AMD acquires Taalas, an AI inference chip startup that etches trained models into silicon; demos show up to 17,000 tokens/second
Advanced Micro Devices is counting on its graphics processing units to drive the bulk of its data center growth as cloud companies snap up all the advanced AI chips they can find. But as the generative artificial intelligence boom approaches its fourth anniversary, it's becoming clear that GPUs don't do everything.
On Thursday, AMD announced an agreement to acquire Taalas, a Toronto-based startup that makes chips for inference. Unlike general-purpose GPUs, Taalas' accelerators are customized and hard-wired for a single AI model. In exchange for that loss of flexibility, Taalas' technology promises a less-expensive chip that can produce output for specific models thousands of times faster than a traditional GPU. AMD did not disclose a purchase price. Taalas has raised a total of $219 million in venture funding since its 2023 founding.
The deal comes just over seven months after Nvidia spent $20 billion acquiring assets from Groq, a designer of high-performance AI chips—Nvidia's largest transaction on record.
Taalas' current chip runs a small version of Meta's Llama 3.1 model, though the company is working on chips for bigger and more advanced models. It's manufactured using an older Taiwan Semiconductor Manufacturing Co. process and uses speedy SRAM memory on the chip itself. CEO Ljubisa Bajic wrote on the company's website that Taalas "developed a platform for transforming any AI model into custom silicon," noting that "from the moment a previously unseen model is received, it can be realized in hardware in only two months."
Alternative chips like those from Taalas and Groq are particularly important for "low-latency" applications, where time to first response from an AI model is critical. "I'm a big believer that there's no one-size-fits-all as it comes to chips," AMD CEO Lisa Su said at a product launch in July. She added that AMD still expects GPUs to make up the majority of the AI chip market because they're flexible enough to support newly developed AI models.
AMD's acquisition reflects the rising importance for leading GPU makers to offer integrated systems with multiple different components and chips rather than just processors. AMD recently began shipping Helios, its first rack-scale rival to Nvidia's integrated server racks, to customers including Meta and Microsoft.
AMD said it would integrate Taalas' chips and technology into its road map, including in systems with its central processors and Instinct GPUs. The company is also supporting other complementary accelerators: in July, AMD announced a partnership with Cerebras to integrate its AI chips into its systems.
AMD has been acquisitive in building out components for its Helios racks. In 2024, it paid $665 million for Silo AI, which develops AI models, and purchased ZT Systems for $4.9 billion for its rack-scale technical foundation. Last year, AMD bought several smaller AI companies, including MK1, which made inference software.