On a Monday when AI stocks were still digesting last week’s selloff, one private chip startup posted a very different headline. Tenstorrent said it closed a $1.5 billion Series E round that values the company at $6.2 billion—more than doubling its prior mark and underscoring how investors still want alternatives to Nvidia’s grip on AI hardware.
The raise lands at a moment when hyperscalers, sovereign buyers, and model labs are hunting for more compute, lower lock-in, and open software stacks. For U.S. readers following the AI chip race, Tenstorrent’s deal is a signal that the hardware story is no longer only about one company’s CUDA ecosystem.
What Tenstorrent Announced and Who Led the Round
In an Oct. 19 company announcement, Tenstorrent said it closed $1.5 billion in Series E funding at a $6.2 billion post-money valuation. Trace Capital and Innovation X led the round. Participants included Agostinelli Group, ARK Invest, Cadence Ventures, Eclipse, Fidelity Management & Research Company, Morgan Stanley, Protagonist, Sandisk, T. Rowe Price, Tether Investments, Virtu Financial, XTX Markets, and others named in the release.
Reuters reporting, carried by outlets including Channel NewsAsia, put the prior December 2024 round at more than $693 million with a $2 billion pre-money valuation led by Samsung Securities and AFW Partners. Techmeme’s summary of Bloomberg coverage described the prior valuation near $2.6 billion after that 2024 raise. Either way, Monday’s $6.2 billion post-money figure is a sharp step up.
CEO Jim Keller—known for work on Apple’s A4/A5, AMD’s Zen architecture, and Tesla’s Full Self-Driving chip—said the company built one architecture it sells “at every size,” from cards and workstations to racks and custom IP. Trace Capital founder Ben Mahdavi cited Keller’s silicon and software experience as a reason for the partnership as Galaxy servers scale.
Why Investors Are Betting on Open RISC-V AI Compute
Tenstorrent builds AI processors and systems on RISC-V, an open instruction-set architecture that lets designers create chips without the same proprietary licensing model that defines some competing ecosystems. The company sells compute across the stack and licenses AI and CPU IP—including its TT-Ascalon RISC-V CPU and Tensix AI cores—to chip designers such as Samsung and LG, according to its announcement.
That pitch resonates as customers try to own more of their AI stack. Tenstorrent said the new capital will fund volume manufacturing and global commercial expansion of Galaxy servers, which combine dozens of its AI chips for applications such as chatbots, coding assistants, and video generation. The company also said it will keep advancing open-weight model workloads and the IP it licenses to customers designing their own silicon.
Protagonist co-founder George Bousis said few firms design both the chip and the software that runs on it, and that Tenstorrent’s video-generation results beat industry benchmarks ahead of the firm’s expectations—one reason Protagonist doubled down in the Series E.
How This Fits the Broader Push Against Nvidia Lock-In
Nvidia still dominates AI accelerators, in large part because much of the industry’s software was built around CUDA. Moving that stack can take months of engineering work, which has kept many buyers tied to Nvidia even as Amazon, Google, Microsoft, and startups build alternatives.
Monday’s news sits beside that wider contest. Separately on Oct. 19, Amazon said it was opening TorchNeuron—a tool that runs existing PyTorch code on AWS Trainium chips after a basic configuration change—to a public beta and planned an open-source release with the PyTorch Conference, GeekWire reported. Google has pursued a similar TorchTPU path. Microsoft has said it plans to run more AI services on its own Maia chips.
Tenstorrent’s raise is smaller than hyperscaler capital budgets, but it targets a complementary niche: open hardware and software for customers who want local, on-prem, or sovereign control rather than only renting cloud GPUs. The company operates from Santa Clara, Austin, Toronto, Belgrade, Tokyo, and Bangalore and lists backers that have included Bezos Expeditions, Samsung, LG Electronics, and Hyundai Motor Group.
What to Watch After the Funding Announcement
Raising money is not the same as winning sockets. The practical questions now are execution and adoption:
Can Galaxy servers ship at scale? Volume manufacturing is the stated use of proceeds. Delivery timelines and customer wins will matter more than the headline valuation.
Will open software convert curiosity into production? An open stack helps developers experiment; production latency, throughput, and tooling decide renewals.
Does the AI chip market keep diversifying? Last week’s semiconductor selloff and CoreWeave credit stress showed how sensitive AI infrastructure valuations can be. Diversified supply can reduce single-vendor risk—but demand still has to clear at profitable prices.
For now, Tenstorrent’s $1.5 billion Series E is a clear vote that investors still see room for challengers in AI silicon, even after a rough week for public AI names.
This is for information only, not financial advice.
Key takeaways
- Tenstorrent closed a $1.5 billion Series E at a $6.2 billion post-money valuation on Oct. 19, 2026 (company announcement).
- Trace Capital and Innovation X led; ARK Invest, Fidelity, Morgan Stanley, Sandisk, T. Rowe Price, Tether, and others participated.
- The prior late-2024 round raised more than $693 million at about a $2 billion pre-money valuation (Reuters/CNA).
- Proceeds target Galaxy server volume manufacturing, global expansion, and AI/CPU IP for customers designing their own chips on RISC-V.
- The raise fits a broader industry push—alongside Amazon Trainium/TorchNeuron and other custom chips—to reduce dependence on a single AI hardware ecosystem.
Sources
- Tenstorrent: Series E announcement — $1.5B at $6.2B valuation
- Reuters via Channel NewsAsia: Tenstorrent more than doubles valuation to $6.2 billion
- Techmeme / Bloomberg summary: Tenstorrent $1.5B Series E
- GeekWire: Amazon TorchNeuron challenges Nvidia software advantage
Comments
Post a Comment