Tenstorrent, the Toronto-based AI hardware company led by chip designer Jim Keller, has closed a $1.5 billion Series E round at a $6.2 billion valuation, according to SiliconANGLE. Trace Capital and Innovation X jointly led the raise. Protagonist, Morgan Stanley, Fidelity Management & Research Co., ARK Invest and T. Rowe Price also took part.
The company began as a chip designer but now sells complete systems. Its range includes Galaxy servers, TT-QuietBox liquid-cooled workstations and Blackhole PCIe cards. Compute comes from Tensix Neo accelerators and Ascalon CPUs, both built on the open RISC-V instruction set. Tenstorrent positions the designs as scalable and energy-efficient, aiming for strong AI performance without runaway power draw.
Galaxy servers are its best-selling line. They can be linked into large clusters for training and running frontier models. Virtu Financial is cited as using them for on-premises AI trading agents, and Equinix makes them available to enterprise customers.
Tenstorrent also develops an open-source software stack. It comprises TT-Forge, a compiler that connects frameworks such as PyTorch to its chips; TT-Metalium, an SDK for hardware optimization and custom kernels; and TT-LLK for low-level kernel work. Together they let developers compile, tune and run models on its silicon.
Keller described the approach as one architecture sold at every size. Protagonist's George Bousis argued that few rivals design both chip and software, and said the company's video generation results beat industry benchmarks ahead of his firm's expectations.
The money is earmarked for higher Galaxy production, for global deployment aimed at large language model and video generation workloads, for research to advance the RISC-V ecosystem, and for global hiring.
Why it matters: Capital on this scale gives a non-GPU vendor room to build supply and support, and Equinix availability suggests a route into enterprise colocation. Still, the source offers no shipment volumes, revenue or independent benchmark data, so how far Tenstorrent can erode Nvidia's position remains unproven.
TwinStack automated analysis, based on reporting by SiliconANGLE. Check the original for full details.