BRIEFINGS · Cloud and AI compute providers
Crusoe and Thinking Machines Lab announce a $65 million annual inference agreement
Crusoe and Thinking Machines Lab said on 23 September they have a $65 million annual agreement to run the lab's open-model inference on Crusoe Cloud.
On 23 September 2026 Crusoe and Thinking Machines Lab announced a $65 million annual agreement to run the lab's open-model inference on Crusoe Cloud. The lab will serve production workloads on a dedicated cluster of NVIDIA HGX B200 systems that Crusoe runs and supports. The companies said they are also looking at batch inference for large-scale synthetic data generation. Crusoe said the customer joins others that have taken its managed inference product past $100 million of contracted annual recurring revenue. We file this because it is a named customer with a stated annual amount for AI computing. How we read it: AI labs are paying for dedicated inference capacity rather than building it themselves. What would change this: the release gives no contract length, so the commitment could be short. Why it matters to you: it shows who is paying for the computing that runs AI models, and how much.
What this doesn't prove
Joint company press release. The annual amount is given but not the contract length, total value or GPU count. The $100 million recurring revenue figure is Crusoe's own.
Next fixture
- Batch inference expansion
- Other labs signing with disclosed amounts
- Contract length
Sources
Every figure above comes from these sources. Every briefing is checked against its sources before it publishes. Nothing here is a recommendation to buy or sell anything, and this is not investment advice. See the disclosures.
One finding from the AI economy in your inbox every Saturday. Free, three minutes, every fact sourced.
This is general information, not investment advice.