SambaNova Closes $1 Billion Series F Round, Valuing AI Hardware Firm at $11 Billion

A second close of SambaNova's Series F is expected in the coming weeks, signaling continued investor interest beyond the initial $1 billion tranche.
The round includes a broad roster of notable investors beyond the lead, such as BlackRock, Intel Capital, the Qatar Investment Authority, Battery Ventures, Vista Equity Partners and Volantis.
JPMorgan Chase will deploy SambaNova's SN40 and SN50 systems for on-premise AI workloads, with the bank emphasizing the need for high performance, control and reliability in AI infrastructure.
SambaNova asserts that its SN40 and SN50 chips can decode the inference step five to ten times faster than competing approaches, potentially freeing capacity for other tasks.
Intel figures prominently in the governance of SambaNova, with Lip-Bu Tan serving as chairman; reports note that Intel and Foxconn have joined the company, underscoring a direct tie to major hardware incumbents.
SambaNova Systems has closed the first tranche of its Series F round at $1 billion, valuing the AI chip startup at $11 billion post-money, according to Quartz. General Atlantic led the round, with notable participation from Seligman Ventures, T. Rowe Price, Capital Group, BlackRock, Intel Capital, and the Qatar Investment Authority. The close marks a fivefold jump from SambaNova's estimated $2.2 billion valuation just five months earlier in February 2026.
Alongside the funding, SambaNova announced that JPMorgan Chase will deploy its SN40 and SN50 chip systems on-premises for private AI workloads. A second close of the Series F is expected in the coming weeks, signaling continued investor appetite, Yahoo Finance reported.
SambaNova's rise has been dramatic. In late 2025, Intel tried to acquire the company for roughly $1.6 billion — including debt. Those talks stalled. SambaNova then raised a $350 million Series E in February 2026 at around $2.2 billion. Now, five months later, it has closed a $1 billion Series F at $11 billion. That is a sevenfold jump from the failed Intel bid.
Instead of being absorbed, SambaNova partnered with Intel. At Computex 2026 in Taipei, Intel CEO Lip-Bu Tan — who also serves as SambaNova's Executive Chairman — announced a three-way alliance with Intel, SambaNova, and Foxconn to build and sell rackscale AI systems. Tan said Intel is "poised to bring the world new innovations from the chip to systems level." He recused himself from all internal deal negotiations to avoid a conflict of interest.
SambaNova builds what it calls Reconfigurable Dataflow Units, or RDUs. Unlike standard graphics chips, RDUs map algorithms directly onto physical circuits. This cuts power use and speeds up inference — the step where an AI model actually generates answers. SambaNova says its SN40 and SN50 chips decode inference steps five to ten times faster than competing chip approaches.
The SN50 chip supports AI models with over 10 trillion parameters and context lengths above 10 million tokens. SambaNova also claims a 3x reduction in total cost of ownership compared to GPU-based systems. CEO Rodrigo Liang said the company is racing to meet "an incredible wave of demand" ahead of SN50 shipments in the second half of 2026.
JPMorgan Chase has chosen SambaNova's SN40 and SN50 systems to run AI workloads inside its own data centers — not on public cloud services. Banks handle highly sensitive data. Running AI on-premises keeps that data behind the bank's own firewalls. It also gives the bank direct control over performance and reliability.
Liang called the JPMorgan deal a milestone. "Having JPMorgan Chase decide they're going to use SambaNova for their inference solution is a big deal," he said. "It sends a message to the banking industry that it's time not to completely depend on cloud services." The global AI inference market is projected to grow from $97 billion in 2024 to $254 billion by 2030, according to Grand View Research.
The SambaNova raise reflects a broader investor push to back chips that can challenge Nvidia — especially for inference. Training AI models is Nvidia's stronghold. But running those models in production can cost 20 times the original training bill. That creates huge demand for cheaper, faster inference chips. General Atlantic Co-President Martín Escobari said "demand for inference is accelerating well ahead of supply."
SambaNova's valuation tripling in months has drawn skepticism from some analysts, who note the company has disclosed no revenue figures. But the Intel and Foxconn alliances provide hardware validation that many pure-software AI startups lack. Liang has also hinted publicly at a future IPO, keeping acquisition offers and an independent path both on the table, Quartz reported.
Publishers
42
Articles
74
Reach
116