Santa Clara-based Tenstorrent has raised $1.5 billion in Series E funding at a $6.2 billion post-money valuation, more than doubling its valuation from its previous major funding round in December 2024.
The financing was co-led by Trace Capital and Innovation X, with participation from ARK Invest, Morgan Stanley, Fidelity Management & Research, T. Rowe Price, Eclipse, Maverick Silicon, Sandisk, Tether Investments and several other institutional investors.
Founded in 2016 by Ljubisa Bajic, Ivan Hamer and Milos Trajkovic, Tenstorrent is led by CEO Jim Keller, a prominent chip architect whose previous work includes Apple's early custom processors, AMD's Zen architecture and Tesla's autonomous-driving hardware.
Tenstorrent is developing an alternative to the proprietary hardware and software ecosystems that currently dominate AI computing.
The company designs AI accelerators, processors and complete computing systems, including workstations, servers and larger-scale infrastructure. It also licenses semiconductor intellectual property to customers developing their own chips.
Its technology combines Tensix AI processing cores with RISC-V-based CPU designs and an open-source software stack intended to give customers greater control over their computing infrastructure.
That strategy differs from simply creating another GPU competitor. Tenstorrent wants customers to be able to purchase its hardware, build systems around its architecture or license its designs directly.
The company already has semiconductor IP relationships with major electronics manufacturers, including Samsung and LG.
Its flagship infrastructure product, Galaxy Blackhole, connects 32 AI processors within a system designed for workloads including large language model inference and AI video generation.
Tenstorrent announced general availability of Galaxy Blackhole in April 2026 and has been expanding deployments of its computing systems. The new funding will support volume manufacturing, global commercialization, additional Galaxy server deployments and further development of its AI and CPU intellectual property.
The company previously raised more than $693 million in December 2024 at a reported $2 billion pre-money valuation, approximately $2.6 billion post-money. Its latest financing therefore represents a substantial increase in investor expectations.
The broader market opportunity comes from growing demand for AI inference, the process of running trained models to serve users. As AI applications expand, inference costs, memory bandwidth, energy consumption and infrastructure flexibility become increasingly important.
Tenstorrent is particularly focused on customers deploying open-weight models and organizations that want to own and control their computing infrastructure rather than depend entirely on a single hardware supplier.
Its approach also targets sovereign AI infrastructure, where governments and domestic technology companies may want greater control over the underlying computing stack.
However, Nvidia's advantage goes well beyond processor performance. Its CUDA software ecosystem, networking technologies, developer tools and production-scale deployments create substantial switching costs.
Tenstorrent must demonstrate that its hardware delivers competitive real-world economics and that developers can run production workloads without sacrificing reliability, performance or software compatibility.
Its mixed business model also creates execution challenges. Building and selling AI servers requires manufacturing capacity and support infrastructure, while licensing semiconductor IP requires long-term relationships with customers designing their own chips.
At a $6.2 billion valuation, Tenstorrent is being financed as a potential major alternative in AI computing. The critical question is whether its technology can move beyond specialized deployments and become a widely adopted production platform.