SambaNova Completes $100M Funding, Valuation Reaches $11B

SambaNova has closed fresh financing and landed major enterprise customers as demand for AI inference infrastructure surges. The company said it secured $100 million in project funding that contributed to a post-money valuation of $11 billion, and also announced completion of a $1 billion Series F financing round that it says places the firm at an $11 billion valuation. The San Jose‑based startup, founded in 2017, likewise confirmed a multi‑year agreement with Intel and that JPMorgan Chase will deploy SambaNova’s SN40 and SN50 systems for secure on‑premises AI inference.

"JPMorgan Chase’s AI infrastructure must meet high standards for performance, control, and reliability," said Darrin Alves, CIO of JPMorgan Chase’s infrastructure platform, underscoring the enterprise requirements that guided the bank’s selection of SambaNova technology.

SambaNova designs a custom RDU (Reconfigurable Dataflow Unit) rather than relying on GPU architectures. The company argues that its dataflow approach reduces costly memory transfers that typically slow and power AI inference on traditional GPUs. The SN50, described as the fifth‑generation RDU, is pitched for "agentic AI inference" use cases such as AI coding agents, enterprise copilots, model hosting, and large‑scale inference services; it can also be scaled via the SambaRack system to support larger models, longer contexts, and multi‑model switching.

The startup’s leadership mixes industry and academic experience: CEO Rodrigo Liang, born in Taipei and raised in Brazil, previously worked on high‑performance processors at Hewlett‑Packard, Sun Microsystems and Oracle. Co‑founders include Stanford professors Kunle Olukotun and Christopher Ré, noted respectively for work in multicore processor design and machine learning, data systems and AI infrastructure.

SambaNova’s path to the current valuation has been uneven. The company raised $676 million in a 2021 Series D led by SoftBank Vision Fund 2, pushing valuation above $5 billion, but later faced fundraising challenges, layoffs and a period in which Intel reportedly negotiated a potential acquisition in the roughly $1.6 billion range that did not materialize. The subsequent pivot toward optimized inference, cloud services and enterprise deployment — along with the JPMorgan order and the Intel partnership — form the basis of the company’s case that market sentiment has rebounded.

• Partnerships and customers: JPMorgan Chase (on‑prem SN40 and SN50 deployments), multi‑year agreement with Intel.
• Product focus: RDU chip architecture, SN50 for agentic inference, SambaRack for scaling.
• Funding and valuation: reported $100 million in project funding, completion of a $1 billion Series F, $11 billion post‑money valuation.

The broader AI inference market is attracting significant capital. In September 2025, Groq raised $750 million, lifting its valuation to $6.9 billion and securing a $1.5 billion commitment from Saudi Arabia for delivery of inference chips. Cerebras — now public — has a market capitalization cited around $49 billion and reported $510 million in revenue last year. Baseten raised $1.5 billion at a $13 billion valuation and reported revenue growth of 20‑fold over the prior year, reflecting heightened demand for inference services.

Outlook: SambaNova is positioning itself as an enterprise‑grade alternative to GPU‑centric inference stacks, arguing that a heterogenous approach — CPUs orchestrating, GPUs handling prompt ingestion, and RDUs driving low‑latency token generation — will allow customers to deploy production AI without rebuilding data centers. Whether that vision translates into sustained commercial traction will depend on broader enterprise adoption, the company’s ability to scale deployments like JPMorgan’s, and continued investor appetite for inference‑focused hardware and software ecosystems.