Mira Murati deal
- Thinking Machines Lab, led by Mira Murati, signed a multi‑billion dollar Google Cloud agreement for Nvidia GB300 infra. - The partnership centers on GB300 hardware deployment to host large AI workloads on Google Cloud. - The arrangement was reported publicly in TechCrunch social posts this week, highlighting cloud spend for new GB300 nodes (x.com).
Mira Murati’s Thinking Machines Lab has signed a multi-billion-dollar agreement with Google Cloud for Nvidia GB300-based artificial intelligence infrastructure. (techcrunch.com) Google Cloud announced the expanded arrangement on April 22 at Google Cloud Next in Las Vegas, where it said Thinking Machines will use A4X Max virtual machines and other cloud services for model research and training. (marketwatch.com) (googlecloudevents.com) A4X Max is Google Cloud’s server offering built around Nvidia GB300 NVL72 systems. Google said each system combines 72 Blackwell Ultra graphics processors with 36 Grace central processors and is designed for large multimodal AI workloads. (cloud.google.com) Thinking Machines is Murati’s artificial intelligence startup, launched publicly in February 2025 after her departure from OpenAI. The company says it is building multimodal systems that are more customizable and easier for people to use in specific domains. (techcrunch.com) (thinkingmachines.ai) The cloud agreement adds to a separate infrastructure push the startup disclosed in March. TechCrunch reported on March 10 that Thinking Machines also signed a multi-year partnership with Nvidia that includes at least one gigawatt of Vera Rubin systems starting in 2027. (techcrunch.com) That spending comes on top of a large cash raise. Thinking Machines said in July 2025 that it raised about $2 billion at a $12 billion valuation, with participation from Nvidia, AMD, Cisco, ServiceNow, Accel, Andreessen Horowitz, and Jane Street. (techcrunch.com) Google has been rolling out GB300 capacity for months as it tries to win more frontier-model customers. In an October 2025 post, the company said A4X Max was “shipping in production” and could scale to tens of thousands of graphics processors in a cluster. (cloud.google.com) For Thinking Machines, the deal shows how much of an artificial intelligence startup’s early life now depends on securing compute before shipping many products. The company’s website says infrastructure quality is a top priority because research productivity depends on reliable, efficient systems. (thinkingmachines.ai) The immediate next step is deployment: Google Cloud said the new capacity will support Thinking Machines’ research, platform development, and frontier model training on its AI Hypercomputer stack. (prnewswire.com)
