AUTO-UPDATED

Cloud startup Volta claims $10B AI lab deal for Norway bit barn

AMD has acquired AI chip startup Taalas to enhance inference performance by etching machine learning models directly into silicon, achieving speeds of up to 17,000 tokens per second.

Key Points

  • AMD purchased Taalas to integrate AI models into custom silicon, significantly boosting inference efficiency.
  • Early technical demonstrations show the hardware processing up to 17,000 tokens per second.
  • The acquisition aims to optimize model-specific integrated circuits for high-performance AI workloads.
  • Taalas specializes in hardware-level acceleration that bypasses traditional software-based model execution.

Why it Matters

This acquisition signals a strategic shift toward hardware-level AI optimization to overcome the latency and power limitations of general-purpose processors. By embedding models directly into silicon, AMD could gain a significant competitive advantage in the rapidly expanding market for high-speed, efficient AI inference hardware.
Theregister.com Published by Dan Robinson
Read original