OpenAI Still Leads Enterprise AI, but Anthropic Is Gaining Fast, According to New Study

Why It Matters

The shifting provider mix signals an emerging oligopoly where trust, rapid model upgrades, and integration drive spend, forcing enterprises to rethink vendor lock‑in and budgeting strategies. CIOs must monitor Anthropic and Google as they close the gap on OpenAI.

Key Takeaways

• OpenAI holds 56% AI model spend now.
• Anthropic usage rose 25% to 44% since May 2025.
• Google and Anthropic each projected 18% share by 2026.
• Microsoft 365 Copilot adopted by 94% of enterprises.
• Enterprise LLM spend hit $7M in 2025, up 180%.

Pulse Analysis

The third‑annual a16z CIO survey, which interviewed 100 senior technology leaders from Global 2000 companies, confirms that enterprise AI is coalescing around a handful of dominant providers rather than fragmenting. OpenAI remains the clear front‑runner with roughly 56 percent of total large‑language‑model (LLM) spend, but its share is projected to dip to 53 percent by 2026 as Anthropic and Google each climb toward an 18‑percent slice. This emerging oligopoly reflects CIOs’ preference for trusted, proprietary models that deliver rapid improvements and meet stringent data‑security requirements. The findings also suggest that open‑source alternatives remain a distant prospect for most large enterprises.

Use‑case analysis reveals a nuanced competitive landscape. OpenAI continues to dominate horizontal applications such as chatbots, knowledge‑management tools, and customer‑support agents, while Anthropic has captured the software‑development and data‑analysis segments, largely thanks to its Sonnet 4.5 and Opus 4.5 models. Google’s Gemini performs solidly across many domains but lags behind in coding capabilities. The survey also shows that 81 percent of enterprises now run three or more model families, indicating a diversification strategy that balances performance, cost, and risk. CIOs cite faster model improvements since late 2024 as a primary driver for switching to newer providers.

Enterprise spending on LLMs exploded from $2.5 million in 2024 to $7 million in 2025—a 180 percent jump that outpaced even the most bullish forecasts. Microsoft’s ecosystem reinforces this trend, with 94 percent of respondents deploying Microsoft 365 Copilot and 74 percent using GitHub Copilot. As budgets swell, vendors that can combine deep integration, reliable governance, and accelerated model updates will capture the most value. For CIOs, the data underscores the need to negotiate flexible contracts and to monitor the accelerating rise of Anthropic and Google as viable alternatives to OpenAI. Analysts predict that by 2027, total AI model spend could exceed $20 million per enterprise, intensifying vendor competition.