Enterprise AI Market Shifts from Model Competition to Agent Infrastructure Control
New survey data shows Microsoft and OpenAI leading enterprise AI agent orchestration platforms, while Anthropic makes first appearance in the emerging market.
The enterprise artificial intelligence market is shifting from competition over AI models to control over the infrastructure where AI agents operate, according to new survey data from VentureBeat's Enterprise Agentic Orchestration tracker. The February survey of 70 qualified enterprise technical decision-makers found Microsoft's Copilot Studio and Azure AI Studio leading with 38.6% adoption, up from 35.7% in January.
OpenAI's Assistants and Responses API held second place with 25.7% adoption, rising from 23.2% the previous month. Anthropic made its first appearance in the tracker, moving from zero adoption in January to 5.7% in February for its Claude tool use and workflow capabilities. While the numbers remain small, industry analysts view this as significant because it represents the first sign of Claude usage expanding beyond the model layer into native orchestration.
The orchestration layer encompasses where AI agents plan workflows, access data, call tools, and maintain audit logs. Unlike swapping AI models, which enterprises can do relatively easily, changing agent runtime infrastructure involves migrating workflows, permissions, credentials, and operational monitoring between providers. This makes switching costs significantly higher and creates potential vendor lock-in scenarios.
Security and permissions ranked as the top selection criteria for orchestration platforms in both January and February surveys, at 39.3% and 37.1% respectively. Control over agent execution rose from 17.9% to 22.9%, while flexibility across models and tools fell from 35.7% to 25.7%. The data suggests enterprises are prioritizing governance over optionality as AI agents handle more sensitive operational tasks.
Microsoft's early lead reflects its existing enterprise distribution through Microsoft 365, Teams, and Azure services. The company's integrated approach allows organizations to manage AI agents within their existing technology stack and identity systems. OpenAI's position stems from its Assistants API, which provided developers early tools for building agent-like systems.
Despite the competitive dynamics, survey respondents indicated preference for hybrid control planes rather than single-vendor solutions. Around 35-36% of enterprises across survey waves expected to combine provider-native orchestration with external orchestration tools. Vendor lock-in emerged as the second-largest concern after security limitations, rising from 23.2% to 25.7% between January and February.