Tenstorrent Just Raised $1.5 Billion to Take on Nvidia With Open Hardware

The most important AI chip story of the week is not about Nvidia at all. It is about the company trying to break Nvidia’s grip on the entire industry. Tenstorrent, the Santa Clara chip startup run by legendary designer Jim Keller, announced on October 19 that it has closed a $1.5 billion Series E round, valuing the company at $6.2 billion post money. That more than doubles its worth from its previous raise in December 2024, when it took in over $693 million at a roughly $2 billion pre money valuation.

## Why this raise is bigger than the money

A billion and a half dollars is eye catching on its own, but the investor list tells the real story. The round was co led by Trace Capital and Innovation X, and it drew in Fidelity Management and Research, T. Rowe Price, Morgan Stanley, Cathie Wood’s ARK Invest, Sandisk, Tether Investments, and existing backers.

When crossover funds like Fidelity and T. Rowe Price show up in a late stage round, it usually means one thing. The company is being priced like future infrastructure, not a long shot startup. Market watchers now treat a public listing within two to three years as the base case. Wall Street is treating Tenstorrent as a company headed for the public markets, and this round is the runway.

## The open hardware bet

Tenstorrent’s pitch is disarmingly simple. Where Nvidia sells you a closed stack of chips, interconnects and software that only work together on its terms, Tenstorrent builds on RISC-V, an open source chip architecture that anyone can use without paying licensing fees.

For a cloud provider or a government, that distinction matters more every year. Running AI on Nvidia hardware means paying the Nvidia tax and living inside the Nvidia ecosystem, with its proprietary software stack as the gatekeeper. Tenstorrent argues that open hardware lets customers own their silicon destiny: design what they need, avoid licensing lock in, and keep the cost curve under control.

The company makes money two ways. It sells Galaxy servers, machines built from dozens of its AI chips aimed at workloads like chatbots, coding assistants and video generation. And it licenses its processor designs and software to customers who want to build their own silicon on its architecture. Both bets need scale, which is exactly what the new money is for.

## Where the money goes

According to the announcement, the $1.5 billion will fund three things: expanding Galaxy server manufacturing and commercial deployments, developing new generations of its AI and RISC-V based chip IP, and hiring globally. The company did not break down the split, and it has not disclosed revenue figures or a forecast, so the market is largely taking the deployment story on trust.

What deployments exist are real, if early. Virtu Financial runs Galaxy servers on premises for AI agents tied to trading and operations. Equinix offers Galaxy systems through its Distributed AI Hub. Smallest.ai runs its Lightning V2 real time text to speech model on Galaxy servers for customers that want voice AI without shipping data to the cloud. And in Japan, a company called ai& has deployed more than 120 Galaxy systems as part of sovereign AI infrastructure, where hardware independence is a genuine procurement category.

## The Jim Keller factor

You cannot write about Tenstorrent without writing about Jim Keller, and investors know it. Keller previously led Tesla’s self driving chip program and held lead design roles on Apple’s A series processors and AMD’s Zen architecture, arguably the most successful CPU turnaround in modern history. That reputation can raise a billion dollars, but it cannot ship a server. Tenstorrent’s central question is execution: can it turn an elegant architecture into dependable production compute at scale? The company has named real deployments, but it has not disclosed revenue or shipment figures. Until it does, the valuation rests on belief.

## The fight it is picking

Nvidia is not sitting still. Its lead in AI compute comes from more than raw silicon. The CUDA software ecosystem is the moat: millions of developers, thousands of optimized libraries, and a decade of tooling that make switching painful. Tenstorrent’s answer is its own open software stack, but ecosystems are built over years, not funded in a day. The race is between Tenstorrent’s open architecture vision and the sheer gravitational pull of Nvidia’s installed base, and the $6.2 billion question is whether openness can overcome a decade of lock in.

## What to watch next

The next twelve months should tell us a great deal. Watchers expect Galaxy systems to ship at volume and a landmark hyperscaler or sovereign customer announcement as the next proof point. The full investor roster and round terms remain undisclosed. What is certain is that the money has landed and the clock has started. Tenstorrent now has the capital of a serious contender. The job is to look like one when it matters, in the data centers of customers choosing between Nvidia’s certainty and an open alternative with something to prove.

## Frequently Asked Questions

### How much did Tenstorrent raise, and what is it worth now?

Tenstorrent announced on October 19, 2026 that it closed a $1.5 billion Series E round at a $6.2 billion post money valuation. The round was co led by Trace Capital and Innovation X, with participation from Fidelity, T. Rowe Price, Morgan Stanley, ARK Invest, Sandisk and Tether Investments, among others. The valuation more than doubles the company’s roughly $2.7 billion post money mark from December 2024.

### What is RISC-V, and why does Tenstorrent use it?

RISC-V is an open source instruction set architecture for processors. Unlike proprietary architectures such as Arm’s, anyone can design chips on RISC-V without paying license fees or depending on a single licensor. Tenstorrent builds its AI chips and licensing business on it, arguing that customers should be able to own and customize their silicon rather than rent access inside a closed ecosystem.

### Can Tenstorrent really compete with Nvidia?

That is the open question. Tenstorrent has real deployments, including Virtu Financial, Equinix and a 120 plus system sovereign AI deployment in Japan, and it now has serious capital. But Nvidia’s lead rests on the CUDA software ecosystem as much as on chips, and Tenstorrent has not disclosed revenue or shipment figures. The next year of production ramp and customer announcements will show whether the open hardware pitch converts into market share.