Google DeepMind CEO Demis Hassabis says Google will add support for Anthropic's Model Context Protocol to its Gemini models and SDK, but did not give a timeline
Just a few weeks after OpenAI said it would adopt rival Anthropic's standard for connecting AI models to the systems where data resides, Google is following suit.
Context & Ripple Effects
Google has been building Gemini for long-context and developer use cases, including its earlier Vertex AI preview with a 1M-token context window. Adding Anthropic's protocol shifts attention from how much context Gemini can hold to how consistently it can connect with external data and tools.
The move also fits Google's broader agent ambition, later visible in Gemini CLI's connection to local codebases. Here, Google is signaling that interoperability can matter alongside control of the model and SDK stack.
First-order effects
- Developers using Gemini models and Google's SDK will eventually be able to use Model Context Protocol rather than build solely to Google's proprietary integration patterns; Google gave no rollout date.
- Anthropic's protocol gains a major additional implementation commitment, making it more practical for tool and data-system providers to consider MCP support.
Second-order effects
- Tool, data, and developer-platform vendors can reduce the cost of supporting multiple leading model ecosystems if Gemini's implementation is compatible with other MCP adopters.
- Google's SDK becomes more competitive for agentic applications whose builders want to preserve model choice, rather than lock integrations to one model provider.
Third-order effects
- If major model vendors converge on MCP, the integration layer around AI agents could become more standardized while competition shifts toward model quality, distribution, and proprietary data access.
- The remaining uncertainty is implementation: an announced commitment without a timeline does not yet establish how portable Gemini-based MCP workflows will be in practice.
The trend: AI model providers are increasingly competing to become interoperable agent platforms, not just providers of proprietary model APIs.