rationale:

The true target here is worldwide enterprise LLM API dollar spend, not usage, adoption breadth, or any single survey's figure. A useful anchor is Menlo Ventures' late-2025 enterprise survey, which placed Anthropic at 40% of enterprise LLM API spend—up from 24% a year earlier—compared to OpenAI's 27% and Google's 21% menlovc.com. It is crucial to distinguish this dollar-spend metric from breadth-of-adoption telemetry. For example, Datadog's report showing OpenAI at 63% share in early 2026 measures the percentage of organizations emitting LLM spans, not the volume of dollars spent datadoghq.com. Breadth metrics structurally favor OpenAI as the default first vendor, but spend metrics favor whoever captures token-heavy, agentic workflows, particularly in coding, where Anthropic has a commanding lead menlovc.com.

Current revenue math implies Anthropic's true dollar share today is likely even higher than Menlo's 40%, potentially sitting in the 50–60% range. As of mid-2026, Anthropic's run rate exceeded $65B reuters.com, with an estimated 70–75% derived from pay-per-token APIs sacra.com. By contrast, OpenAI's model-as-a-service/API line makes up a much smaller fraction of its ~$40B run rate sacra.com. Even accounting for Google's Gemini API growth and the long tail of alternative providers, Anthropic is the clear market leader in B2B API spend right now, fueled by sticky enterprise deployments and massive cloud distribution deals with AWS and Google anthropic.com.

However, projecting this dominance forward 2.4 years requires applying substantial mean reversion. The market is highly unstable, having swung 25+ points in single years. Sustaining a near-monopoly share of global spend will be incredibly difficult against aggressive competitors. OpenAI is actively executing heavy price cuts (e.g., dropping GPT-5.6 Luna prices by 80%) businessinsider.com and pushing enterprise-focused products openai.com. Google brings a structural TPU cost advantage and bundled default distribution through Google Cloud and Workspace cloud.google.com. Furthermore, as enterprises mature, they are pivoting from raw premium consumption to utilizing caps, tiers, and model routers cnbc.com. The increasing viability of open-weight and cheaper Chinese models—which are heavily discounted but capturing massive token volumes—will also serve as a structural drag on the dollar denominator of premium closed APIs 2 sources.

Balancing Anthropic’s current momentum against the inevitable commoditization and price wars—and aligning this forecast with expected trajectories for OpenAI's competitive API pricing and Anthropic's footprint in the coding assistant market—yields a stable median estimate of 41%. This lands modestly above Menlo’s last published baseline but represents a material decline from the current revenue-implied peak. The uncertainty bands are set to account for potential multi-model fragmentation, defining an interquartile range from 31% to 51%. The upper tail, reaching 59% at the 90th percentile, accounts for scenarios where agentic coding remains the undisputed driver of API spend and Anthropic maintains its frontier edge. The lower tail, dipping to 21% at the 10th percentile, accounts for a successful enterprise pivot by OpenAI, massive multi-model fragmentation, or a scenario where Anthropic's reported revenue is revealed to be heavily skewed by gross-reseller accounting rather than net true API consumption.