AI lab valuation-to-revenue multiples, from 14x to 76x
Dividing each AI lab's latest closed valuation by its revenue run-rate gives multiples from 14.4x (Anthropic) to 76.4x (xAI). What the numbers say and don't.
Blog
How the score works, where the numbers come from, and what they do and don't mean. RSS feed
Best AI model odds on Polymarket, October 2026 history
How Polymarket's best-AI-model markets moved from 10 to 22 Oct 2026: Anthropic and Google trade the 2026 lead, OpenAI's price halves to 4.9%.
Chinese AI labs ranked: Alibaba, Moonshot, Z.ai, DeepSeek
As of 22 Oct 2026 the top Chinese lab on the scoreboard is Alibaba's Qwen at #5 (50.5), ahead of Moonshot, Z.ai and DeepSeek. Model ranks, revenue and odds.
DeepSeek valuation and funding in 2026
DeepSeek closed its first outside round at a $51.82B valuation and is seeking about $75B in a follow-on of up to RMB100B. Revenue is a $1B estimate.
Claimed vs reported revenue, gross vs net
Lab revenue numbers mix company disclosures, third-party estimates, gross vs net accounting, and figures the scoreboard annualizes itself from one quarter.
How the scoreboard's 0-100 score works
The scoreboard weighs five 0-100 components, Arena, Artificial Analysis, revenue, valuation and Polymarket odds, then rescales when a lab lacks data.
Why Arena and Artificial Analysis disagree
Google leads the Arena leaderboard but ranks 9th on Artificial Analysis, while Anthropic is the reverse: 1st on AA and 3rd on Arena.
Why Google and Meta are scored on 3 of 5
Google DeepMind and Meta have no standalone revenue or valuation figures, so the scoreboard rescales their weights across the three components it has.
What public AI compute numbers actually say
Epoch AI tracks xAI's Colossus 2 at about 946 MW of IT power, ahead of Anthropic-Amazon's 910 MW and Meta's 471 MW, but the board doesn't score compute.