SiFive Raises $400M at $3.65B Valuation With NVIDIA Backing to Build Open RISC-V CPUs for AI Data Centers

SiFive, the company founded by the UC Berkeley engineers who created the RISC-V open chip architecture, has closed an oversubscribed $400M Series G round at a $3.65B valuation. NVIDIA is among the investors. The financing will fund open-standard CPU designs for AI data centers, directly competing with Intel and AMD in the infrastructure layer that runs enterprise agent workloads.

The Round

The Series G was led by Atreides Management, the fund run by former Fidelity investor Gavin Baker that also backed Cerebras Systems’ $1B Series H. Other investors include Apollo Global Management, Point72 Turion, T. Rowe Price, and returning backers Prosperity7 Ventures and Sutter Hill Ventures, according to TechCrunch.

SiFive hadn’t raised since March 2022, when it brought in $175M led by Coatue Management at a $2.33B pre-money valuation. That round included Intel Capital, Qualcomm Ventures, and Aramco Ventures. The jump from $2.33B to $3.65B reflects the strategic value of open-architecture CPU IP as AI data center buildouts accelerate.

Why NVIDIA Is Backing an Open CPU

SiFive’s business model mirrors what ARM’s was before its March 2026 pivot to manufacturing its own chips: it licenses RISC-V processor designs that customers modify for their own needs. The designs are open-standard, hardware-neutral, and not dependent on Intel’s x86 or ARM instruction sets.

The NVIDIA connection is the strategic core. SiFive’s RISC-V designs integrate with NVIDIA’s NVLink Fusion rack server system and are compatible with NVIDIA’s CUDA software stack, according to SiFive’s press release. This means enterprises and cloud providers building GPU-centric AI data centers can plug open-design RISC-V CPUs alongside NVIDIA accelerators.

As TechCrunch noted, “as rivals Intel and AMD seek to compete with Nvidia’s GPU, Nvidia is backing an 11-year-old startup that can design CPUs on an open and completely alternate technology.”

The Agentic Workload Framing

SiFive is explicitly framing RISC-V as infrastructure for agentic AI. CEO Patrick Little said in the press release: “As the industry urgently evolves toward agentic AI, SiFive is doubling down on the data center.”

The company’s argument: CPUs handle the orchestration, coordination, and system-level tasks that GPUs are not designed for. As AI moves from batch inference toward complex agentic workflows with branching decisions, tool calls, and multi-step execution, efficient CPU performance becomes a bottleneck. SiFive claims RISC-V’s lower power consumption and standards-based architecture let data centers scale compute capacity within existing power envelopes.

“The CPU is suddenly exciting again, especially for applications in the data center,” said Dan Newman, CEO of The Futurum Group, in SiFive’s announcement.

The Investment Will Fund

SiFive plans to allocate the $400M across three areas: advanced R&D on high-performance RISC-V CPU, accelerator, and system IP; software ecosystem development including CUDA, RedHat, and Ubuntu ports; and customer enablement for NVLink Fusion deployment, per the press release.

The timing matters for the broader AI infrastructure stack. ARM launched its first in-house AI chip in March 2026, developed with Meta and sold to OpenAI, Cerebras, and Cloudflare. RISC-V entering the same market with NVIDIA’s backing creates a three-way architecture competition (x86, ARM, RISC-V) for the CPU layer of AI data centers.