Tenstorrent Raises $1.5 Billion Series E to Scale Galaxy AI Servers and RISC-V Licensing

Tenstorrent, the AI processor and RISC-V CPU designer led by chip architect Jim Keller, said on Oct. 19, 2026, that it has raised $1.5 billion in a Series E financing led by Trace Capital and Innovation X. The company, which has engineering teams in Santa Clara, Austin, Toronto, Belgrade, Tokyo and Bangalore, plans to use the capital to deploy its Galaxy servers at scale, serve open-weight AI models and expand the processor designs it licenses to other chipmakers.

Company background

Tenstorrent was founded in Toronto in 2016 to build processors for machine learning workloads. Keller, whose earlier career includes processor design work at Apple, AMD and Tesla, joined the company in 2020 and later became chief executive. Under his leadership, Tenstorrent has pursued two lines of business in parallel: selling its own AI hardware systems and licensing its processor designs to companies that want to build their own chips.

The licensing side rests on RISC-V, an open instruction set architecture that any company can implement without paying a license fee for the instruction set itself. Tenstorrent's TT-Ascalon is a high-performance CPU core built on that architecture, while its Tensix cores handle AI computation. In the funding announcement, Keller described the strategy in a single line: the company "built one architecture and sell it at every size." He added that the future of compute is open, scalable and owned by the people building on it.

The systems side is anchored by Galaxy, the company's server platform. Tenstorrent said a central use of the new money is deploying those servers at scale and using them to serve open-weight models for text and video generation, which moves the company closer to operating AI infrastructure rather than only supplying components.

Industry implications

The round is a large private financing for an AI chip designer, and it arrives as data center operators, cloud providers and enterprises look for alternatives to the dominant GPU-based accelerator stack. Investors are funding a company whose pitch rests on openness at several layers: an open instruction set, an open-source software stack and support for open-weight models. That combination targets buyers who want more control over their hardware roadmaps and less dependence on a single supplier's proprietary tools.

The financing also signals that institutional capital is willing to back a hybrid model that mixes hardware sales, IP licensing and model serving. Each of those businesses has different capital needs. Server deployment requires inventory and supply commitments, while licensing depends on long design cycles at customer companies. A large cash reserve gives Tenstorrent room to pursue all three without relying on near-term revenue from any one line.

For the RISC-V ecosystem, the round adds weight to the argument that the architecture can move from embedded and microcontroller uses into data center and high-performance computing designs. Licensing deals with consumer electronics groups such as Samsung and LG suggest that demand reaches beyond data centers into devices and other edge products.

Competitive and supply chain implications

Tenstorrent competes on two fronts. In AI accelerators, it faces established GPU suppliers, custom chips built in-house by hyperscalers and a field of well-funded startups. In processor IP, its RISC-V cores compete with Arm-based designs that dominate mobile and are expanding in servers. A company that can supply both AI cores and CPU cores under one licensing relationship offers chip designers a way to build complete systems on a chip without stitching together IP from several vendors.

The investor list carries supply chain meaning as well. Sandisk, a flash memory supplier, is among the participants, which points to the growing importance of memory and storage in AI systems that serve large models. Several large asset managers, including Fidelity, T. Rowe Price and Morgan Stanley, also joined, a mix often seen in late-stage rounds ahead of a possible public listing, although Tenstorrent has not announced any such plan.

Tenstorrent did not disclose its manufacturing partners in the announcement. Scaling Galaxy server deployments will depend on access to advanced foundry capacity, packaging and high-bandwidth memory, all of which remain tight across the industry. How the company secures those inputs will shape how quickly the new capital turns into installed systems.

Regional implications

In North America, the round strengthens a company with roots in Toronto and major operations in Santa Clara and Austin, two of the main U.S. hubs for chip design. It also gives Canada's technology sector a well-funded semiconductor design company.

In Asia, the named licensees Samsung and LG tie Tenstorrent closely to South Korea's electronics industry, and its Tokyo team positions it near Japanese efforts to rebuild advanced chip design capability. Its Bangalore operation taps India's large pool of semiconductor design engineers, many of whom already work for global chip companies. In Europe, the Belgrade team gives the company an engineering base in Serbia, a smaller but growing center for software and hardware talent.

What comes next

• Expansion of Galaxy server deployments, which the company named as a primary use of the new capital.
• Launch or scaling of open-weight model serving for large language and video generation workloads.
• Further TT-Ascalon and Tensix licensing agreements beyond existing licensees such as Samsung and LG.
• Disclosure of manufacturing and supply partners as volume production increases.
• Updates on any additional strategic investors or customer partnerships following the close of the round.

Sources

• Tenstorrent: Tenstorrent Announces a $1.5B Series E
• TNGlobal: Tenstorrent raises $1.5B Series E

Image: Tenstorrent CEO Jim Keller, by James Keller via Wikimedia Commons, CC BY-SA 4.0. Cropped and edited by Credence Research.