Thinking Machines Lab signs $65 million annual deal with Crusoe

Thinking Machines Lab, an AI startup founded by former OpenAI CTO Mira Murati, has entered into a $65 million annual agreement with Crusoe to power the inference of its open models. The partnership utilizes Crusoe Managed Inference to run production workloads on a dedicated deployment of NVIDIA HGX B200 systems equipped with NVIDIA Quantum-2 InfiniBand networking.

The agreement covers a variety of models, including the lab’s proprietary Inkling models, as well as GLM 5.2 and 5.3 variants. Through this tailored deployment, Thinking Machines Lab will utilize a benchmarked, SLA-backed endpoint, allowing the lab to scale its research and production without managing its own inference stack.

With this contract, Crusoe announced that its managed inference business has surpassed $100 million in annual recurring revenue (ARR) within less than a year of the product's launch. The two companies are also exploring the use of batch inference for large-scale synthetic data generation to support future research efforts.