Source: Mira Murati's TML signed a deal with Google Cloud, valued in single-digit billions, to access Google's latest AI systems built on Nvidia's GB300 chips
Context & Ripple Effects
Thinking Machines Lab has been assembling the inputs for a large-scale AI buildout: recruiting senior technical talent, pursuing major financing, and reportedly securing a separate Nvidia chip-supply commitment. The Google Cloud arrangement adds a managed deployment route alongside that hardware relationship.
The reported scale matters because it ties a young model developer’s ability to build products to long-duration access to scarce, current-generation compute rather than to a small, on-demand cloud footprint.
First-order effects
- TML gains access to Google Cloud-hosted AI capacity built on Nvidia GB300 systems, giving it an immediate path to train and run workloads without relying solely on infrastructure it operates directly.
- Google Cloud adds a large reported commitment for its AI infrastructure, while Nvidia benefits from another channel for deployment of its latest systems.
Second-order effects
- The deal makes TML’s compute strategy more hybrid: its reported Nvidia supply agreement can coexist with cloud capacity, reducing dependence on a single delivery model while increasing the coordination burden across suppliers and platforms.
- Large commitments from AI labs can make cloud providers compete not just on chip availability but on the surrounding deployment stack, including the systems and services that turn hardware access into usable model capacity.
Third-order effects
- If similar arrangements persist, frontier-model development will increasingly be shaped by multi-year capacity contracts and a small set of cloud-and-chip providers, raising the capital and procurement threshold for independent labs.
- The pattern points toward AI compute becoming a capacity market: access is secured in advance through large commitments, rather than acquired only as variable cloud usage when models are ready to run.
The trend: AI labs are locking in scarce next-generation compute through a mix of direct chip agreements and cloud-platform contracts, turning infrastructure procurement into a core competitive strategy.