I would rather expect Akida as an accelerator in the N3 / V3 Version.
The CSS-N3 offers a 20% performance-per-Watt improvement per core over the CSS-N2. This CSS design comes with between 8 and 32 cores, with the 32-core version using as little as 40 W. It’s intended for telecoms, networking, DPU, and cloud applications and can be used with
on-chip or separate AI accelerators. The new N3 core is based on Armv9.2 and includes 2 MB private L2 cache per core. It supports the latest versions of PCIe, CXL and UCIe.
When switching from N2 to N3, the maximum L2 cache doubles to 2 MByte. AI algorithms benefit the most from this - the performance leap is achieved without new data formats. The common INT8 and Bfloat16 variants are already supported by the 2 generation.
View attachment 57466
"Today we are proud to announce that Arm Total Design has expanded to include over 20 partners, all committed to ensuring broad accessibility of performant, efficient solutions that will help meet the computing demands of an AI-accelerated future. These partners are already working together on everything from verifying IP, and customizing firmware, through to building chiplets on the world’s most advanced process nodes – all using Neoverse CSS."
Link:
https://www.eetimes.com/arm-updates-css-designs-for-hyperscalers-custom-chips/
Link:
https://www.heise.de/news/ARM-blaest-zum-Angriff-auf-AMD-Intel-und-teils-Nvidia-9634864.html
Link:
https://newsroom.arm.com/news/enabling-ai-infrastructure-on-arm