This assessment estimates the true operational compute allocation at Anthropic as of December 31, 2026, specifically the share dedicated to "research experiments." Under this three-way taxonomy, this bucket captures all internal compute that is neither external inference serving nor the final training runs of released models. This includes ablations, scaling and derisking runs, synthetic data generation, reinforcement learning environment rollouts, and internal agentic inference workflows.
The
Weighing this question against expected enterprise API market shares and automated AI research trends led to a slight upward adjustment, balancing massive external serving loads against the growing resource demands of internal agentic workflows.
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 me
Aligning this forecast with expected trajectories for OpenAI's competitive API pricing and Anthropic's footprint in the coding assistant market kept the median stable near 41% while refining the tails to account for potential multi-model fragmentation.
Resolution Basis and Current Baseline
The forecast target requires a mutually exclusive share of worldwide enterprise LLM API spend that sums with competitors to roughly 100%. This rules out adoption breadth metrics—like Datadog's finding that OpenAI is used by 63% of organizations datadoghq.com or an a16z CIO survey showing 56% wallet share 2 sources—because these involve overlapping multi-model usage that sums well above 100% datadoghq.com. The most appropriate anchor is a Menlo Vent
Balancing this projection against forecasts for competing providers' API market shares and broader revenue-scaling expectations led to a slight upward adjustment in the median and lower tail.
What We Are Measuring This estimate reflects a revenue-weighted market share, rather than simply developer penetration. The numerator is direct calendar-2027 revenue attributable to Claude Code, while the denominator includes the total paid coding-assistant market (OpenAI, Cursor, GitHub Copilot, Google, plus a smaller long tail). While penetration proxies are useful—an August 2026 JetBrains survey placed Claude Code at 39%, Copilot 21%, Codex 16%, Cursor 12%, and Google at 6% [c6baa8, 2c3f2
Aligning this estimate with projections for Anthropic's broader enterprise API footprint and the annualized revenue of its coding products kept the median share stable near 42% while slightly tightening the tails.
Ask a followup
Sign in to run · $20 free credit, no card · every claim cited