Nvidia backs SiFive
Nvidia led a $400 million financing round for SiFive to accelerate open RISC‑V CPU designs targeted at AI data centres. (parameter.io) The investment highlights how hardware and open CPU work remain central to the infrastructure that supports large‑scale AI research and deployment. (parameter.io)
Nvidia has joined a new $400 million funding round for SiFive, a chip-design company building RISC-V processor technology for artificial intelligence data centers. (sifive.com) SiFive announced the oversubscribed Series G round on April 9, 2026, and said it values the company at $3.65 billion. Atreides Management led the financing, with Nvidia, Apollo Global Management, Point72 Turion, T. Rowe Price Investment Management, Prosperity7 Ventures, and Sutter Hill Ventures also participating. (businesswire.com) SiFive said the money will speed up its “high-performance data center roadmap” and expand its RISC-V central processing unit and artificial intelligence intellectual property for server systems. Reuters reported the company is treating this as its final private round before a planned initial public offering. (sifive.com) (msn.com) RISC-V is the basic instruction set that tells a processor how to do its work, like a common grammar for chips. RISC-V International describes it as an open standard, which means companies can build around it without relying on a single owner such as Arm or Intel. (riscv.org) SiFive does not run chip factories; it sells processor designs and related intellectual property that other companies can license and turn into silicon. The company says its products target data centers, edge computing, mobile devices, automotive systems, and aerospace hardware. (sifive.com) Nvidia’s interest did not start with this financing round. On January 15, 2026, SiFive said it would integrate Nvidia NVLink Fusion, Nvidia’s chip-to-chip connection technology, into SiFive’s data-center-class RISC-V platforms so those processors can link more tightly with Nvidia graphics processing units and other accelerators. (sifive.com) That pairing fits how artificial intelligence servers are built in 2026. Graphics processing units handle most of the model training math, but central processing units still manage memory, storage, networking, scheduling, and the software layers that keep large clusters running. (sifive.com) (techcrunch.com) SiFive was founded in 2015 by Krste Asanović, Yunsup Lee, and Andrew Waterman, researchers tied to the University of California, Berkeley work that helped launch RISC-V. More than a decade later, the company is raising one of the largest recent bets on open central processing unit designs for artificial intelligence infrastructure. (wikipedia.org) (thenextweb.com)
