Google Cloud expands AI computing resources for Thinking Machines Lab to accelerate AI model training

Google Cloud signed a new agreement with Thinking Machines Lab to provide expanded access to its AI Hypercomputer—a specialized infrastructure designed to train large AI models (the kind of systems that power ChatGPT-like assistants). The deal increases Thinking Machines' capacity and footprint on Google's servers.

The AI Hypercomputer bundles high-speed chips, specialized networking, and software tools that work together to cut the time and cost of training frontier models—meaning Thinking Machines can experiment faster and reach production deployments weeks or months sooner than on standard cloud infrastructure.

For businesses and researchers working on AI projects, this signals that Google Cloud is competing harder to lock in AI-focused customers by offering turnkey infrastructure rather than making them assemble parts themselves. It also means Thinking Machines can now build and launch AI applications faster, potentially bringing new AI tools to market sooner than competitors using less optimized setups.