US Tech Giants Unite to Battle China's Open-Source AI Dominance
Anthropic, OpenAI, and Block have formed the Agentic AI Foundation under the Linux Foundation to develop open standards for AI agents. The foundation's core projects include the Model Context Protocol (MCP) from Anthropic, AGENTS.md from OpenAI, and Goose from Block. Platinum members of the foundation include AWS, Bloomberg, Cloudflare, Google, Microsoft, Anthropic, and OpenAI. Gold and Silver members involve a broad range of industry players such as Cisco, Datadog, Docker, IBM, Oracle, SAP, Snowflake, Twilio, Hugging Face, Uber, and SUSE.
The foundation promotes neutral, vendor-neutral governance for agentic AI infrastructure, ensuring no single company controls its direction. MCP enables AI models to leverage tools beyond simple API calls and currently has substantial traction with over 10,000 active servers and 97 million monthly SDK downloads. It is supported across multiple platforms including ChatGPT, Gemini, Microsoft Copilot, and Visual Studio Code. OpenAI's AGENTS.md provides a lightweight specification offering standardized instructions for AI agents and is employed by about 60,000 repositories. Block contributed Goose, a local-first agent framework. All three projects are now governed by the Linux Foundation to enhance interoperability.
This initiative is positioned as a strategic response to China's growing share of open-source AI downloads. According to a December 2025 MIT study, China accounted for 17.1% of global open-source AI downloads compared to the US's 15.8%, with Chinese companies like DeepSeek and Alibaba expanding high-performance open AI models.
The policy context underscores the potential geostrategic value of open-source and open-weight AI models. Both the White House and the Trump-era AI Action Plan acknowledge the risks of reliance on Chinese open AI models and endorse standards that aim to secure global AI development and innovation.
