Tenstorrent is changing leadership roles

Toronto-based startup grows rapidly, swaps CEO and CTO roles, lays out compelling AI roadmap and RISC-V.

Tenstorrent, a Toronto-based startup that raised more than $200 million in venture funding at a $1 billion valuation in May 2021, has now grown to more than 280 employees and is backed by Eclipse Ventures and Real Ventures, among others. The company hired legendary chip designer Jim Keller as CTO, early angel investor and advisor in early 2021. At the time, he said Tenstorrent had “the best architecture” for AI.

Now Jim and founder Ljubisa Bajic have switched roles. Jim became CEO and founder Ljubisa Bajic took over as CTO. In our opinion, these roles are really what everyone has been doing all along, so the change will be barely noticeable. But let’s look at the early impact Jim has already had.

When Jim joined the company, he saw the changes he wanted to make to the accelerator plan, which had already produced two production chips, notably the high-performance Greyskull. But aside from learning about AI while designing Tesla’s FSDs, Jim’s forte has always been CPU core designs. He led the AMD Zen effort that led to AMD’s amazing market gains, especially in HPC with the EPYC series.

So when you put a CPU designer on a design team, it shouldn’t surprise you that he’s designing a new CPU. In addition to the high-performance AI accelerators already in development, Jim envisioned an opportunity to tap into the momentum RISC-V is enjoying and potentially lead the enhanced core market as a second line of business. His team came up with CPUs that could be built as chips, chiplets in accelerators, or perhaps even made available to multiple customers as chiplet IPs. Companies like Meta and Google have been cajoling AMD and Intel for decades, pushing them for specific add-on features. They could now choose to work with Tenstorrent and use the Tenstorrent RISC-V IP as a solid foundation from which to upgrade. You can see below that this is much more than just a business for Tenstorrent as they plan for CPUs that will run as clients, in edge processors and in HPC and server applications.

Tenstorrent’s Buda software stack is CPU-type agnostic, so the accelerator can connect to Intel, AMD, or RISC-V CPU clusters. Below you can see part of the plan, which, similar to the AMD MI300 and NVIDIA Grace Hopper, combines AI and CPU cores on a single die or chiplet. So if you’re wondering why Tenstorrent is going the RISC-V route, you can see the value the company is creating with both AI and CPU.

So what is Ljubisa working on now you may ask? Basically the same as before. The founder leads the development of the Tenstorrent software stack, which enables the company’s ease-of-use mantra and provides target-independent AI development with an open-source stack that developers need to compile, run, and optimize when needed. Python and now Python 2.0. Most ambitiously, Ljubisa plans to build AI models that can scale across an arbitrarily large array of nodes, greatly simplifying the development and deployment of even the largest language models currently in development.

Conclusions

We’ve barely touched the surface of what Tenstorrent is up to in this short article, and we plan to cover more about Tenstorrent in the near future. We believe Tenstorrent has both a differentiated AI hardware and software story and a compelling RISC-V CPU roadmap. In the emerging era of chiplets, Tenstorrent is well positioned for success and we believe that Jim and Ljubisa’s new roles are a natural fit.