SambaNova banks $1bn at $11bn valuation as inference chips heat up

The News

SambaNova Systems has pulled in $1 billion in the first close of a Series F round that values the chip designer at $11 billion, the Palo Alto company confirmed on 8 July. The raise lands barely five months after its $350 million Series E in February, an unusually short gap that underlines how fiercely investors are chasing exposure to AI inference silicon.

General Atlantic led the round. The cap table reads like a roll-call of global capital: BlackRock, T. Rowe Price, Capital Group, Vista Equity Partners, Battery Ventures, the Qatar Investment Authority and, notably, Intel all took part. That last name is striking, because Bloomberg reported in December 2025 that Intel had explored buying SambaNova outright for roughly $1.6 billion before the talks stalled. The chipmaker is now a shareholder rather than an owner.

Founded in 2017, the nine-year-old firm designs specialised chips tuned for running large AI models rather than training them. Its SN40L launched in September 2023, and its newer SN50, unveiled in February, begins shipping in the second half of 2026, with SoftBank as the first deployment partner. Chief executive and co-founder Rodrigo Liang said the money would go toward locking down supply: "We're using that capital to secure the supply chain," he said, framing it as essential to filling orders over the coming 12 months.

Why It Matters

For most of the current cycle, the AI hardware story has been a single-name story: Nvidia. SambaNova's valuation, up sharply in months, is evidence that capital is now hunting aggressively for the second and third names, particularly in inference, the workload that dominates once a model is actually deployed at scale.

The shape of the round matters as much as the size. When a rumoured acquirer ends up writing a minority cheque instead, it usually signals that the target sees a bigger prize going it alone. The last time a chip challenger commanded this kind of pre-revenue enthusiasm, it foreshadowed a wave of custom-silicon spending by hyperscalers. SambaNova is betting that enterprises wanting to run models on their own premises, rather than renting cloud GPUs, are a large and underserved market. Its named customers, from JPMorgan Chase to Saudi Aramco, suggest regulated and sovereign buyers are the early believers.

Indian Angle

On-premises inference is precisely where India's regulatory reality bites. RBI data-localisation rules and the sectoral caution of Indian banks and insurers have made cloud-only AI a hard sell for the most sensitive workloads. A chip family built to run large models inside a customer's own data centre maps neatly onto that constraint, and SambaNova's JPMorgan deployment is the kind of reference an Indian private bank will study closely.

There is a sovereign dimension too. The IndiaAI Mission has committed to pooling tens of thousands of GPUs for domestic developers, and MeitY has leaned heavily on Nvidia-class imports to get there. A credible inference-only alternative, especially one backed by a sovereign fund like QIA, gives Indian procurement teams a reason to diversify away from a single supplier and hedge against export controls and rupee-denominated cost spikes.

For domestic chip ambitions, the read is double-edged. Efforts such as Ola's Krutrim, which is building both models and silicon, now have a fresh $11 billion benchmark for what focused inference hardware can be worth, useful for fundraising, sobering on execution.

FAQ

How much did SambaNova raise and at what valuation?

$1 billion in the first close of a Series F, at an $11 billion valuation, led by General Atlantic and announced on 8 July 2026.

Why is Intel's involvement notable?

Intel reportedly tried to acquire SambaNova for about $1.6 billion in December 2025. That deal stalled, and Intel has instead joined this round as a minority investor.

What does SambaNova actually sell?

Specialised inference chips, the SN40L and the newer SN50, designed to run large AI models efficiently, often inside a customer's own data centre rather than in the public cloud.

Why should Indian enterprises care?

On-premises inference suits RBI data-localisation rules and sovereign-compute goals, offering banks, insurers and the IndiaAI Mission an alternative to cloud-only, single-vendor GPU stacks.

Where can I read the original announcement?

TechCrunch published the first detailed report of the Series F first close.

This story was reported by TechCrunch. Read the full original coverage at TechCrunch.