SambaNova Systems raised $1 billion in the first close of its Series F round on July 8, 2026, valuing the custom AI chip company at $11 billion and simultaneously announcing JPMorgan Chase as an inference-infrastructure partner that will deploy its hardware on-premises inside the bank’s own data centers. The round was led by General Atlantic, with participation from Intel Capital, BlackRock, and the Qatar Investment Authority, and comes just five months after SambaNova’s previous major fundraise — a pace that reflects how rapidly enterprise demand for AI inference hardware is scaling.
What Happened
SambaNova disclosed the Series F first close on the same day it announced JPMorgan Chase as a deployment partner — a dual announcement that signals both capital strength and a marquee commercial win. JPMorgan Chase will deploy SambaNova’s SN40L systems alongside the next-generation SN50 when it begins shipping to customers in the second half of 2026. SoftBank was previously named as the SN50’s first deployment partner. The Series F remains open; additional investors are expected to join in a second close in the coming weeks.
Intel Capital, a long-time SambaNova backer, remains an investor in the new round alongside newer participants including Cambium Capital, BlackRock, the Qatar Investment Authority, A&E Investment, Assam Ventures, Battery Ventures, Kabila Capital, QFO Capital, Vista Equity Partners, and Volantis. The involvement of BlackRock and the Qatar Investment Authority is significant: both are sovereign-adjacent capital allocators whose participation signals growing institutional conviction in AI inference infrastructure as an asset class.
Why It Matters
JPMorgan Chase’s selection of SambaNova for on-premises inference rather than a hyperscaler cloud service is a pointed commercial endorsement. Financial institutions operate under strict data residency and audit requirements that complicate public cloud AI deployments. Running models inside the bank’s own data centers gives JPMorgan Chase full control over where customer data travels, which regulators can audit what, and how inference workloads are logged — controls that cloud inference cannot replicate with the same fidelity. SambaNova’s architecture was designed with this regulated-enterprise use case in mind, and a partnership with one of the world’s largest banks by assets creates a credibility signal that will resonate with other banks, insurers, healthcare systems, and government agencies facing the same constraints.
The funding round also reflects escalating competition across the AI silicon market. Anthropic is in early talks with Samsung to develop a custom 2nm AI chip, while OpenAI partnered with Broadcom to announce its Jalapeño inference processor. SambaNova occupies a different niche — not an AI lab building chips for its own models, but a specialized hardware company selling inference infrastructure to enterprises that want to run AI at scale without depending on Nvidia or a public cloud.
Background and Context
SambaNova was founded in 2017 by Stanford researchers and former Oracle executives with a thesis that general-purpose GPU architectures carry significant inefficiency for AI inference, where the model structure is fixed and the bottleneck shifts from pure computation to moving data between memory and compute units. The company’s reconfigurable dataflow architecture addresses this memory-bandwidth problem at the hardware level, producing chips that are more efficient than GPUs for the specific patterns of inference workloads. Its SN40L is already in production deployments; the SN50, unveiled in February 2026, extends that architecture to handle even larger models at lower latency.
The $11 billion valuation reflects the broader premium that AI infrastructure plays now command. Google’s AI buildout drove a record 37 percent surge in its electricity consumption last year, a figure that illustrates how expensive running AI at scale has become and why enterprises are willing to pay significant premiums for hardware that can run inference more efficiently. For SambaNova, each improvement in inference efficiency translates directly into lower operating costs for its customers — a compelling value proposition in an environment where AI compute bills are growing faster than most IT budgets.
What Comes Next
SambaNova’s CEO indicated the company is fielding acquisition inquiries but that its current growth trajectory points toward an eventual IPO. A public offering would subject the company to a rigorous accounting of customer revenue, margins, and competitive positioning relative to both Nvidia and the growing number of custom-chip efforts from AI labs themselves — a test that the company appears confident it can pass given the JPMorgan Chase validation.
The enterprise AI infrastructure market is consolidating quickly. Microsoft’s $2.5 billion Frontier Company initiative embeds AI engineers directly inside enterprise customers, creating sustained demand for reliable, high-performance on-premises inference hardware from vendors those customers can trust for regulated workloads. As enterprises move from AI pilots to production deployments at scale, hardware choices for inference are becoming strategic decisions locked in for years — and SambaNova, with JPMorgan Chase as a reference customer, is positioning itself to win that enterprise conversation before its competitors can.
