AMD acquires Taalas, a startup specializing in hardwiring AI models into custom inference chips.

AMD has entered into a definitive agreement to acquire Taalas, a Toronto-based startup that designs custom chips for AI inference. The company was founded in 2023 and has raised $219 million in venture funding to date.

Taalas' approach differs fundamentally from conventional chip design. Rather than building general-purpose processors, the startup creates chips tailored to specific AI models, optimizing inference dataflows and reducing compute and memory bottlenecks. This specialization comes with a tradeoff: each chip is locked to the model it was designed for, meaning deploying a different model requires commissioning entirely new silicon.

However, Taalas contends that this constraint is less limiting than it appears. The company estimates that fewer than a handful of a chip's hundred-plus layers need to be altered between designs, and its proprietary tooling enables the company to move from design to finished silicon in approximately two months. Taalas' first chip ran a version of Meta's Llama 3.1 model.

AMD plans to integrate Taalas' technology into its accelerator roadmap and develop system-level products alongside its Instinct GPUs, EPYC processors, Helios rack-scale platform, and ROCm software. "AMD is building a full-stack AI platform that gives customers the flexibility to deploy the right compute solutions for every AI workload," said Vamsi Boppana, senior vice president of AMD's Artificial Intelligence Group. "Taalas' technology and world-class engineering team strengthen our AI portfolio by delivering differentiated inference performance and efficiency."

Ljubisa Bajic, Taalas' co-founder and CEO, said the company's founding mission was "to rethink AI inference from the ground up by building the hardware around the model." He noted that joining AMD would provide the scale and resources needed to accelerate development.

The acquisition reflects AMD's broader strategy to expand its AI capabilities. The company reported record revenue of $11.5 billion in its most recent quarter, with its data center segment more than doubling year over year to $6.7 billion. AMD has already secured Helios commitments from major customers including Meta and Microsoft, which agreed to deploy the rack-scale system on Azure to support frontier model inference.

AMD stock rose approximately 1.5% following the announcement. The transaction remains subject to customary closing conditions and regulatory approvals.