Supermicro will sell SambaNova’s inference chips alongside Nvidia GPUs

Supermicro will build and sell AI inference systems that pair SambaNova’s SN50 chips with Nvidia GPUs, the companies said on Tuesday. The collaboration initially plans for 10,000 SN50 chips, according to SambaNova’s announcement.

The racks will join Supermicro’s Data Center Building Block Solutions, the ready-made data centre systems it sells to customers.

Running an AI model has two stages. First the system reads the prompt, then it writes the answer piece by piece. In the new racks, Nvidia GPUs do the reading and SN50 chips do the writing, SambaNova said. The company says AI agents work nonstop, so how fast they act depends on how fast the model responds.

“Agentic AI is the next wave of inference, and it has to run everywhere,” said Rodrigo Liang, co-founder and chief executive of SambaNova. “Now customers can order SN50 RDUs alongside GPUs, from a world-leading partner they trust.”

Who does what

Supermicro will put the systems together, test them, deliver them and support customers, including with on-site service. SambaNova engineers will get AI models running and tuned on SN50 and help with deployments.

The companies will start with neoclouds, the cloud providers built for AI, and with sovereign clouds. Those keep data and computing inside one country. They then plan to sell to companies that run AI in their own data centres.

“Supermicro is committed to delivering flexible, high-performance AI infrastructure that supports the evolving requirements of our customers,” said Charles Liang, president and chief executive of Supermicro.

Built for existing data centres

SN50 is SambaNova’s fifth-generation Reconfigurable Dataflow Unit, or RDU, a chip designed for inference. The company says it can serve the largest models without needing extreme power density in each rack. A SambaRack SN50 draws an average of 20 kW, SambaNova said last week.

Operators can add inference capacity in data centres they already run, according to SambaNova. They do not have to wait for new builds held back by power and cooling. It says they can also keep using their existing GPUs for longer. The two companies say they can deliver much more inference capacity in 2027 as a result.

SambaNova’s other recent deals

The Supermicro deal follows another one with Oracle. Oracle said at its AI World conference on 27 October that it plans to deploy SN50 chips in its cloud data centres. It will offer the capacity to AI model builders.

Two days later, SambaNova signed Emergys, a digital transformation services company, to deploy and run its platform for enterprise customers.

SambaNova, founded in 2017 and based in San Jose, raised new funding at an $11 billion valuation in July. In June, Foxconn and Intel joined it to build rackscale AI infrastructure.