Quick Facts
- 69% of enterprises with OpenAI’s Agents SDK or Responses API in their stack name it their primary platform, versus 38% for Anthropic’s Claude Platform.
- 85% of enterprises now run two or more orchestration platforms, with a mean of 3.1 platforms per organization.
- Anthropic leads all platforms in consideration-to-current-use ratio, with 36 enterprises planning to adopt it against 45 current users.
Enterprise AI buyers are choosing OpenAI’s agent platform and sticking with it. Of the 75 enterprises that have OpenAI’s Agents SDK or Responses API somewhere in their stack, 52 — or 69% — name it their primary orchestration platform. That figure comes from the August wave of VentureBeat’s VB Pulse agentic orchestration tracker, which drew 221 responses and qualified 169 at organizations with 100 or more employees.
Anthropic’s numbers tell a different story. Of the 45 enterprises running Anthropic’s Claude Platform and Agent Skills, only 17, or 38%, name it their primary platform. That 38% is the lowest conversion rate of any platform in at least 25 enterprise stacks examined, and the pattern holds across every company size and industry group in the survey.
The Primary Platform Rankings
Among the 162 enterprises that named a primary platform, OpenAI leads at 33%, with 53 organizations. Google’s Enterprise Agent Platform ranks second at 24%, with 39 enterprises. Anthropic holds third at 11%, with 18 enterprises naming it primary.
Microsoft AI Foundry and Copilot Studio appears in 70% of all stacks surveyed. OpenAI’s Agents SDK shows up in 68%, and Anthropic’s Claude Platform in 47%. The survey reflects a market where no single vendor dominates exclusively.
Why Enterprises Are Running Multiple Platforms
The data points to a clear trend: orchestration has become plural. 85% of enterprises run two or more orchestration platforms. 64% run three or more. The mean sits at 3.1 platforms per organization.
Flexibility across models and tools is the top purchase driver at 29%, nearly three times the share naming model alignment as their primary criterion at 10%. Enterprises are not anchoring to the orchestration layer that ships with their preferred model. They are choosing platforms that avoid locking them in.
Mitch Ashley, VP and Practice Lead at The Futurum Group, framed the shift: “Enterprise teams must weigh execution and governance primitives alongside model quality.”
Anthropic’s Pipeline Advantage
Despite the low conversion rate, Anthropic has one number working in its favor. Among the 121 enterprises that named platforms they are considering adopting in the next 12 months, 36 named Anthropic. That gives it the highest ratio of consideration to current use of any platform with 40 or more enterprises in the survey.
By comparison, OpenAI drew 41 prospective adopters against 75 current users. Google drew 30 against 65. Microsoft drew 15 against 31. The consideration data suggests Anthropic’s installed base could grow faster in the near term than any rival platform.
Broader Competitive Signals
The survey results land as spending data shifts in OpenAI’s favor. OpenRouter’s weekly figures show spending on OpenAI models passing spending on Anthropic models in the week of September 7, 2026, the first time since February 2024. The shift is measured in dollars, not tokens.
At the model level, OpenAI’s Astra accounts for roughly 13% of enterprise AI spend tracked by Ramp, compared to 8% for Anthropic’s Fable. Ramp’s September 2026 AI Index does show Anthropic ahead on one metric: 43.8% of businesses in Ramp’s base transacted with Anthropic last month versus 39.8% for OpenAI, though that measure skews toward smaller, tech-forward buyers rather than the large enterprises surveyed by VB Pulse.
On revenue, Anthropic’s annualized run rate exceeded $65 billion by July. OpenAI’s surpassed $40 billion. Both companies are scaling fast, but the question for enterprise buyers is which platform they trust enough to build on as a foundation rather than a fallback.