JPMorgan Chase Taps SambaNova for Secure On-Premises AI Inference in Multi-Year Deal #

JPMorgan Chase has selected AI chip startup SambaNova Systems as its inference-infrastructure partner, deploying the company’s SN40 and SN50 Reconfigurable Data Unit (RDU) systems to run AI workloads inside the bank’s own facilities. The multi-year agreement was announced July 8 at the RAISE summit in Paris, alongside SambaNova’s disclosure of a $1 billion Series F funding round.

Darrin Alves, JPMorgan Chase’s Chief Infrastructure Officer, said the bank is “excited to deploy SambaNova’s RDU architecture” and will evaluate its speed and security characteristics across enterprise AI workloads, according to the company’s announcement. JPMorgan chose on-premises hardware rather than a public cloud service, keeping models and data within its own facilities.

SambaNova CEO Rodrigo Liang told TechCrunch the deal “sends a message to the banking industry that it’s time not to completely depend on cloud services,” adding that major banks are investing in private, heterogeneous infrastructure to complement existing cloud commitments. He cited larger AI models, data-privacy demands, and on-site power constraints as factors driving demand for on-premises hardware.

The announcement came alongside news of SambaNova’s latest fundraise. The Palo Alto-based company closed the first tranche of its Series F at a post-money valuation of $11 billion, up from the roughly $1.6 billion acquisition price Intel had reportedly considered earlier in the year. General Atlantic led the round, with significant commitments from Seligman Ventures, T. Rowe Price Associates, and Capital Group. BlackRock, Intel Capital, the Qatar Investment Authority, Vista Equity Partners, and Battery Ventures also participated. SambaNova said a second close is expected in the coming weeks.

The company plans to use the proceeds to expand manufacturing capacity, accelerate chip and software development, and grow deployment programs for enterprises, neo-cloud operators, sovereign AI projects, and managed service providers globally. SambaNova’s enterprise customers include Saudi Aramco and several Japanese companies. The company competes with Nvidia for organizations that need dedicated on-site inference hardware rather than GPU-based cloud capacity.

Source: TechCrunch