Exclusive: Google deepens Thinking Machines Lab ties with new multi-billion-dollar deal

Editor's take

Thinking Machines Lab has secured a multi-billion dollar commitment from Google Cloud for AI compute resources, leveraging Nvidia's upcoming GB300 Blackwell chips. This significant infrastructure pact underscores the escalating demand for specialized hardware to train and deploy increasingly sophisticated AI models, particularly as companies like OpenAI, backed by Microsoft, continue to push the boundaries of large language models and multimodal AI.

The agreement highlights Google Cloud's strategy to attract major AI players to its platform, directly competing with Azure's existing relationship with OpenAI. The sheer scale of the deal suggests a long-term commitment to AI development, potentially accelerating Thinking Machines' ability to iterate on its proprietary models and compete with established leaders. The inclusion of Nvidia's next-generation chips signals an arms race in AI hardware, with performance and efficiency being paramount.

Future developments will reveal the specific performance gains Thinking Machines achieves with the GB300 architecture and how this infrastructure advantage translates into model capabilities. Observing whether Google Cloud can maintain this level of hardware investment and service quality for other major AI labs will be crucial in understanding the evolving cloud AI landscape. The success of this partnership could set a precedent for future multi-billion dollar infrastructure commitments.