Supermicro will integrate, validate, deliver, and support disaggregated inference systems that pair NVIDIA GPUs with SambaNova SN50 RDUs for agentic AI inference in data centers around the world. The systems will roll out in phases, beginning with an initial batch of 10,000 SN50 chips. They will be distributed first to NeoClouds and sovereign clouds, then expand to enterprises running AI in their own data centers.
The partnership addresses a changing market, as agentic AI is expected to dramatically increase demand for fast, efficient inference. Agents make thousands of calls to LLMs, and each call can take minutes to complete, so even small improvements in inference speed and efficiency are extremely valuable.
LLM inference consists of two phases. The prefill phase processes the prompt and is compute-bound, which is where NVIDIA GPUs excel. The decode phase generates the response one token at a time and is memory-bound, which is what SambaNova SN50 RDUs were built for. Because agentic workloads spend most of their time in decode, SN50 RDUs are particularly well suited to accelerating them.
That is the logic behind the new systems. By using the best hardware for each phase, disaggregated inference with SN50 RDUs and NVIDIA GPUs delivers faster and more efficient inference than either type of hardware alone. And because SN50 RDUs run high-speed, high-throughput inference of large models without requiring extremely high rack power densities, customers can deploy them in existing data centers.
“Supermicro builds AI infrastructure for data centers globally,” said Rodrigo Liang, SambaNova co-founder and CEO. “Now customers can order SN50 RDUs alongside GPUs, from a world-leading partner they trust. That’s how fast, efficient inference gets to scale.”
“Supermicro is committed to delivering flexible, high-performance AI infrastructure that supports the evolving requirements of our customers. Our collaboration with SambaNova expands our broad DCBBS portfolio by enabling heterogeneous architectures that integrate GPU computing with specialized inference acceleration, helping customers deploy efficient, scalable AI solutions,” said Charles Liang, president and CEO of Supermicro.
Getting the systems to customers is where Supermicro comes in. It will handle integration, validation, delivery, and support, including on-site service. SambaNova engineers will work alongside Supermicro on customer deployments and will ensure that models run well and are tuned for best performance on SN50 RDUs.
For customers, the net effect is simplicity. They can now purchase SN50 RDUs the same way they purchase the rest of their AI infrastructure: from a trusted channel partner. SambaNova expects this partnership will result in significantly more high-performance inference capacity being deployed in 2027 than would otherwise be possible.