Fetching from the wire…
Public story · 2026-02-20 · source-backed
Canadian startup Taalas unveiled its HC1 chip — a hardwired implementation of Llama 3.1 8B that generates 17,000 tokens per second per user, 73x faster than NVIDIA's H200 at one-tenth the power. Using aggressive quantization (3-bit and 6-bit) on TSMC N6 at 815mm2 die size, drawing ~250W per card. Over $200M raised, ~25 employees. Second chip for a reasoning model by early summer. This represents a radical approach: model-specific silicon trading flexibility for extreme speed.
Each link below shares sources, entities, or timing with this story.
Taalas released HC1 / Shared entities / Same source / Shared topic
Linked by a graph relationship (Taalas released HC1); both cover HC1, Taalas; cite the same source (HC1 chip).
Meta released Llama / Shared entity: NVIDIA / Shared topic / What happened next
Linked by a graph relationship (Meta released Llama); both cover NVIDIA; overlapping topics (aggressive, chip, custom, silicon).
Taalas released HC1 / Shared entities / Shared topic / What happened next
Linked by a graph relationship (Taalas released HC1); both cover Llama, Taalas; overlapping topics (chip, taala).
Meta released Llama / Shared entity: Nvidia / Shared topic / What happened next / Tension
Linked by a graph relationship (Meta released Llama); both cover Nvidia; overlapping topics (chip, custom).
Meta released Llama / Shared entity: Nvidia / Shared topic / What happened next
Linked by a graph relationship (Meta released Llama); both cover Nvidia; overlapping topics (chip, custom, silicon).
Meta released Llama / Shared entities / What happened next
Linked by a graph relationship (Meta released Llama); both cover Llama, Using; picks up the Llama thread on 2026-05-01.
NVIDIA released Blackwell / Shared entities / What happened next
Linked by a graph relationship (NVIDIA released Blackwell); both cover NVIDIA, Tokens; picks up the NVIDIA thread on 2026-04-03.
Meta released Llama / Shared entity: Nvidia / Shared topic / What happened next
Linked by a graph relationship (Meta released Llama); both cover Nvidia; overlapping topics (chip, custom).