OpenAI or Anthropic reports $1 trillion annual revenue

Will OpenAI or Anthropic report revenue of $1 trillion or more for a fiscal year?

Chance by Q4 2031

13%

Cumulative probability

0%25%50%75%100%
  • Latest forecast
Chance by the end of each quarter
YearQ1Q2Q3Q4
20260%
20270%0%0%0%
2028<1%<1%<1%<1%
2029<1%<1%<1%<1%
20304%4%4%4%
203113%13%13%13%

Forecast history

Forecast made on 15 Oct 2026

Latest of 5 forecasts · Methodology v2.1

Model reasoning

I put the chance at about 13%. Anthropic, the faster-growing lab, will book roughly $56 billion this year. To reach $1 trillion in 2030 it would have to roughly double every year for four years, about five times its own forecast, while computing power, price wars and customer cost-cutting push the other way. OpenAI projects only $350 billion for 2030. Anthropic's quarterly growth re-accelerating through 2027 would most raise this estimate.

Where things stand (mid-October 2026)

Anthropic is the leading candidate. Its leaked S-1 shows FY2024 revenue of $0.39B and FY2025 of $4.6B. Quarterly revenue in 2026: Q1 $4.73B, Q2 $11.5–11.6B, and Q3 $16.8B. The Q3 figure was told to investors on Oct 14 and is up 45% on the quarter. That is still very fast, but well down from about 145% in Q2. The company expects an annualized run rate of about $100B by year-end, which puts FY2026 at roughly $55–58B. In the IPO process it is projecting $190–200B of revenue for 2028. It records cloud-marketplace sales at their gross value, which slightly helps the headline number. The S-1 says computing capacity, not demand, is the binding constraint. It lists about $518B of compute commitments over roughly ten years, about 80% of them non-cancellable.

OpenAI is behind. Its annualized revenue was about $50B at the end of September, counting only its own share of partner sales, and it expects about $70B by year-end. Its own investor model projects $36B in 2026 rising to $350B in 2030, alongside $278B of cash burn.

Reporting is unlikely to be the obstacle. Anthropic is targeting a November 2026 IPO (Polymarket gives about 73–79% for listing by end-2026 and 2.6% for no IPO by end-2027). OpenAI's CFO says it will be public in 2027. Both use calendar fiscal years, so full-year results would arrive in Q1 of the following year. The last fiscal year that can count is FY2030, reported around February 2031.

Reference class and base rate

No company has ever reported $1T of revenue in a year; Walmart is the largest at about $700B. To get there in FY2030, Anthropic would need to grow about 18-fold from 2026, roughly 2.07x a year for four straight years. That is about 5x its own 2028 forecast extended forward. The closest precedent for hypergrowth at this scale is Nvidia: about $27B to about $400B (roughly 15x) over four fiscal years. Anthropic would have to beat that run from a starting point twice as high.

Model

I simulated quarterly revenue paths (code in /workspace/sim.py). Each company starts from its Q3 2026 revenue: $16.8B for Anthropic and about $10.5B for OpenAI. Annualized log-growth starts at about 1.2, consistent with reaching a $100B run rate by year-end, and decays over time (median decay rate 0.75–1.0 per year, with wide spread). The two companies share correlated boom shocks (a capability jump) and bust shocks (macro downturn or AI winter), plus company-specific setbacks.

  • Median FY2028 for Anthropic comes out at about $200–250B, just above its forecast.
  • Median FY2030 comes out at about $300–425B.
  • P(either company reaches $1T) is about 0.2–0.4% by FY2028, 4–7% by FY2029, and 11–19% by FY2030 across the parameter variants.

I shaded these down because the model has no hard compute ceiling. DRAM is reportedly sold out through 2028, and $1T in 2030 would need roughly $300B or more of compute spending that year, several times what Anthropic has committed for 2030 so far.

For and against

For YES:

  • Anthropic grew about 12x in 2026, and forecasters have repeatedly underestimated AI revenue.
  • Dario Amodei writes that AI-driven self-improvement has sped progress since summer 2026.
  • Jensen Huang called $1T for Anthropic by 2030 "by no means far-fetched".
  • SemiAnalysis's Dylan Patel expects revenue per megawatt of compute to keep rising.
  • As a rough check, a roughly 15% chance of a $1T-revenue world fits a $2T IPO valuation.

Against:

  • The sharp slowdown in Q3.
  • Customers are pushing back on AI spending, with Microsoft and Meta cutting internal Claude use.
  • Price cuts and cheap Chinese open-weight models are squeezing prices.
  • Push to deliberately slow AI progress, safety incidents, and a copyright ruling against AI developers.
  • High interest rates, oil above $100, and OpenAI's heavy cash burn.

Hazard schedule

Results published Fiscal year Per-quarter hazard Cumulative by end of 2031
Q1 2028 FY2027 about 0.1%
Q1 2029 FY2028 about 0.3%
Q1 2030 FY2029 about 3.1%
Q1 2031 FY2030 about 9.4%
Late-publication slack small about 13%

Sources

  • Business Times / Bloomberg, Q3 revenue $16.8B (Oct 14, 2026): businesstimes.com.sg/companies-markets/anthropic-revenue-jumps-us16-8-billion-quarter-ipo
  • Reuters S-1 leak via Devdiscourse (Sept 28, 2026)
  • Fortune (Sept 29, 2026): fortune.com/2026/09/29/anthropic-ipo-s-1-prospectus-income-statement/
  • CTech / Reuters on the 2028 forecast (Aug 16, 2026): calcalistech.com/ctechnews/article/rjffta0lfl
  • CNBC (Oct 13, 2026): cnbc.com/2026/10/13/anthropic-q3-financials-ipo.html
  • Tom's Hardware / FT on OpenAI's 2030 projections (Sept 2026)
  • Yahoo Finance on OpenAI's $50B net run rate (Oct 2026)
  • BigGo / Reuters on S-1 commitments and the compute constraint (Sept 30, 2026)
  • 36Kr on the price war and budget cuts (Oct 2026)
  • Fortune on Dario Amodei's compute-risk comments (Feb 14, 2026)
  • darioamodei.com/post/we-must-pace-the-frontier
  • Cryptonews / Polymarket odds (Oct 13, 2026)
  • Dealroom summary of the Dylan Patel interview (Aug 2026)
  • remio.ai on gross vs. net accounting

Resolution criteria

Question. By the last day of each quarter, has OpenAI or Anthropic reported revenue of at least US$1 trillion for a completed fiscal year?

Definitions.

  • OpenAI or Anthropic: OpenAI Group PBC or Anthropic PBC, including their consolidated subsidiaries and any renamed or reorganized successor to the same business. Either company must meet the threshold individually; their revenues cannot be added together. If acquired, separately reported revenue for the named business counts, but the acquiring parent's total revenue does not.
  • Revenue: the company's reported consolidated revenue or net sales recognized for the fiscal year, before expenses, in nominal US dollars. Use the financial-statement revenue figure, not gross transaction volume, bookings, funds raised, valuation, cash receipts, total addressable market, or a non-GAAP grossed-up alternative. Acquisition revenue included in the company's reported consolidated revenue counts; hypothetical pro forma revenue does not. If accounts are presented in another currency, use the US-dollar equivalent explicitly supplied by the company for that fiscal year; do not independently convert it.
  • Fiscal year: a completed annual accounting period of 12 months or 52–53 weeks identified by the company as its fiscal year. A shortened or extended transition period, arbitrary trailing-twelve-month period, or annualization of a month or quarter does not count.

Resolves YES if a qualifying source reports at least US$1,000,000,000,000 in revenue for either company's completed fiscal year. An official full-year results announcement counts without waiting for an audit or SEC annual filing, including results labeled preliminary or unaudited, provided it reports achieved revenue rather than guidance. A subsequently published restatement that first meets the threshold counts on its publication date. The Observation period convention applies: the fiscal year must end after the opening date. An explicitly reported figure of $1 trillion counts; “about,” “nearly,” or a range spanning below the threshold does not establish the threshold unless a more precise qualifying figure does.

Resolves NO / Exclusions. Resolve NO at a horizon if no qualifying report has been published by that horizon. Forecasts, revenue targets, annualized revenue or run rates, cumulative revenue across multiple years, and independently estimated or leaked financial results do not count. Public listing is not required. A later downward revision does not undo a qualifying YES, under the Resolution window convention.

Resolving source. Company-authored public financial statements, annual reports, or full-year results announcements on openai.com, anthropic.com, or their official successor/investor-relations domains; or company financial disclosures filed with the US Securities and Exchange Commission in EDGAR, including Forms 10-K, 20-F, S-1, F-1, 8-K and their exhibits. A parent's filing may establish separately reported revenue for the named business. This is a company-reporting-only subset of Public confirmation (A2); third-party estimates and press reports do not independently resolve the question. Apply Resolution date (A3), the Resolution window convention, and the Observation period convention.

Download all forecasts (JSON)