“OpenAI’s inference spend with Microsoft Azure between CY2024 and Q3 CY2025 was $12.43 billion. That is an astonishing figure, one that dramatically dwarfs any and all reporting, which, based on my analysis, suggested that OpenAI spent $2 billion on inference in 2024 and $2.5 billion through H1 CY2025. In other words, inference costs are nearly triple that reported elsewhere.”
Our reading, the sentence that was checked:
OpenAI spent $12.43B on Azure inference from 2024 through Q3 2025, nearly triple what was reported elsewhere.
Assessment: Unresolved Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, Exclusive: Here's How Much OpenAI Spends On Inference and Its Revenue Share With Microsoft, paragraph 84.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-oai-docs-96467e-a2
We recorded 10 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2025-11-12. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
From the speaker
- inference spend with Microsoft Azure
- between CY2024 and Q3 CY2025
- $12.43 billion
- $2 billion on inference in 2024
- $2.5 billion through H1 CY2025
- nearly triple that reported elsewhere
Our interpretation
- The spending period is January 2024 through September 2025 inclusive.
- Assess the $12.43 billion Azure inference total and the comparison with the two identified earlier figures.
- Read 'nearly triple' ordinarily as approximately three times, not an exact equality.
- The stated numerator covers 21 months while the stated comparison covers 18 months. Assess the literal numerical comparison while recording that it cannot establish a like-for-like revision.
- Do not silently equate cost of revenue, total compute, cash burn, training expense, or cloud commitments with Azure inference expense.
- Accounting basis is estimate because these are journalist-reported figures from nonpublic documents. The words do not resolve credits, accruals, or cash settlement.
Counts as supported if: Independent records establish $12.43 billion of Azure inference expense for January 2024–September 2025. Earlier reports substantiate the identified comparison figures and their scope sufficiently to justify the claimed ratio.
Counts as contradicted if: Comparable records establish a materially different Azure inference total, or verified comparison records establish that the claimed ratio or expense classification is materially wrong.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.
- Evidence as of
- Deadline state when assessed
- No horizon
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
The claim's stated inputs yield about 2.76 times, but compare January 2024–September 2025 with January 2024–June 2025; this is conditional arithmetic, not verification [e5]. Epoch reports a $1.8 billion inference estimate for 2024 and cautions that its source figures include projections for the final quarter [e5, e6]. The Register's November 12, 2025 report relays an accounting-completeness concern without supplying a reconciled alternative total [e7]. These items establish neither the claimed $12.43 billion Azure inference total nor a comparable total that contradicts it [e5, e6, e7].
Limitations
- No independently authenticated Azure inference ledger covers January 2024–September 2025.
- The original H1 2025 comparison report was unavailable; its equivalence to inference expense remains unestablished.
- The comparison periods differ, and cash, credits, accruals and expense categories remain unreconciled.
Evidence (3 items)
Context only: Epoch AI,
These reports indicate that OpenAI spent $3 billion on training compute, $1.8 billion on inference compute, and $1 billion on research compute amortized over “multiple years”.
Why it matters here: Provides an earlier 2024 inference estimate near the comparison's $2 billion. It covers only 2024 and cannot verify or contradict the 21-month Azure total.
- Publisher
- Epoch AI
- Kind of source
- Secondary summary
- Measures
- OpenAI, Estimated 2024 inference compute expense, USD billions, estimate, 2024-01-01 to 2024-12-31
- Arithmetic
- 12.43 / (2 + 2.5) = 2.762222... (inputs: Claim-stated January 2024–September 2025 total: $12.43 billion, independently unverified.; Claim-stated comparison amounts: $2 billion for 2024 and $2.5 billion for H1 2025.; Approximately 2.76 times. The numerator covers 21 months and the denominator 18 months; this is only a conditional numerical comparison.)
- Describes events of
- Public from
Context only: Epoch AI,
These figures come from documents shared to investors in September 2024 and shown to journalists. They should not be considered completely reliable; for one, they rely on projections for the final quarter of 2024.
Why it matters here: The earlier comparison is partly forecast-based. It should not be treated as a verified final spending total.
- Publisher
- Epoch AI
- Kind of source
- Secondary summary
- Describes events of
- Public from
Context only: The Register,
Another source familiar with OpenAI suggested to the FT that the figures offered an incomplete accounting of the two company's dealings.
Why it matters here: Raises a material completeness concern but supplies no reconciled alternative total. This is secondary reporting of an unnamed source, not decisive financial evidence.
- Publisher
- The Register
- Kind of source
- News report
- Describes events of
- Public from
What would settle it
Authenticated Azure inference records for January 2024–September 2025, reconciled for credits, cash and accruals, plus the original comparison reports and enough detail to compare the same expense categories and periods.
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, Exclusive: Here's How Much OpenAI Spends On Inference and Its Revenue Share With Microsoft
OpenAI’s inference spend with Microsoft Azure between CY2024 and Q3 CY2025 was $12.43 billion. That is an astonishing figure, one that dramatically dwarfs any and all reporting, which, based on my analysis, suggested that OpenAI spent $2 billion on inference in 2024 and $2.5 billion through H1 CY2025. In other words, inference costs are nearly triple that reported elsewhere.
- Where
- paragraph 84
- Published
- 2025-11-12
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-oai-docs-96467e-a2: Current. Outcome: Unresolved. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.