AMD acquires AI chip startup Taalas to boost inference performance by etching models into silicon
Summary
Early tech demos show model-specific integrated circuits churning out up to 17,000 tokens a second
Discussion on
Trending posts from X.
🚨 BREAKING: AMD just acquired Taalas
— NIK (@ns123abc) August 6, 2026
Nvidia bought Groq for $20B
> SRAM-based LPUs
> model-agnostic, low-latency
> run any model, fast
Now AMD buys Taalas
>model-specific ASICs
>hardwire the model into the chip
>one model, max efficiency, zero flexibility
AI inference wars… pic.twitter.com/M9fBhdUuoa
$AMD agrees to acquire Taalas which it calls a “pioneer in specialized AI inference silicon.”
— Shay Boloor (@StockSavvyShay) August 6, 2026
The deal expands AMD beyond GPUs and deepens its push into dedicated inference hardware as competition accelerates. pic.twitter.com/2d0D6ZzrZA
We are pleased to share that Taalas has agreed to join AMD.
— Taalas Inc. (@taalas_inc) August 6, 2026
We built Taalas to rethink AI inference from the ground up: hardware designed around the model, rather than the other way around. The result is the world's fastest and most cost-effective inference silicon.
Joining AMD…
$AMD just announced it has agreed to acquire Taalas
— Evan (@StockMKTNewz) August 6, 2026
"a pioneer in specialized AI inference silicon" pic.twitter.com/ca8zer1XR2