AI Industry

Jim Keller's Tenstorrent Raises $1.5B at a $6.2B Valuation

Oct 20, 2026·4 min read·By Rahish Saifi

Open AI accelerator server on a workbench showing a grid of chips with black heatsinks, with data-centre racks behind

On October 19, 2026, Tenstorrent said it has closed $1.5 billion in Series E funding at a $6.2 billion post-money valuation. In its announcement, the company said the round was led by Trace Capital and Innovation X, with participation from new and existing investors including Agostinelli Group, ARK Invest, Brett Barna, Cadence Ventures, Eclipse, Fidelity Management & Research Company, Maverick Silicon, Morgan Stanley, Protagonist, Sandisk, T. Rowe Price (accounts advised by T. Rowe Price Associates), Tether Investments, Troluce Capital, Virtu Financial, and XTX Markets.

The deal lands as inference-heavy workloads and open-weight models push buyers to look beyond a single GPU vendor. Reuters reported the round more than doubled Tenstorrent’s valuation from December 2024—a useful benchmark because outlets disagree on the exact prior mark (see below).

Why Keller stayed independent

TechCrunch reported that the round comes five months after the company reportedly received acquisition interest from Intel and Qualcomm at valuations of as much as $5 billion. CEO Jim Keller told Bloomberg, per TechCrunch, that Tenstorrent decided to remain independent and raise more capital after demonstrating that its Galaxy servers could run DeepSeek models.

On June 15, 2026, Reuters reported that Qualcomm was in talks to buy Tenstorrent for $8 billion to $10 billion, citing The Information. Keller later reportedly denied talks with Qualcomm, per Seeking Alpha (June 30, 2026). Treat those M&A threads as background, not settled fact.

What the money is for

Tenstorrent said it will use the capital to deploy Galaxy servers at scale, keep serving open-weight models across LLMs and video generation, and advance the AI and CPU IP it licenses to customers designing their own silicon.

The company sells AI compute at every level—from a single card or workstation to racks, superclusters, and custom IP—with an open-source software stack so developers can run frontier models locally and enterprises can deploy on-prem at what it calls industry-leading latency and throughput.

One architecture, many sizes

“We built one architecture and sell it at every size,” Keller said in the release. “The future of compute is open, scalable, and owned by the people building on it. Our investors believe in that model too, and we appreciate their support.”

George Bousis, co-founder and managing partner at Protagonist, added: “Few companies design the chip and the software that runs on it. Tenstorrent does both, and its video generation results beat industry benchmarks ahead of our expectations.” Protagonist said it is doubling down in the Series E.

Tenstorrent builds RISC-V-based AI processors and also licenses its TT-Ascalon RISC-V CPU and Tensix AI cores to chip designers including Samsung and LG. Backers listed in its boilerplate include Bezos Expeditions, Samsung, LG Electronics, Hyundai Motor Group, and Fidelity Management and Research Company, among others. Offices span Santa Clara, Austin, Toronto, Belgrade, Tokyo, and Bangalore—the last of which matters for readers tracking India’s role in global AI silicon supply chains.

Toronto roots and headcount

The Globe and Mail noted Tenstorrent was founded in Toronto about 10 years ago, employs roughly 1,000 people (about 200 in Toronto), and raised US$693 million in late 2024. The Globe tied that prior round to a US$2 billion valuation; TechCrunch cited $2.6 billion. Reuters says the December 2024 round valued the company at $2 billion pre-money, so the two figures likely differ by pre- vs post-money; either way, the new round more than doubled the valuation.

Key deal facts

Item	Detail
Round	Series E
Size	$1.5B
Post-money valuation	$6.2B
Lead investors	Trace Capital; Innovation X
Announced	Oct 19, 2026
Stated use of funds	Galaxy server scale-out; open-weight LLM and video inference; CPU/AI IP licensing

Why it matters

For inference buyers: Tenstorrent is betting that open-weight models and on-prem latency will reward a vertically integrated stack—custom RISC-V silicon plus open software—rather than default GPU rentals. Galaxy as a proof point for DeepSeek-class models is the commercial story Keller is telling investors.

For the competitive map: A $6.2B independent challenger still trails Nvidia by orders of magnitude, but it signals that sovereign buyers, hyperscaler-adjacent labs, and chip designers licensing IP want credible second sources. Tenstorrent’s dual revenue lines—systems and IP—combine Arm-style IP licensing with selling finished systems.

For Indian readers: Tenstorrent lists Bangalore among its six offices—alongside Santa Clara, Austin, Toronto, Belgrade, and Tokyo—underlining that AI silicon engineering is not confined to Silicon Valley.

Bottom line

Tenstorrent’s $1.5B Series E at $6.2B post-money, co-led by Trace Capital and Innovation X, funds scale-up of Galaxy servers and open-weight inference while doubling down on RISC-V CPU and Tensix licensing. Keller chose independence over reported takeover talks—and investors priced the company as a full-stack open compute alternative, not a feature on someone else’s balance sheet.