AMD acquires Taalas to hardwire AI models into silicon - SiliconANGLE
AMD acquired Taalas to embed AI models in silicon; HC1 achieved 17K tokens/sec on Llama 8B—73x H200 speed, 1/10 power. Third deal in 9 months; signals shift to specialized inference over generalist GPUs—reshaping CTO infrastructure strategy.
Financial terms or expected timeline for completion were not disclosed
Early tech demos show model-specific integrated circuits churning out up to 17,000 tokens a second
AMD acquires Taalas to bake models in silicon: 17K token/sec on Llama 3.1 without memory stack or cooling. Industry trend toward model-specific chips for inference vs Nvidia's general-purpose GPUs; trade-off is lock-in (2-month re-spins).
AMD acquires Taalas, a startup with model-specific chips that etch weights into silicon; HC1 serves Llama 3.1 8B at 16,960 tokens/sec—48x NVIDIA's speed. Lower inference costs could drive developers to extend reasoning times, reshaping test-time scaling economics.
AMD acquires Taalas ($219M total funding) to add specialized inference chips to its full-stack AI platform with Instinct GPUs. The acquisition reflects the industry shift from training to inference deployment and intensified competition with Nvidia on cost optimization for enterprise AI workloads.
AMD acquires Taalas, an inference-chip startup, to add HC1 silicon optimized for Llama models with 1/10 the power of Nvidia H200. The acquisition strengthens AMD's AI portfolio against Nvidia while validating market demand for model-specific inference hardware.
AMD is buying Canadian startup Taalas, which hard-codes model weights directly into inference chips. That makes them extremely fast but locks each chip to a single model. A demo chip hit over 16,000 tokens per second…
Taalas' current chip runs a small version of Meta's Llama 3.1, though the company is working on chips for bigger and more advanced models.
Aug 6 : Advanced Micro Devices said on Thursday it would buy chip startup Taalas for an undisclosed amount, strengthening its technology as it taps into the rapidly growing market…
AMD acquired Taalas, a startup that bakes model weights into custom chips. Early benchmarks show 17,000 tokens/sec. This could change AI inference economics.
Toronto-based Taalas develops specialized silicon designed to reduce computing and memory bottlenecks in AI inference, the process of running trained AI models to generate…
The Inference Inflection is HEATING up.
AMD has acquired AI chip startup Taalas, leveraging its groundbreaking technology to embed model weights into silicon for dramatically enhanced inference performance.
AMD ha acquisito Taalas, startup canadese specializzata in acceleratori per l'inferenza AI progettati attorno ai singoli modelli. La tecnologia, pensata per ridurre colli di…
AMD acquisisce Taalas e punta sulla Hard Coded Inference: pesi dei modelli AI integrati nel chip per ridurre memoria, latenza e costi.
AMD is buying Toronto startup Taalas, which etches a single AI model into silicon for fast, cheap inference, its latest bid to challenge Nvidia.
The Toronto-based startup, founded in 2023, has raised $219 million and builds chips hardwired for specific AI models
AMD announced Thursday that it will buy Taalas, the latest Canadian semiconductor startup to be acquired by a U.S. company. Read more
AMD is buying Canadian startup Taalas, which hard-codes model weights directly into inference chips. That makes them extremely fast but locks each chip to a single model. A demo…