Part of: this research — view all 4 rows
Metrics and Starting Baseline The most direct baseline for this metric is Menlo Ventures’ September 2026 survey, which estimates OpenAI’s share of enterprise API spend at 31%, behind Anthropic’s 55% and ahead of Google’s 9% 2026: The State of Enterprise AI | Menlo Ventures 2026: The State of Enterprise AI | Menlo Ventures. It is crucial to distinguish this dollar-spend metric from adoption figures, such as Datadog’s 63% observation, which measures the share of organizations using OpenAI models rather than the portion of global spending they capture State of AI Engineering State of AI Engineering State of AI Engineering. Furthermore, while headline revenue run-rates are informative, they are complicated by the inclusion of consumer subscriptions and seat-based enterprise plans, as well as accounting disparities: Anthropic books cloud-marketplace sales gross, while OpenAI recognizes partner sales net. Adjusting for these factors, a 30% baseline for OpenAI’s API dollar share is highly coherent.
OpenAI's Momentum and Upside Catalysts Despite trailing Anthropic in recent surveys, there is substantial evidence of a competitive OpenAI rather than an inevitable displacement. Newer spending indicators from late 2026 show signs of recovery; Ramp’s Token Spend Management sample saw OpenAI's share of token spending rise to 46.1% by early November, while OpenRouter reported OpenAI reaching approximately half of the combined OpenAI–Anthropic spending Ramp AI Index The Token State of Play - Inferred. OpenAI's enterprise run-rate also grew 107% in Q3, and the company has announced broad Google Cloud distribution OpenAI models are coming to Google Cloud, giving it multiple avenues to expand its enterprise footprint even in a crowded market. Additionally, OpenAI claims an unreleased model is significantly more capable than GPT-6 Astra, which could trigger a rapid capability leapfrog.
Competitive Headwinds and Dilution However, OpenAI faces formidable headwinds that constrain its median trajectory. Anthropic currently holds the capability lead, with Claude Fable 5.2 taking the top spot on the Artificial Analysis index, and dominates the critical coding segment (representing $34B of a $77.3B application market). Anthropic is also leaning heavily into an acceleration narrative, reporting a capability growth speed-up attributed to recursive self-improvement that could widen its lead if sustained. Furthermore, the broader market dynamics are increasingly dilutive. Google's Gemini 4 Carbon sits solidly at #2, and its 103% Cloud revenue growth signals a strong enterprise rebound. Widespread cost-cutting, model routing, and the proliferation of cheaper open-weights models and competitors like SpaceXAI are placing downward pressure on prices, making it harder for any single premium-priced leader to monopolize dollar share.
Market Volatility and Outlook A median of 31% reflects an equilibrium where Anthropic currently leads, but OpenAI maintains a massive, well-funded position with robust enterprise distribution. Anticipated shifts in primary vendor adoption, weighed against competitor capability compounding and growth in total enterprise LLM spending, favor a modest gain over the 30% baseline rather than a decisive recovery in share. The wide uncertainty bounds account for the extreme historical volatility of this market, where leadership has flipped repeatedly and organizational loyalty is thin—roughly half of large organizations on OpenRouter switched their top model in September alone https://openrouter.ai/data/newsletter/model-portfolios. I place the 25th and 75th percentiles at 23% and 40%, and the 10th and 90th percentiles at 16% and 49%. On the downside, sustained Anthropic RSI advantages and a stronger Google recovery could squeeze OpenAI's share to the mid-teens. On the upside, a dominant next-generation model release coupled with aggressive pricing and deep cloud integrations could allow OpenAI to reclaim the majority of enterprise spend.
Evaluating this estimate against anticipated shifts in primary vendor adoption, competitor capability compounding, and total enterprise LLM spending shifted the median slightly upward to 31% while maintaining wide uncertainty bounds.
Ask a followup
Sign in to run · $20 free credit, no card · every claim cited