ユーザー数は減っても財布は太い、これが Anthropic が OpenAI を LLM 収益面で上回る理由

Anthropic is pulling in more LLM revenue than OpenAI, despite having a fraction of the users.

New stats from Counterpoint Research put Anthropic on top of the global LLM revenue table in Q1 2026, with a 31.4 percent share, narrowly ahead of OpenAI on 29 percent, but the real story lies beneath that near-tie. Anthropic is doing it with around 134 million monthly users, compared to OpenAI's roughly 900 million, and extracting far more revenue from each one.

Counterpoint puts Anthropic's average monthly revenue per active user at $16.20, far ahead of Microsoft's $5, OpenAI's $2.20, and Google's $1.10. That's not a subscription price so much as a rough average spread across total users, but it still shows where the money is landing. The companies winning in terms of popularity are not the ones making the most from it.

You can see that in Meta's numbers. The company leads on raw scale with about a billion users, but barely registers on direct monetization at $0.10 per head.

Investors, however, are being asked to take that on faith. Meta has raised its expected 2026 capital expenditure to between $125 billion and $145 billion as it pours money into AI infrastructure, up from an earlier forecast of $115 billion to $135 billion. The company beat expectations on revenue and profit in the first quarter, but the heavier AI spending knocked its shares down 7 percent in after-hours trading, a reminder that engagement gains do not automatically translate into confidence.

It's not alone. Combined, Alphabet, Amazon, Microsoft, and Meta are expected to spend a $725 billion on infrastructure this year, up sharply from an already hefty $410 billion last year. That's a 77 percent jump, which would be easier to swallow if the returns were showing up evenly across the market.

But they are not. The biggest platforms are racking up users and engagement while footing ever larger infrastructure bills. Meanwhile a smaller set of players is already turning AI into something that looks a lot more like a traditional software business.

• Amazon chips no longer just a side dish, they're a $20B biz

• Databricks can't seem to shake authors' copyright claim that could result in 'extraordinary' damages

• Microsoft's GitHub shifts to metered AI billing amid cost crisis

• Anthropic's magic code-sniffer: More Swiss cheese than cheddar, for now

Anthropic, for example, appears to have carved out a premium lane where customers are paying for outputs rather than novelty, with Counterpoint noting that the firm has "successfully captured the high-end professional market."

China's players are not sitting this out either. The analysts peg Baidu as pulling in about $1.30 per user, more than Google and Meta combined, while Tencent and Alibaba are building out their own mixes of scale and services behind the Great Firewall.

All in all, more than 3.8 billion people are now using LLMs each month, generating roughly $20.7 billion in quarterly revenue, but not all of those users are worth much. Billions are happily tapping away for free while a smaller crowd racks up the actual bill. ®