Skip to content
Longterm Wiki
Index
Fact·f_DtyJCdGd1g·Fact

OpenAI — Annual Expenses: $22B

Verdictpartial75%
1 check · 5/18/2026

1 → partial

Our claim

entire record
Subject
OpenAI
Property
Annual Expenses
Value
$22B
As Of
December 2025
Notes
Total anticipated spending of ~$22B in 2025 against $13.1B revenue; inference costs of $8.4B, R&D of $6.7B in H1 alone, plus $2.5B in stock-based compensation in H1

Source evidence

1 src · 1 check
partial75%primaryHaiku 4.5 · 5/18/2026

NoteThe claim states annual expenses of $22B (exact: 22,000,000,000) as of 2025-12. The source provides several relevant data points: (1) H1 2025 R&D of $6.7B + H1 stock compensation of $2.5B = $9.2B for half-year, which would annualize to ~$18.4B for those two categories alone; (2) inference costs of $8.4B in 2025; (3) projected cash burn of $8.5B-$9B in 2025. However, the source does NOT provide a single consolidated 'annual expenses' figure for 2025. The claim's $22B appears to be an inference or projection based on combining multiple cost categories (inference $8.4B + R&D $6.7B annualized + stock compensation $2.5B annualized + other costs), but the source does not explicitly state total annual expenses of $22B. The source's stated 'cash burn' figures ($8.5B-$9B) are lower than the claimed $22B, though cash burn and total expenses are different metrics. The component figures cited in the claim's context are present in the source, but the aggregate $22B figure is not directly confirmed.

Case № f_DtyJCdGd1gFiled 5/18/2026Confidence 75%
Source Check: Annual Expenses | Longterm Wiki