SambaNova and Supermicro are partnering to build and sell AI inference systems combining SambaNova’s SN50 chips with Nvidia GPUs, with Supermicro initially planning for 10,000 SN50 chips.
The architecture splits inference workloads between two types of processors. Nvidia GPUs handle prefill, processing incoming prompts, while SambaNova’s fifth-generation reconfigurable dataflow units handle decode, generating responses. SambaNova said the combination can improve performance and efficiency while extending the useful life of existing GPUs.
"Agentic AI is the next wave of inference, and it has to run everywhere," said Rodrigo Liang, co-founder and CEO of SambaNova. "Supermicro builds AI infrastructure for data centers globally. Now customers can order SN50 RDUs alongside GPUs, from a world-leading partner they trust.”
Supermicro will integrate, test, deliver and support the systems for customers, including providing on-site service. SambaNova engineers will get models running and tuned on SN50 and help Supermicro with deployments. The companies will start with NeoClouds and sovereign clouds, then expand to enterprises that run AI in their own data centers.
Read the full press release here.