RDU
Reconfigurable Dataflow Units (RDUs) — purpose-built for AI
Delivering fast & energy efficient AI inference
Our fourth-generation RDU, the SN40L, is designed to deliver extraordinary performance. Uniquely capable of powering the most demanding workloads, the SN40L does it all with the highest efficiency rate.

Headline here

From chips to racks
The combination of 16 SN40L RDUs creates a single, high-performance rack that can run the largest models, such as DeepSeek R1 671B and Llama 4 Maverick, with the fast inference. These racks can be seamlessly integrated into any existing air-cooled data center.
Learn more →
Seamlessly achieve high performance
From chips to racks
The combination of 16 SN40L RDUs creates a single, high-performance rack that can run the largest models, such as DeepSeek R1 671B and Llama 4 Maverick, with fast inference. These racks can be integrated seamlessly into any existing air-cooled data center.
Learn more →
Dataflow architecture
Our innovative compute and memory chip layout enables seamless data flow between operations when processing AI models. This approach results in high-speed data traffic and significant gains in performance and efficiency.
Learn more →
Three-tier memory for efficiency
The SN40L design enables multiple models to run in memory and switch models in microseconds. This unique layout enables SambaNova to scale to the largest models, like DeepSeek and Llama 4 — all on a single rack.
Learn more →Future-proof your infrastructure
Our fourth-generation RDU SN40L is the heart of the SambaNova solution platform.

Speed
RDUs are the only solution that run the largest AI models on a single system with blazing fast performance.
Learn more →
Energy
RDUs deliver the highest tokens per kilowatt-hour, which is ideal for data centers of all sizes.
Learn more →
Agentic
Three-tier memory architecture enables multiple models to run while switching between them. Perfect for AI agents.
Learn more →Related resources
