LaunchAI ModelsNovember 17, 2026

Axelera unveils Titania inference chiplet with native FP8 and FP4

Read original source →futurumgroup.com

Each Titania chiplet packs 32 AI cores and 128 MB shared SRAM on TSMC N4C for 1.25 PFLOPS FP8; an 8-package AX8 server claims 20 PFLOPS FP8 at ~6 kW. Production starts Q3 2027 against over $300M in confirmed commitments.

1 source

More stories today

Open the live feed