Dell Technologies secured a $150 million deal with AI start-up Imbue to build a compute cluster for training foundation models optimised for reasoning, which marked a customer win for its server business.

Imbue stated it is already using the cluster to train AI models and develop prototype agents capable of correcting bugs in code and analysing lengthy documents. The set-up uses Dell’s PowerEdge XE9690 servers along with Nvidia’s H100 Tensor Core GPUs.

The AI company is training its foundation models to have improved reasoning and problem-solving abilities, which require large amounts of compute power.

Instead of tapping large cloud providers to rent cloud compute capabilities, Imbue is building its own model from scratch.

The companies stated they designed the system to include smaller clusters which, combined, would enable rapid experimentation for developing AI systems.

Imbue also using a larger Dell network cluster to enable efficient training of large-scale foundation models.

Josh Albrecht, CTO at Imbue, stated Dell “helped us deploy a custom cluster much more quickly than other providers could have as we pursue our work to create AI systems with much stronger reasoning abilities”.

Imbue’s system is managed by cloud compute provider Voltage Park.

The AI start-up completed a $200 million Series-B funding round in September with investors including Nvidia and changed its name from Generally Intelligent to Imbue.

The company stated it was valued at $1 billion after the round.