New data from Counterpoint Research puts Anthropic at the top of the global LLM revenue table for the first quarter of 2026, with a 31.4 percent share against OpenAI's 29 percent. The near-tie at the top masks a striking difference in how both companies got there. Anthropic achieved its position with approximately 134 million monthly users. OpenAI served roughly 900 million. The companies are within a few percentage points of each other on revenue while being separated by nearly seven to one on scale.
The per-user numbers reveal the strategy gap. Counterpoint estimates Anthropic's average monthly revenue per active user at $16.20, compared to $2.20 for OpenAI, $5 for Microsoft, and $1.10 for Google. At the other end of the table, Meta leads on raw scale with around a billion users and monetises at roughly $0.10 per head, generating almost nothing directly from its AI offerings despite the enormous user base. The pattern suggests the LLM market is sorting into at least two distinct models: platforms that treat AI as a user-acquisition tool and hope monetisation follows, and companies that have built high-value relationships with enterprise and professional customers willing to pay substantially for the capability they get.
Anthropic is now pursuing a funding round that would value it at more than $900 billion, according to Bloomberg, which would put it ahead of OpenAI's $852 billion valuation from its March round. The speed of that climb is notable: Anthropic was valued at $380 billion as recently as February. The driver cited in reporting is Claude Code, the company's AI coding assistant, which has accelerated enterprise adoption since its launch and contributed to annualised revenue that now runs at around $30 billion. The Mythos model, Anthropic's frontier cybersecurity AI, has also drawn high-profile attention from government and financial institutions, even as its restricted release limits its immediate revenue contribution.
The infrastructure commitments behind Anthropic's growth are substantial. Amazon has agreed to invest up to $25 billion and is providing five gigawatts of compute capacity. Google has committed up to $40 billion and is supplying an additional five gigawatts, alongside Broadcom, set to come online next year. These are not conventional investment rounds; they are structural partnerships that give Anthropic the compute required to operate frontier models at scale and the relationships that put Claude into a large share of enterprise AI infrastructure. The capital it is now raising is partly intended to support Mythos's compute-intensive requirements.
The broader AI infrastructure picture adds context to why the revenue numbers matter. The Register's analysis of Counterpoint's data notes that Alphabet, Amazon, Microsoft, and Meta combined are expected to spend $725 billion on infrastructure in 2026, up from $410 billion last year. Meta, despite leading on user numbers and investing the most heavily, saw shares fall 7 percent after its most recent earnings as investors weighed whether the engagement it is building will ever translate into the kind of direct monetisation Anthropic is demonstrating. The companies winning on popularity are not, at this moment, the ones making the most money from AI.
There is an asterisk on Anthropic's position that investors appear largely to have set aside. The Pentagon designated Anthropic a supply chain risk, requiring government contractors to avoid doing business with the company. Litigation over the designation is pending, and its implications for Anthropic's government relationships remain unresolved. Shares on secondary markets are already trading at prices that imply a valuation near $1 trillion, suggesting the financial community is treating the Pentagon situation as manageable rather than structural. Whether that confidence is warranted will depend on how the litigation resolves and whether any restrictions expand.
The revenue story is useful not just as a competitive scorecard but as a signal about where value is accumulating in the AI ecosystem. A company that generates $16.20 per user per month has built something enterprises consider genuinely indispensable. A company that generates $2.20 with seven times the users is still, at least in part, giving away value to capture attention. Both strategies can work in the long run, but the enterprise-focused path is generating cash now, which matters in an environment where the compute costs of running frontier models are themselves enormous.