Tenstorrent
Open RISC-V AI compute at every scale, from a single card to superclusters.
About Tenstorrent
Tenstorrent is a next-generation computing company that builds AI processors, systems and chip intellectual property on the open RISC-V architecture. Its Galaxy servers combine dozens of Blackhole processors, each built around Tensix AI cores, into rack-scale systems that run chatbots, coding assistants and video generators. One architecture spans every size, from a single workstation card to large clusters of networked machines.
Alongside its own hardware, Tenstorrent licenses its TT-Ascalon RISC-V CPU cores and Tensix AI cores to other companies building custom silicon. Samsung and LG are among the licensees using its designs in their own products, and its software stack is open source. The company is led by CEO Jim Keller, the chip architect behind Apple’s A4 and A5 chips, AMD’s Zen processors and Tesla’s self-driving computer. Tenstorrent was founded in Toronto in 2016 and is headquartered in Santa Clara, with offices in Austin, Toronto, Belgrade, Tokyo and Bangalore.
Founders
Funding History
Frequently Asked Questions
What does Tenstorrent do?
Tenstorrent builds AI processors and rack-scale Galaxy servers on the open RISC-V architecture, and licenses its TT-Ascalon CPU cores and Tensix AI cores to other companies building custom silicon.
Who leads Tenstorrent?
Tenstorrent is led by CEO Jim Keller, the chip architect behind Apple's A4 and A5 chips, AMD's Zen processors and Tesla's self-driving computer. The company was founded in 2016 by Ljubisa Bajic, Milos Trajkovic and Ivan Hamer.
How much funding has Tenstorrent raised?
Tenstorrent has raised about $2.7 billion in total, including a $1.5 billion Series E in October 2026 that valued the company at $6.2 billion, and a $693 million Series D in December 2024.
How is Tenstorrent different from Nvidia?
Tenstorrent builds on RISC-V, an open architecture with no licensing fees, instead of a proprietary stack. It also licenses its chip designs to other companies and sells one architecture at every scale, from a single card to superclusters.
