Claude's enterprise market share nearly doubled in nine months, and the retention data explains why

The market share figure that Harry Stebbings put on the table is stark. Claude went from 21% to 48% enterprise market share in the last nine months. Gemini moved too, from 27% to 40% over the same period. But the Claude number is the one that warrants attention, because it is not just growth: it is the pace of growth relative to a starting position that was already meaningful, and it arrives alongside retention and spend data that describe a customer base that is deepening its commitment rather than experimenting.

Krishna Rao places Anthropic’s net dollar retention above 500% on an annualized basis. That figure is unusual enough to require a moment of context. Net dollar retention measures how much existing customers spend in a later period relative to an earlier one, accounting for churn and expansion. A number above 100% means expansion outpaces losses. A number above 500% means the existing customer base is, on average, spending five times what it spent a year ago. That is not a metric that describes a product people tolerate. It describes a product around which customers are reorganizing their operations.

The spend data from individual operators reinforce the picture. Nico Laqua, describing Korgi’s AI infrastructure, notes that the company spends roughly $400,000 per month on Anthropic and nothing on OpenAI. That is a single data point, but it illustrates the kind of vendor concentration that emerges when a product becomes load-bearing in a company’s stack. When the monthly bill reaches that level, switching costs are no longer hypothetical.

Andrew Lee adds a behavioral signal from the other direction. At Tasklet, when users churn, 80% of them migrate to an Anthropic product, primarily because they already hold a Max subscription. That is a telling dynamic: Claude is not only winning enterprise accounts from the top down, it is also capturing churned users from adjacent products because those users have already built a personal relationship with the model. The enterprise sale and the individual adoption are reinforcing each other.

Our net dollar retention rate is over 500% on an annualized basis. Krishna Rao

Martin Casado traces part of the strategic explanation back to a deliberate bet. Anthropic, having raised less capital than its main rivals, chose to concentrate on coding. That focus, Casado argues, succeeded. The decision to go deep on a specific high-value use case rather than spread thin across every surface appears to have created a compounding technical and reputational advantage in the developer and enterprise segment, precisely the segment where spending and retention data now appear most favorable.

Rao points to a further compounding dynamic in the product layer. Co-work, built on Claude, is achieving faster adoption than Claude Code did when measured at the equivalent point after launch. If that trajectory holds, it suggests the base that Claude Code built is now accelerating adoption of subsequent products, rather than each product starting from scratch. That is how platform dynamics behave when they are working.

Gavin Baker captures the competitive realignment in model quality terms: where Google held the dominant position on the cost-versus-intelligence frontier nine months ago, that frontier is now shaped by Anthropic and OpenAI, with Grok also present. The market share data and the model quality data are telling the same story from different angles.

What the combined evidence describes is not a company winning on price or timing alone. The retention rate suggests customers are finding more to do with the product over time, not less. The churn behavior at Tasklet suggests personal familiarity with Claude is becoming a durable asset that travels with users across products. The spend concentration at companies like Korgi suggests that, for some operators, the vendor question is already settled. The strategic question for rivals is not whether the shift happened, but whether the compounding dynamics now embedded in Claude’s customer base can be interrupted before they become self-reinforcing at a scale that forecloses competition in the segment.