Cloudflare AI Cloud is a platform for building and scaling AI applications and agents on Cloudflare’s global network. It combines data storage, GPU inference, and serverless infrastructure so you can run models close to users worldwide.
Serverless GPU inference
Cloudflare AI Cloud supports serverless model inference without managing clusters, with automatic scaling and global response times under 100 ms. This helps teams deploy both lightweight models and more complex AI systems in production.
AI agents and integrations
Using the Cloudflare Agents SDK and MCP (Model Context Protocol), developers can build AI agents that coordinate tools, plan tasks, and achieve goals inside the Workers environment.
- Build agents that orchestrate workflows across tools
- Run agent logic in Cloudflare Workers
- Use MCP to connect agents to external context and capabilities
Data storage for training
The platform includes R2 for storing training data with egress-free access to GPUs in a multi-cloud setup. This can reduce data transfer costs and simplify training and fine-tuning pipelines on the same network Cloudflare uses for its own AI workloads.
