MOUNTAIN VIEW, Calif. - FuriosaAI announced a partnership with Broadcom to develop its third-generation AI accelerator chip.
The collaboration combines FuriosaAI's Tensor Contraction Processor architecture with Broadcom's networking technology to create a multi-die chiplet system for AI inference workloads in data centers.
FuriosaAI's current product, RNGD, is a 180-watt, PCIe-based accelerator manufactured using TSMC's 5nm process. The chip is in mass production and deployed by Samsung SDS and LG AI Research.
"Inference performance is no longer defined solely by raw compute. It is increasingly a function of data reuse and communication efficiency across servers and racks," said Charlie Kawwas, president of Broadcom's Semiconductor Solutions Group.
The third-generation chip will feature a 2nm compute die and HBM4/4E memory, using Broadcom's packaging technology to integrate multiple silicon dies.
"Having proven the performance and efficiency of our architecture with RNGD, our second-generation chip now in mass production with TSMC, we will deliver a third-generation inference solution," said June Paik, FuriosaAI cofounder and CEO.
Sampling of the third-generation chip is scheduled to begin in the first half of 2028.
