China's Top 7 AI Companies Earn a Tenth of What OpenAI and Anthropic Make Combined — Despite Building World-Class Models
Benchmark leaderboards tell one story about the AI race. Revenue tells a completely different one.
New estimates from Rhodium Group, reported by the South China Morning Post, put a number on a gap the industry has talked around for years:
→ China's seven leading AI developers — ByteDance, Alibaba, DeepSeek, Moonshot AI, Z.ai, MiniMax, and Kuaishou — generate a combined ~$10.7 billion in annual recurring revenue
→ OpenAI and Anthropic together report over $100 billion in combined ARR — roughly 10x the entire Chinese field put together
→ ByteDance leads China's pack at ~$4B ARR, followed by Alibaba at ~$2.4B
→ These are run-rate figures, not audited annual revenue — useful as a scale indicator, not a precise accounting comparison
Here's the part worth sitting with: this isn't a capability gap. Chinese labs have shipped models that compete on benchmarks with anything from the US. It's a business-model gap. American labs built their revenue on expensive enterprise contracts, premium subscriptions, deep API demand, and tight cloud-distribution partnerships. Chinese developers, by contrast, operate in a brutally price-competitive domestic market and have leaned hard into open-weight releases — which spread adoption fast but don't convert directly into revenue the way a locked-down API does.
What this really points to is two different AI business models forming in parallel: one optimized to monetize every token, the other optimized to maximize reach and ecosystem lock-in first, revenue later (or never, in the traditional sense). Betting on which model "wins" long-term is a much harder call than reading a benchmark chart — because open-weight distribution can still translate into leverage, just not the kind that shows up on an ARR slide.
If you're evaluating AI companies — as an investor, a partner, or a buyer — benchmark scores are starting to tell you less than they used to. The real question is becoming: how does this lab actually make money, and is that model built to last?
Which do you think wins the next 5 years — the monetization-first US model, or China's distribution-first open-weight bet?
#AI #China #OpenAI #Anthropic #TechEconomics #AIIndustry #TechNews
— 𝔖𝔞𝔫𝔡𝔢𝔢𝔭 ℜ𝔞𝔦𝔷𝔞