Nvidia backs SiFive

Nvidia has backed SiFive in a $400 million financing round, giving the RISC-V chip-design company a $3.65 billion valuation. (sifive.com) SiFive announced the Series G round on April 9, 2026, saying it was oversubscribed and led by Atreides Management, with Nvidia, Apollo Global Management, Point72 Turion, T. Rowe Price Investment Management, Prosperity7 Ventures, and Sutter Hill Ventures also participating. (businesswire.com) The company said it will use the money to expand its data-center roadmap, add engineering staff, and build more software around its processor designs, including existing ports of CUDA, Red Hat, and Ubuntu. (sifive.com) SiFive does not manufacture chips. It sells processor blueprints, known as intellectual property, that customers can customize for their own semiconductors, and Chief Executive Patrick Little told Reuters the new round is likely its last before a public offering filing. (usnews.com) Those blueprints are built on RISC-V, an open instruction set architecture, which is the basic rulebook that tells a processor how to run software. RISC-V International, the nonprofit that governs the standard, says the architecture is open for anyone to use as a building block in proprietary or open products. (riscv.org) That open model puts SiFive in a different lane from Arm and x86, the dominant processor families in servers. Reuters reported that Little sees an opening now that Arm has started selling its own chips, turning a longtime supplier into a potential competitor for some of its licensees. (usnews.com) Arm made that shift explicit on March 24, 2026, when it launched the Arm AGI central processing unit, its first production silicon product, with Meta as the first customer. Arm said the chip is aimed at artificial intelligence data centers and agentic artificial intelligence workloads, where software systems coordinate many steps and tools on their own. (newsroom.arm.com) SiFive is pitching the same control layer. In its funding announcement, the company said central processing units handle orchestration tasks that graphics processing units and accelerators are not designed to manage efficiently, especially as artificial intelligence systems add more routing, scheduling, and state management. (businesswire.com) Nvidia’s role is not limited to writing a check. SiFive said it is working with industry partners on deployment paths that include Nvidia NVLink Fusion, a program Nvidia announced on March 31 with Marvell to let more third-party chips plug into Nvidia’s artificial intelligence infrastructure. (sifive.com) (nvidianews.nvidia.com) The immediate result is that Nvidia gains a stake in a company trying to supply the general-purpose processors that sit next to artificial intelligence accelerators in data centers. SiFive gets fresh cash, a higher valuation, and a clearer path toward testing whether public investors will buy into open-standard central processing unit designs for the artificial intelligence buildout. (usnews.com) (sifive.com)