FuriosaAI scales inference beyond chips

Broadcom partnership targets hyperscale AI inference clusters.

May 22, 2026

Jon Peddie

FuriosaAI is moving beyond stand-alone AI accelerators and into full-scale inference infrastructure. The South Korean start-up has partnered with Broadcom to develop a third-generation AI platform that combines chiplets, HBM4 memory, Ethernet fabrics, advanced packaging, and large-scale cluster interconnects. The effort builds on Furiosa’s mass-production RNGD processor and reflects a broader industry shift toward inference-optimized architectures designed for agentic AI and large language models. The company aims to deliver greater token throughput, improved performance per watt, and simplified software deployment for hyperscale data centers. FuriosaAI enters its next phase FuriosaAI has spent nearly a decade developing AI processors focused on

...

Enjoy full access with a TechWatch subscription!

TechWatch is the front line of JPR information gathering service, comprising current stories of interest to the graphics industry spanning the core areas of graphics hardware and software, workstations, gaming, and design.

A subscription to TechWatch includes 4 hours of consulting time to be used over the course of the subscription.

Already a subscriber? Login below

FuriosaAI scales inference beyond chips

Enjoy full access with a TechWatch subscription!

This content is restricted

Login