Blog · 23 Oct 2026 · Data as of 22 Oct 2026

AI lab valuation-to-revenue multiples, from 14x to 76x

How much are investors paying for each dollar of AI lab revenue? Six labs on the scoreboard have both a closed valuation and a revenue figure, so the ratio can be computed directly. The division is ours; both inputs are sourced and dated on each lab page.

The multiples

LabClosed valuationAs ofRevenue run-rateAs ofValuation ÷ revenue
Anthropic$965B28 May 2026$67.2B (Q3 x 4)30 Sep 202614.4x
OpenAI$852B31 Mar 2026$50B30 Sep 202617.0x
Mistral AI$24B8 Sep 2026about $1B8 Sep 2026about 24x
Moonshot AI$50B6 Oct 2026$1B6 Oct 202650.0x
DeepSeek$51.82B17 Jul 2026$1B (estimate)30 Sep 202651.8x
xAI$250B2 Feb 2026$3.272B (Q1 x 4)31 Mar 202676.4x

Google DeepMind, Meta Super­intelligence Labs and Alibaba's Qwen team sit inside public companies that do not disclose a lab-level valuation, and Z.ai has no sourced financials on the board yet, so none of them has a multiple.

Bigger labs trade at lower multiples

The two largest labs by revenue have the lowest multiples: Anthropic at 14.4x and OpenAI at 17.0x. The four smaller labs all sit between roughly 24x and 76x. With revenue at $1B to $3.3B, their valuations are priced mostly on expected growth rather than current sales.

Why the multiples are rough

Rounds in talks would raise the numbers

Two labs are negotiating at higher prices. Neither round has closed, so the scoreboard doesn't score them:

The multiple is not part of the 0-100 score: revenue (20 points) and valuation (10 points) are scored separately on log scales. The live table, with total funding raised, is on AI company valuations.

Sources

All blog posts →