matters more than the headline ARR number. A $1 trillion valuation on $25 billion of revenue and $14 billion of annual losses is a bet on the 2029-2030 inflection actually arriving on schedule — not a bet on today's financials. That's a fundamentally different risk profile than most late-stage software investments, and it's worth underwriting explicitly rather than assuming the growth-at-all-costs playbook that worked for SaaS in the 2010s applies cleanly to frontier AI labs burning double-digit billions a year on compute.
The Bottom Line
OpenAI's $25 billion ARR is real and enormous by any historical software standard, but the $14 billion loss next to it — and the $115 billion in cumulative losses projected through 2029 — is the more important number for anyone pricing the company today. Profitability isn't expected until around 2030, Anthropic has already taken the revenue lead on a smaller compute budget, and the IPO that was supposed to anchor a $1 trillion valuation this September now looks more likely to slip into 2027. The revenue growth story is intact; the profitability story is still years away from being tested.
Track AI company valuations against real revenue multiples on the AI Valuations dashboard and hyperscaler capex on the Big Tech Earnings dashboard at Value Add VC.
Get VC data most people never see
— 100% free
Weekly benchmarks, valuations, and fund data. Join 5,000+ investors. No spam.