SambaNova at RAISE Summit 2026

The world's first heterogenous disaggregated inference demo for AI agents was just demonstrated live and will be coming to RAISE. See how GPUs for prefill, SambaNova RDUs for decode, and Intel Xeon processors for orchestration deliver premium inference — faster, more efficient, and at speeds GPUs alone can't reach.

WEDNESDAY - JULY 8 - 5PM

End of Day 1 Keynote

Join Rodrigo's keynote on how SN50 is redefining premium inference with industry-leading performance and economics. We will be hosting a Champaign Reception at booth 15B at the conclusion of Day 1.

JULY 8 & 9

Meet Us at Booth #15B

Visit Booth 15B to see how SambaRacks and our RDU architecture help deliver premium inference for AI agents. Stop by for a demo and conversation with the team.

devos-header
TUESDAY - JULY 7 - 7:30pm CET

Inference Above Paris

Join SambaNova and MiniMax for an evening of networking with AI infrastructure leaders, innovators, investors, and partners shaping the future of premium inference.

TUESDAY - JULY 7

Inference Above Paris

An exclusive rooftop reception with SambaNova and MiniMax. Experience breathtaking views of the Pantheon during this premium networking event.

Register now →

JULY 8 & 9

Meet the SambaNova Team

Meet the SambaNova team to discuss AI agents, inference infrastructure, sovereign AI, and premium inference strategies. Schedule time with our experts during RAISE.

Book a meeting now →

devos-header
Meet with SambaNova at RAISE to experience the architecture that Artificial Analysis verified as the fastest enterprise inference — and learn what it means for premium AI experiences at scale.

PREMIUM INFERENCE FOR AGENTIC AI

AI agents demand fast decode on the largest models — long contexts, hundreds of turns, tool calls between every step. SambaNova RDUs deliver 500+ tok/s decode at speeds GPUs physically cannot reach, turning the decode bottleneck into a competitive advantage.

Learn more →

10X THROUGHPUT, BETTER ECONOMICS

At 500 tokens/sec/user on MiniMax M2.7, a B300 + SN50 configuration generates 10x the throughput of GPU-only decode — lowering cost-to-serve while improving the experience. Faster inference and better margins can reinforce each other.

Learn more →

THE RIGHT CHIP FOR THE RIGHT WORKLOAD

Don't pile more machines into the wrong part of the factory. GPUs excel at prefill. RDUs are purpose-built for decode. Intel Xeon orchestrates the agent loop. Disaggregated inference puts each chip where it performs best.

Learn more →

Related Resources

The First Disaggregated Inference Demo for AI Agents Is Live
Blog

The First Disaggregated Inference Demo for AI Agents Is Live

June 3, 2026
SambaNova and Intel Announce Blueprint for Heterogeneous Inference: GPUs For Prefill, SambaNova RDUs for Decode, and Intel® Xeon® 6 CPUs for Agentic Tools
SambaNova | Intel
News

SambaNova and Intel Announce Blueprint for Heterogeneous Inference: GPUs For Prefill, SambaNova RDUs for Decode, and Intel® Xeon® 6 CPUs for Agentic Tools

April 8, 2026
Introducing the SN50 RDU: Purpose-Built for Agentic Inference
Blog

Introducing the SN50 RDU: Purpose-Built for Agentic Inference

February 24, 2026
sambanova_favicon
Time To Start Building
Let’s Go