SambaNova Raises $1 Billion at $11 Billion Valuation for AI Inference

SambaNova Systems completed the first close of $1 billion in Series F financing at an $11 billion post-money valuation, providing new capital to expand production capacity and deployments of its purpose-built AI inference infrastructure. General Atlantic led the round, with significant investments from Seligman Ventures, T. Rowe Price Associates, and Capital Group, alongside a broad group of new and existing investors.

The financing follows SambaNova’s earlier 2026 announcement of more than $350 million in funding, its SN50 processor introduction, and a strategic collaboration with Intel focused on heterogeneous AI inference infrastructure. SambaNova said it will use the Series F proceeds to increase capacity, accelerate product development, expand customer programs, and scale deployments across enterprises, neoclouds, sovereign AI projects, and service providers.

SambaNova also announced that JPMorganChase selected the company as an AI inference infrastructure partner and plans to deploy SN40 and SN50 systems for secure, on-premises AI inference. The deployment gives SambaNova a major enterprise reference customer for its Reconfigurable Dataflow Unit (RDU) architecture as financial institutions evaluate dedicated inference accelerators alongside GPU-based infrastructure.

• Financing: First close of $1 billion as part of SambaNova’s Series F round.

• Valuation: $11 billion post-money.

• Lead investor: General Atlantic.

• Other significant investors: Seligman Ventures, T. Rowe Price Associates, and Capital Group.

• Additional participants: A&E Investment, Assam Ventures, Battery Ventures, funds and accounts managed by BlackRock, Cambium Capital, Intel Capital, Kabila Capital, QFO Capital, Qatar Investment Authority, Vista Equity Partners, and Volantis.

• Capital deployment: Capacity expansion, product development, customer programs, and global deployments across chips, systems, software, and full-stack AI infrastructure.

• Enterprise deployment: JPMorganChase plans to deploy SambaNova SN40 and SN50 systems for on-premises AI inference.

• Architecture: SambaNova’s RDU platform targets high-throughput inference workloads and competes for deployments where enterprises seek alternatives or complements to GPU-centric AI infrastructure.

“SambaNova’s $11 billion valuation highlights the central role that fast inference now plays in the enterprise AI stack,” said Rodrigo Liang, co-founder and CEO of SambaNova.

🌐 Analysis: The $1 billion first close gives SambaNova substantial capital to scale its RDU architecture as AI infrastructure investment increasingly shifts toward production inference, cost per token, power efficiency, and workload-specific accelerators. The JPMorganChase deployment also strengthens SambaNova’s enterprise position as Intel, AMD, NVIDIA, Google, AWS, Microsoft, and specialized accelerator vendors expand competing inference platforms and heterogeneous computing architectures.

Deep Dive: SambaNova AI Inference Stack
Architecture, memory hierarchy, rack-scale systems and heterogeneous inference strategy • Updated July 8, 2026
Core Architecture	SambaNova builds its AI infrastructure around the Reconfigurable Dataflow Unit (RDU). The architecture maps AI computation graphs onto reconfigurable hardware resources and streams data through the execution fabric rather than relying exclusively on the conventional GPU execution model.
SN50 Generation	The SN50 RDU is SambaNova’s latest announced accelerator generation. The company positions SN50 for high-throughput AI inference, agentic AI workloads and large-scale enterprise deployments.
Primary Design Goal	Reduce the data-movement bottlenecks that limit inference throughput and increase power consumption. SambaNova emphasizes keeping data close to compute resources and efficiently streaming model data through the execution fabric.
Memory Architecture	The SN50 platform uses a tiered memory architecture spanning on-chip memory, high-bandwidth memory and system memory. SambaNova’s software stack manages model placement and data movement across these memory tiers.
Inference Pipeline	SambaNova promotes Disaggregated Inference, separating compute-intensive prompt processing from the memory-intensive token-generation phase and assigning each stage to infrastructure optimized for that workload.
Prefill Stage	Compute-intensive prompt processing can run on existing accelerator infrastructure, including GPU platforms, allowing operators to retain and better utilize existing AI compute investments.
Decode Stage	SN50 RDUs target the memory-bound decode phase of LLM inference, where models repeatedly access weights and KV-cache data while generating tokens sequentially.
Rack-Scale Platform	SambaRack SN50 integrates SN50 accelerators, host processors, networking, memory and SambaNova’s software stack as deployable AI inference infrastructure.
Host CPU	Intel Xeon 6 processors serve as host CPUs in SN50 systems under SambaNova’s collaboration with Intel, supporting system orchestration and general-purpose processing alongside the RDU infrastructure.
System Scaling	SambaNova is expanding deployments from rack-level systems toward larger multi-rack configurations. The company has discussed configurations scaling from 16 RDUs to 128- and 256-RDU clusters for enterprise, service provider and sovereign AI infrastructure.
Data Center Deployment	SambaNova emphasizes air-cooled infrastructure and deployment into conventional enterprise data center environments, reducing dependence on the liquid-cooling systems increasingly associated with high-density GPU clusters.
Software Stack	SambaNova provides an integrated software environment for compiling, mapping, deploying and serving AI models on RDU infrastructure. The full-stack model spans chips, systems, software and AI services.
Enterprise Deployment	JPMorganChase selected SambaNova as an inference infrastructure partner and plans to deploy SN40 and SN50 systems for secure, on-premises AI inference workloads.
Target Markets	Enterprise AI Neoclouds Sovereign AI Service Providers
Strategic Position	SambaNova positions the SN50 stack as specialized inference infrastructure that can operate alongside existing GPU clusters. Its architectural focus centers on dataflow execution, memory efficiency, high-throughput token generation and heterogeneous AI infrastructure.
Source: SambaNova Systems announcements and technical materials. Updated July 8, 2026.