Taalas bets on hard‑wired models to beat GPUs at inference

Toronto start-up Taalas fixes trained LLMs directly in silicon.

April 14, 2026

Jon Peddie

Using a lean, GPU‑veteran team and more than $200 million in funding to chase ultra‑fast, low‑cost inference for stable, high‑volume models, Taalas builds AI chips with a focused premise: fix a trained model in silicon and remove the overhead of general-purpose compute. Founded in 2023 in Toronto, the company brings together experienced chip architects from Tenstorrent, AMD, and Nvidia. With more than $200 million in funding and a small, engineering-heavy team, Taalas targets inference efficiency, latency, and cost per token. Its HC1 chip shows how specialization can reshape system design and challenge GPU-centered infrastructure assumptions. Taalas formed in August 2023

...

Enjoy full access with a TechWatch subscription!

TechWatch is the front line of JPR information gathering service, comprising current stories of interest to the graphics industry spanning the core areas of graphics hardware and software, workstations, gaming, and design.

A subscription to TechWatch includes 4 hours of consulting time to be used over the course of the subscription.

Already a subscriber? Login below

Taalas bets on hard‑wired models to beat GPUs at inference

Enjoy full access with a TechWatch subscription!

This content is restricted

Login