SambaNova Systems just secured $1 billion in the first close of its Series F funding round, pushing its valuation to $11 billion. This marks not only another vote of confidence in AI chipmakers but also signals a shift in demand toward specialized enterprise AI systems that run large models on-premises, reducing reliance on public clouds.
The funding round is led by General Atlantic, with more investors expected when the round fully closes. Based in Palo Alto, SambaNova says the new capital will fuel business expansion and strengthen its supply chain. With AI demand skyrocketing faster than many can scale production, this timing reflects the growing urgency in the 2026 AI infrastructure market.
SambaNova designs custom AI accelerators and turnkey systems optimized for running large AI models. This latest round arrives just months after the launch of the SN50 accelerator and a $350 million Series E raise. Investors are rewarding infrastructure players with concrete hardware products and established enterprise clients, not vague AI concepts.
One prominent customer is JPMorgan Chase, which selected SambaNova’s SN40L and SN50 systems to perform AI inference-processing trained models on new data-within its secured infrastructure. For industries like banking, pharma, and government agencies, deploying generative AI internally without sending sensitive data to the public cloud is critical.
SambaNova’s pivot and enterprise AI accelerator competition
Interestingly, SambaNova once contemplated a $1.6 billion acquisition by Intel, but the deal never materialized. Now Intel is an investor and technology partner in the new round, collaborating on AI inference solutions integrating SambaNova accelerators with Intel Xeon processors.
This signals a shift from the notion that Nvidia dominates all AI accelerator turf. Enterprise clients increasingly want localized systems with predictable total cost of ownership, support for specific models, and smooth integration into existing data centers. Beyond SambaNova, startups like Groq and Cerebras-and several newer inference chipmakers-are also vying for this specialized niche.
CEO and co-founder Rodrigo Liang emphasizes SambaNova’s focus on density: their systems target trillion-parameter models, packing massive computational load into a single data center rack. For buyers, practical metrics matter-space, power, and required personnel-not just flashy specs.
The product roadmap is straightforward: the SN40L launched in 2023, with SN50 shipments planned for late 2024. Key customers and partners include Saudi Aramco, Intel, and several Japanese firms. As the customer list grows, SambaNova aims to sell not just chips but full infrastructure solutions with higher margins.
Amid the generative AI boom, SambaNova positions itself between cloud giants and traditional server vendors. Industry analysts estimate Nvidia still commands the majority of AI accelerator market share for data centers, but growing inference demand is pushing enterprises to seek alternatives. If SambaNova scales SN50 production and secures its footing with banks, oil & gas, and government clients, the next funding round will likely test whether it can carve out a sizable share from the market leaders.