SambaNova Valuation Explodes to $11B After Intel Rumor

Five months after SambaNova Systems unveiled its SN50 chip and raised $350 million in Series E funding, the company is now raising $1 billion at an $11 billion valuation, a violent repricing for a startup reportedly discussed as an Intel acquisition target at roughly $1.6 billion in December.

That gap is the story. The SambaNova valuation is not just a funding headline. It signals how aggressively investors and enterprise buyers are repricing private AI infrastructure when the asset is scarce, strategic, and tied to inference rather than just model training. The round was led by General Atlantic in the first close of SambaNova’s Series F, according to TechCrunch, with more investors expected to join soon.

SambaNova's $11B valuation turns a near-sale rumor into an AI chip land grab

The timing makes the deal sharper. In February 2026, the Palo Alto company announced its SN50 chip alongside a $350 million Series E. By July 8, 2026, it had locked in a first close of $1 billion at an $11 billion post-money valuation.

That is roughly five months between mega rounds. It also follows a December Bloomberg News report, cited by TechCrunch, that Intel had been in acquisition talks with SambaNova around a deal valuing the company at roughly $1.6 billion.

If those talks were serious, the market’s view of SambaNova changed fast. Either Intel was looking at a cheaper strategic asset before the next funding wave hit, or investors now believe the company’s position in AI inference deserves a far richer multiple. Both can be true.

SambaNova CEO and co-founder Rodrigo Liang did not frame the new financing as a declaration of permanent independence. Asked whether the Series E and Series F meant the company had decided against a sale, he left the door open.

“We’re always being approached.”

He also told TechCrunch that growth and momentum would most likely push SambaNova toward “being public at some point.”

XOOMAR analysis: that answer matters. SambaNova is not behaving like a company trying to disappear into a strategic buyer at yesterday’s price. It is using private capital to buy time, supply, customer credibility, and optionality. The SambaNova valuation now makes any acquisition conversation much more expensive.

July 8 funding math: $1B raised, $11B valuation, and a five-month repricing

The financing itself is unusually dense with signals.

The first close includes General Atlantic, Seligman Ventures, T. Rowe Price Associates, and Capital Group. TechCrunch also lists new and existing investors including A&E Investment, Assam Ventures, Battery Ventures, Cambium Capital, BlackRock, Kabila Capital, QFO Capital, Qatar Investment Authority (QIA), Vista Equity Partners, and Volantis. Intel also participated.

Liang said more capital is still expected.

“In the next few weeks, a few more investors will be coming in, and the second close is likely to finish up,”

An $11 billion valuation changes the burden of proof. At that level, investors are not just underwriting a promising chip. They are underwriting a company that must turn into a scaled infrastructure supplier. That means repeat enterprise deployments, stronger gross margin evidence, reliable supply, and customer concentration risk that does not make public-market investors flinch.

The proceeds are aimed at capacity and delivery, not vague corporate expansion. Liang told TechCrunch the money will help secure materials and fulfill orders over the next 12 months.

“We’re using that capital to secure the supply chain,”

That supply-chain focus lines up with a broader hardware reality: chip ambition is constrained by manufacturing access, advanced packaging, inspection, and long supplier commitments. XOOMAR readers tracking that constraint should also see our coverage of €91M Bet Pushes QuantumDiamonds Chip Inspection Into Fabs and Apple Broadcom Deal Pulls $30B Chip Bet Back to U.S., two separate examples of how strategic chip capacity and production quality have become board-level issues.

XOOMAR analysis: the risk is that the SambaNova valuation becomes a trap. A big private round can solve procurement and hiring problems, but it also raises the failure threshold. If SN50 shipments slip, if customers test but do not expand, or if the software stack fails to hide hardware complexity, the same valuation that gives SambaNova credibility could later narrow its exit path.

February's SN50 launch put SambaNova directly into the inference fight

SambaNova’s pitch centers on “premium inference”, Liang’s phrase for running the largest models quickly. He told TechCrunch that today’s frontier models span trillions of parameters and that SambaNova was built to handle them at that scale.

The company says its systems can fit multi-trillion-parameter models onto a single rack, helping them run quickly. Its next-generation SN50, unveiled in February 2026, is scheduled to begin shipping to customers in the second half of 2026. SoftBank is the first deployment partner.

This is where the deal moves from financing story to infrastructure strategy. Training large models gets the glory, but inference is where many enterprises experience AI as a recurring operating cost and deployment challenge. SambaNova is trying to sell into that moment with systems designed for private, high-performance inference rather than relying only on cloud-hosted AI services.

The company names three customer categories:

• Sovereign clouds: Governments funding local partners to build private clouds.
• Neoclouds: Specialized cloud providers focused on AI infrastructure.
• Enterprises: Companies building AI systems for their own use.

Its named customers include JPMorganChase, Saudi Aramco, Intel, and other Japanese firms.

Software will decide how much of this promise survives procurement. A custom AI accelerator can look strong on paper, but enterprise buyers need model support, compiler maturity, deployment tooling, and operational reliability. They won’t rebuild their AI operations just to accommodate a chip vendor’s preferred architecture.

XOOMAR analysis: this is the hardest part of challenging incumbent compute stacks. Hardware gets funded. Software gets adopted only when it reduces friction. SambaNova’s value proposition depends on whether customers can bring real workloads over without treating every deployment as a science project.

Intel's role now looks less like a clean acquisition path

Intel’s relationship with SambaNova is complicated in the most interesting way. It was reportedly in acquisition talks valuing the company at roughly $1.6 billion, according to the December Bloomberg News report cited by TechCrunch. It has also been a backer since SambaNova’s Series C and participated in this latest round.

Five months before the Series F first close, SambaNova announced a multi-year partnership with Intel to support AI inference development based on Intel’s Xeon chip. TechCrunch reports the two companies now co-develop products and take them to market together.

That makes Intel both investor and strategic partner, while not owning the asset outright.

For Intel, the July round can be read three ways:

• Missed bargain: If acquisition talks around $1.6 billion were real, the price has moved dramatically.
• Strategic hedge: Participation in the round preserves access to SambaNova’s trajectory without requiring full ownership.
• Market signal: Intel’s continued involvement shows that AI inference remains important enough to support external specialists.

For SambaNova, Intel brings scale and customer channels. It also brings a question. If a major incumbent is close enough to co-develop and sell with SambaNova, public-market investors will eventually ask how dependent the startup is on that relationship.

Liang’s comments suggest the company wants optionality rather than a fixed path. A sale remains possible in a dynamic AI market, but the public-company route is now clearly on the table. After the $11 billion mark, the buyer list gets shorter and the diligence bar gets higher.

JPMorganChase gives the private AI pitch a marquee proof point

The customer announcement may matter more than the funding number. Alongside the new round, SambaNova said JPMorganChase selected it as an “inference-infrastructure partner.” Its SN40L and SN50 systems are set to power secure, on-premises AI inference at the bank.

Liang framed the win as a message to finance and other regulated industries.

“Having JPMorgan Chase decide they’re going to use SambaNova for their inference solution is a big deal,”
“It sends a message to the banking industry that it’s time not to completely depend on cloud services. These banks want heterogeneous [infrastructure].”

That is the cleanest articulation of SambaNova’s commercial wedge. The company is not asking every enterprise to abandon cloud AI. It is arguing that some workloads, especially sensitive models inside banks, governments, and large industrial companies, will move toward private infrastructure controlled by the buyer.

JPMorganChase is useful for SambaNova because it gives the company a reference customer in a sector that tends to demand performance, control, security, and reliability. It also supports Liang’s broader claim that enterprises and governments are still “just starting their AI journey,” while much of the growth to date has been concentrated among model makers and frontier labs.

XOOMAR analysis: this is where the SambaNova valuation becomes more defensible. A chip startup valued at $11 billion needs more than technical claims. It needs proof that conservative buyers will run important workloads on its systems. JPMorganChase does not prove broad adoption by itself, but it is the kind of customer name that helps open procurement doors.

Still, buyers won’t switch because General Atlantic wrote a large check. CIOs and cloud architects will demand measurable gains in deployment speed, inference cost, uptime, security posture, and compatibility with the AI tools they already use. SambaNova has won attention. Now it has to win renewals and expansions.

Nvidia is not the immediate casualty, but specialized challengers keep adding pressure

SambaNova’s rise should not be read as a direct knockout threat to Nvidia. The supplied source material does not show that. In fact, related reporting cited in the source context says SambaNova’s chips are designed to work alongside Nvidia products rather than replace them.

The sharper reading is narrower: specialized AI infrastructure vendors are trying to carve off workloads where their architecture fits better than a generic GPU deployment. SambaNova’s stated target is large-model inference, especially in private environments.

Other challengers are attacking different slices of the compute stack. The related source context names Cerebras and Groq as inference-silicon competitors, with Cerebras having completed a Nasdaq IPO in May 2026 and Groq remaining independent after a reported Nvidia licensing and asset deal in December 2025. Those examples matter because they show that investors and strategic buyers are not treating inference hardware as a side market.

SambaNova’s differentiation, as presented by Liang, is the ability to run very large models fast and fit multi-trillion-parameter models onto a single rack. If that advantage holds in production, it could make the company attractive for banks, sovereign AI projects, and large enterprises that want controlled inference capacity.

XOOMAR analysis: Nvidia’s moat is still reinforced by software adoption, developer familiarity, and existing procurement patterns. SambaNova does not need to break that moat everywhere. It needs to prove that enough high-value private AI workloads justify a separate buying decision.

The next deadline is the second close, then SN50 shipments in the second half of 2026

SambaNova’s next test is not another valuation headline. It is execution.

Liang told TechCrunch a second close is likely in the next few weeks. After that, the company has to turn capital into supply, supply into shipped systems, and shipped systems into repeat customers. The SN50 is due to begin shipping in the second half of 2026, with SoftBank as the first deployment partner.

The execution risks are concrete:

• Supply: SambaNova says it is using capital to secure the supply chain for the next 12 months.
• Product cycle: SN50 must move from February unveiling to customer deployments in the second half of 2026.
• Customer proof: JPMorganChase is a marquee win, but one bank does not make a market.
• Software maturity: Buyers need model support and operational simplicity, not just chip performance.
• Exit path: An $11 billion valuation raises the bar for an IPO or any future acquisition.

The most important evidence to watch is not whether another famous investor joins the Series F. It is whether named customers expand deployments after testing real workloads. A strong confirmation signal would be additional regulated enterprises, sovereign cloud partners, or neocloud customers committing to SN50 systems after initial evaluations. A weakening signal would be shipment delays, vague customer language, or dependence on one or two flagship accounts.

SambaNova now has cash, credibility, Intel proximity, and a JPMorganChase proof point. It also no longer has the luxury of being judged like a scrappy underdog. At $11 billion, the market is asking SambaNova to become a real AI infrastructure supplier, not merely an attractive chip startup with a better story than last quarter.

The Bottom Line

• SambaNova’s jump from a rumored $1.6B acquisition value to an $11B valuation shows how quickly AI infrastructure assets are being repriced.
• The round highlights investor demand for scarce AI chip companies focused on inference, not just training.
• The funding gives SambaNova more leverage as both enterprise demand and strategic acquisition interest intensify.