Anthropic's competition with OpenAI has entered a heated stage, shifting focus from simple model capability battles to a comprehensive game of commercialization and infrastructure. In April 2026, Anthropic's annualized revenue (ARR) surpassed $30 billion for the first time, exceeding OpenAI's $25 billion during the same period, making it the world's highest-earning independent large model company. On the enterprise side, reflecting real commercial competitiveness, Anthropic's API market share has soared from 12% in 2023 to 40%, overtaking OpenAI's 27%; among new AI service purchases by U.S. companies, Anthropic has captured about 73% of the market share.

The core engine behind this explosive performance is the AI programming tool Claude Code—its annualized revenue has skyrocketed from $17 million in April 2025 to over $2.5 billion, holding 54% of the AI programming market, while OpenAI only accounts for 21%. Meanwhile, Anthropic is building a highly strategic and deep compute infrastructure, maintaining partnerships with the three major cloud platforms—AWS, Google Cloud, and Microsoft Azure—and jointly securing about 3.5 gigawatts of TPU compute power with Google and Broadcom, forming a "multi-cloud, multi-chip" flexible architecture. More profound differences lie in their values: Anthropic adheres to a "safety-first" philosophy, refusing to open its models to the U.S. Pentagon and voluntarily abandoning billion-dollar military contracts; whereas OpenAI quickly followed suit by signing defense contracts, attempting to seize early opportunities in the government market. These two very different development paths are pushing the AI race into a deeper institutional contest. #Anthropic与OpenAI竞争升级