Skip to content

SambaNova Pulls In $1B — and JPMorgan Is Already Running Its Chips

Category
News
Published
Reading time
4 min

SambaNova Systems closed a $1B Series F at an $11B valuation — five times higher than five months ago. JPMorganChase picked its SN40L and SN50 chips for secure on-premises AI inference.

Five months ago, SambaNova Systems closed a $350 million Series E. Now it's back with a $1 billion Series F at an $11 billion valuation — a fivefold jump, with a headline-grade customer to show for it.

JPMorganChase has named SambaNova its inference infrastructure partner, deploying SN40L and SN50 chips directly on the bank's own servers. No data leaves the building. For an institution handling transactions for hundreds of millions of customers, that's not a technical preference — it's a compliance requirement that off-premises cloud compute can't satisfy.

General Atlantic led the round, joined by Intel Capital, BlackRock, the Qatar Investment Authority, Battery Ventures, Vista Equity Partners, and several others. Intel CEO Lip-Bu Tan chairs SambaNova's board — a signal this is more than a portfolio bet. Two chipmakers in strategic alignment against a common rival.

Nvidia owns model training, but inference — where a trained model does its actual job every day — is more contested ground. Two years of H100 and H200 scarcity pushed enterprises to rethink single-supplier dependence. SambaNova built its SN50 squarely for that opening: token-efficient, on-premises, designed to slot into existing data centers without a full overhaul.

Proceeds go toward production expansion, faster product development, and scaled deployments. CEO Rodrigo Liang has signaled IPO ambitions. The Series F second close is still open. Whether SambaNova can cut into Nvidia's enterprise hold is now a question of execution, not positioning.

Questions and answers

Frequently asked questions about this article

SambaNova Systems is a Palo Alto company that makes chips for AI inference — running already-trained models efficiently inside enterprise data centers. Nvidia dominates model training; SambaNova targets the inference market where enterprises need on-premises, cloud-independent AI compute.