← All companies

Chips / Company profile

SambaNova

What SambaNova does

SambaNova is a private AI infrastructure company that designs a full stack for fast enterprise AI inference, centered on its Reconfigurable Dataflow Unit (RDU) accelerators and a dataflow software stack that maps models onto the hardware. Its commercial offerings span purpose-built on-prem and dedicated deployments (SambaStack, SambaRack), developer-oriented inference in the cloud (SambaCloud), and managed inference options intended to let data centers operationalize an “AI inference cloud” without building and operating the entire system themselves (SambaManaged).

News

Company record

Jul 22, 2026 · Official · SambaNovaSambaNova Joins the DOE Genesis Mission Consortium

SambaNova announced it joined the Genesis Mission Consortium, a public-private partnership led by the U.S. Department of Energy, contributing its RDU chip and full-stack inference expertise for national science, security, and energy challenges.

Jul 08, 2026 · Official · SambaNovaSambaNova Completes First Close of $1B Financing at $11B Valuation

SambaNova announced the first close of a $1 billion strategic Series F financing, valuing the company at $11 billion post-money, and referenced JPMorganChase selecting SambaNova RDUs for fast on-prem AI inference.

Jul 08, 2026 · Official · SambaNovaSN50 Runs the Fastest MiniMax Speeds in the World (RAISE Summit 2026 demo)

SambaNova reported a heterogeneous, disaggregated inference demo combining an NVIDIA H200 rack for prefill with a SambaRack SN50 (16 RDU chips) for decode, sharing decode-speed results for MiniMax M2.7 benchmarks at RAISE Summit 2026.

2026-06-?? · Official · SambaNovaThe First Disaggregated Inference Demo for AI Agents Is Live

SambaNova described an agentic inference demo using GPUs for prefill and SambaNova’s SN40 RDU for decode, reporting 2X inference speed versus B200-only configurations.

May 11, 2026 · Official · SambaNovaBuild Faster Coding Agents with SambaNova’s Responses API

SambaNova announced support for a Responses API across SambaCloud, SambaStack, and SambaManaged, describing initial /v1/responses support and listing model starters for coding-agent workflows.

Apr 08, 2026 · Official · SambaNovaSambaNova and Intel Announce Blueprint for Heterogeneous Inference

SambaNova and Intel described a heterogeneous inference design using GPUs for prefill, SambaNova RDUs for decode, and Intel Xeon processors as host/action CPUs for agentic tools and enterprise integration.

Feb 24, 2026 · Official · SambaNovaSambaNova Unveils Fastest Chip for Agentic AI, Collaborates with Intel, and Raises $350M+

SambaNova introduced the SN50 AI chip, announced a collaboration blueprint with Intel for heterogeneous inference, and disclosed a $350M+ oversubscribed Series E led by Vista Equity Partners and Cambium Capital.

2026-??-?? · Official · SambaNovaIntroducing the SN50 RDU: Purpose-Built for Agentic Inference

SambaNova announced its fifth-generation SN50 RDU and SambaRack SN50 system, positioning the SN50 for agentic inference and noting shipping timing mentioned in the post.

Source map · 2 recurring channels · 16 references

Still resolving: Newsroom · Official X

Funding

Latest disclosed valuation$11B
Tracked capital$2.2B5 sourced rounds

What it builds

SambaRack SN50

SambaRack SN50 is SambaNova’s purpose-built, air-cooled inference rack that integrates 16 SN50 RDU chips and is designed for agentic inference workloads, including scaling out across multiple racks under SambaNova’s full-stack solution.

source ↗
SambaCloud

SambaCloud is SambaNova’s cloud inference platform powered by its RDU hardware, positioned for fast inference on supported open-source models and developer/enterprise integration.

source ↗
SambaStack

SambaStack is SambaNova’s full-stack enterprise AI platform that combines SambaNova hardware and software into an integrated solution for inference, supporting deployment either on-premises or in the cloud.

source ↗
SambaManaged

SambaManaged is positioned as a fully managed inference cloud intended to let organizations run a high-speed AI inference cloud from their data center infrastructure, with SambaNova managing deployment and ongoing operation.

source ↗
Reconfigurable Dataflow Unit (RDU) / SN50

The SN50 RDU is SambaNova’s fifth-generation purpose-built inference processor for agentic workloads, using its Dataflow Architecture and tiered memory to support large models and fast decode.

source ↗
SambaFlow

SambaFlow is SambaNova’s software component of its Reconfigurable Dataflow Architecture, designed to help map model dataflow graphs onto the RDU architecture with reduced low-level tuning compared with other approaches.

source ↗

Key metrics

Milestones & partnerships