News
SiFive to adopt Nvidia technology for high-speed chip-to-chip interconnects
Thursday, January 15, 2026 at 02:44 PM
SiFive is integrating Nvidia technology to enhance high-speed chip-to-chip interconnects, specifically targeting the integration of RISC-V processors with AI accelerators.
Context
SiFive has announced it will adopt Nvidia’s NVLink-C2C (Chip-to-Chip) interconnect technology to enable high-speed communication between its RISC-V processors and custom chiplets. This integration allows SiFive to link its open-standard CPU cores directly to Nvidia’s AI accelerators, bridging the gap between the RISC-V ecosystem and high-performance AI hardware. The move is designed to support the growing demand for modular, custom silicon in data centers and generative AI infrastructure.
The NVLink-C2C technology provides a massive leap in performance, offering up to 900GB/s of bidirectional bandwidth. It is estimated to be 25x more energy-efficient and 90x more area-efficient than standard PCIe Gen 5 connections. For investors, this partnership solidifies SiFive as a primary challenger to Arm in the high-end compute space, providing chip designers with a scalable, low-latency path to build next-generation AI systems using Nvidia’s industry-standard interconnect fabric.
Related Companies
Nvidia
NVDA