Anthropic has surpassed OpenAI to become the leading enterprise large language model (LLM) provider, capturing 32% of business adoption as of mid-2025, according to a comprehensive new study by Menlo Ventures.
This represents a significant shift in the competitive landscape, with OpenAI now holding 25% market share, Google at 20%, and Meta Llama trailing at 9%. DeepSeek occupies just 1% of the enterprise market, with other providers collectively making up the remainder. The findings are based on a summer 2025 survey of 150 technical decision-makers across enterprises and startups actively building AI applications.
Menlo Ventures, an early-stage venture capital firm that has invested substantially in Anthropic through multiple funding rounds including its Series D and a $3.5 billion Series E that valued the company at $61.5 billion, conducted the research. While acknowledging Menlo’s financial stake, industry analysis from AI Magazine corroborates Anthropic’s ascendancy, noting the company “has established itself as the premier enterprise AI company through its Claude family of LLMs, achieving remarkable 1,000% year-over-year growth to reach $3 billion in annual recurring revenue.”
(Disclosure: Ziff Davis, ZDNET’s parent company, filed an April 2025 lawsuit against OpenAI alleging copyright infringement in training and operating its AI systems.)
Three primary factors underpin Anthropic’s rapid growth. Crucially, Menlo Ventures identifies code generation as “AI’s first killer app,” with Anthropic dominating this critical sector. The Claude models now hold 42% of the programming tools market—double OpenAI’s 21% share. Concrete business outcomes demonstrate this impact: Claude-powered GitHub Copilot evolved into a $1.9-billion ecosystem within a single year. The 2024 release of Claude Sonnet 3.5 further catalyzed innovation, enabling entirely new product categories including AI IDEs like Cursor and Windsurf, app builders including Lovable and Bolt, and enterprise coding agents like Claude Code and All Hands.
Technical innovations differentiate Anthropic’s approach. The company employs reinforcement learning with verifiable rewards (RLVR), a training methodology where models receive binary feedback on output correctness—particularly effective for programming applications where code either functions or fails. Anthropic also pioneered the Model Context Protocol (MCP), an open-source framework allowing LLMs to integrate external tools like search engines, calculators, and coding environments. This positions Claude models as advanced AI agents capable of iterative self-improvement and real-time data integration, moving beyond simple text generation.
Market dynamics increasingly favor performance over cost, accelerating Anthropic’s adoption. Research indicates companies prioritize capability when switching LLMs, with Menlo Ventures noting: Even as individual models drop 10x in price, builders don’t capture savings by using older models; they just move en masse to the best-performing one.
This pattern may shift as models mature and performance gaps narrow, but currently, enterprises demonstrate willingness to pay premium prices for cutting-edge capabilities.
The broader enterprise AI landscape shows accelerated adoption of production deployments. Startups lead this transition with 74% reporting most AI workloads now in production, while large enterprises follow closely with 49% indicating most or nearly all workloads are operational. This marks a strategic pivot from experimental model development to practical implementation across industries.
Meanwhile, open-source LLM usage has declined significantly, dropping to 13% of AI workloads from 19% just six months prior. Despite new model releases from DeepSeek (V3, R1), Bytedance (Doubao), Minimax (Text 1), Alibaba (Qwen 3), Moonshot AI (Kimi K2), and Z AI (GLM 4.5), open-source adoption faces dual challenges. Performance continues to trail frontier, closed-source models,
and Western businesses exhibit wariness toward Chinese-developed LLMs that dominate the open-source segment. Meta’s Llama remains the open-source leader despite debates about its licensing truly qualifying as open-source.
Market volatility persists as foundation model capabilities advance and costs plummet. Menlo Ventures concludes that conditions are ripe for a new generation of enduring AI businesses,
though the ultimate composition of the foundational AI landscape—whether dominated by Anthropic, OpenAI, Google, Meta, or others—remains uncertain. As enterprise reliance on production AI intensifies, competition increasingly hinges on delivering measurable performance advantages rather than theoretical capabilities.
