AMD (AMD.US) acquires AI inference chip startup Taalas to enhance its data center product matrix.

U.S. chip manufacturer AMD (AMD.US) is further expanding its artificial intelligence (AI) chip layout. As the wave of generative AI enters its fourth year, GPUs remain the core hardware for AI training and inference, but the industry is increasingly recognizing that a single GPU cannot meet all AI application scenarios. On Thursday, AMD announced that it has reached an agreement to acquire Toronto-based AI inference chip startup Taalas, further strengthening its competitiveness in the inference chip sector and enhancing its data center AI product matrix. AMD did not disclose the transaction amount. According to reports, Taalas, founded in 2023, has raised approximately $219 million in total funding. This transaction comes just about seven months after NVIDIA Corporation (NVDA.US) spent $20 billion to acquire the assets of high-performance AI chip design company Groq, reflecting that AI inference chips are becoming the new focus of industry competition. With the rapid popularization of generative AI applications, the industry's focus is gradually shifting from model training to model inference, which is the computational process through which AI models respond to user requests in practical business scenarios. Unlike general-purpose computing chips like GPUs, Taalas develops dedicated inference accelerators (ASICs) that are hardware-customized for specific AI models, optimizing the chip for "hardwiring" to a particular model during design. Therefore, while flexibility is lower, inference efficiency can be significantly enhanced, and costs can be reduced. According to Taalas, its technology can quickly convert any AI model into dedicated silicon chips, completing hardware implementation in about two months from receiving a completely new AI model. Currently, Taalas's products primarily support the Llama 3.1 small model from Meta (META.US) and are developing new products suitable for larger-scale models. Its chips are manufactured using mature processes from Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR (TSM.US) and integrate high-speed SRAM storage to reduce inference latency. Industry experts believe that these dedicated inference chips are particularly suitable for low-latency scenarios, such as AI chat applications, real-time search, autonomous driving, and smart assistants, all of which require rapid response. AMD CEO Lisa Su has previously stated that she has always believed that there is no "one chip fits all applications" situation in the AI chip market. She pointed out that GPUs, due to their high versatility, will still hold the majority share of the AI chip market, as they can support the continuously evolving new generation of AI models. However, at the same time, different application scenarios require more specialized accelerators to collaboratively build a complete AI computing ecosystem. Currently, demand for GPUs remains strong, propelling NVIDIA Corporation's market value to exceed $5 trillion, making it the most valuable publicly traded company globally. This acquisition also reflects that AMD is shifting from merely selling GPUs to providing complete AI computing system solutions. In recent years, competition in AI data centers has expanded from individual chips to complete rack systems. This year, AMD began delivering its first rack-level AI system, Helios, to customers such as Meta and Microsoft Corporation (MSFT.US), competing with NVIDIA Corporation's DGX systems. AMD stated that it will integrate Taalas's inference chips and related technologies into the company's product roadmap, developing in synergy with EPYC CPUs, Instinct GPUs, and complete system products to provide customers with a more comprehensive AI infrastructure for data centers. In fact, AMD has been continuously improving its AI ecosystem through acquisitions. In July of this year, the company announced a partnership with AI chip company Cerebras to integrate its AI accelerators into AMD systems. Previously, AMD has completed several AI-related acquisitions, including a $665 million acquisition of AI model developer Silo AI in 2024, and a $4.9 billion purchase of server manufacturer ZT Systems, laying the technical foundation for the Helios rack product. Additionally, last year, the company acquired several AI startups, including inference software developer MK1. Analysts believe that as the AI industry gradually transitions from model training to large-scale commercial deployment, the demand for inference computing is rapidly growing. By acquiring Taalas, AMD is further enhancing its capabilities in dedicated inference chips, which is expected to strengthen its competitive position in the enterprise AI infrastructure market against NVIDIA Corporation.