Key facts
- AMD is acquiring Taalas, a startup focused on AI inference chips.
- Taalas' technology aims to reduce computing and memory bottlenecks in AI inference.
Advanced Micro Devices announced its acquisition of Taalas, a chip startup specializing in AI inference workloads. The deal aims to bolster AMD's AI platform and compete with rivals like Nvidia in the rapidly growing market for deploying trained AI models.

The acquisition strengthens AMD's position in the competitive AI chip market, particularly in the crucial area of AI inference, and signals continued investment in specialized hardware for AI applications.
Advanced Micro Devices (AMD) announced on Thursday its acquisition of Taalas, a Toronto-based chip startup specializing in silicon designed to reduce computing and memory bottlenecks for AI inference workloads. The deal, for an undisclosed amount, aims to enhance AMD's technology offerings as the company competes in the rapidly expanding market for deploying trained AI models.
Specialized inference chips are becoming crucial for semiconductor manufacturers as the AI industry shifts focus from model training to real-time, high-volume deployment, driving a race to lower computing costs. AMD plans to integrate Taalas' technology into its accelerator roadmap and develop system-level solutions utilizing its Instinct GPUs. Vamsi Boppana, senior vice president of AMD's Artificial Intelligence Group, stated that the acquisition strengthens AMD's AI portfolio by delivering differentiated inference performance and efficiency, contributing to a full-stack AI platform.
Taalas, founded in 2023, had previously raised $169 million in February, bringing its total funding to approximately $219 million. This acquisition follows a series of strategic moves by AMD to bolster its AI business, including the recent purchases of AI software startup MK1, MEXT, and FastFlowLM.