SiFive has raised a new round that values the company at $3.65 billion, according to a TechCrunch report published April 11, 2026. The startup secured $400 million in an oversubscribed round led by Atreides Management. Nvidia participated as an investor, bringing attention to a CPU design approach built on RISC-V—an alternative to the two dominant CPU instruction set families that currently power data center systems: x86 and ARM.
The investment signals engineering and platform alignment: SiFive’s chip designs are open and licensable, and the company states its CPUs will integrate with Nvidia’s software and rack-level interconnect ecosystem. For developers and data-center builders, the question is how quickly an open, non-x86/ARM CPU supply chain could integrate into existing Nvidia GPU-based workflows.
SiFive’s funding round and valuation
SiFive, founded in 2015 by UC Berkeley engineers who created an open source chip design, landed a $400 million oversubscribed round that values the company at $3.65 billion. The round was led by Atreides Management, founded by former Fidelity investor Gavin Baker. Atreides previously invested in Cerebras Systems‘ $1 billion round.
Alongside Atreides, the round included Nvidia as an investor, plus a broad set of venture capital, private equity, and hedge fund participants. Named investors include Apollo Global Management, D1 Capital Partners, Point72 Turion, and T. Rowe Price Sutter Hill Ventures, among others. Nvidia’s participation is notable given the company’s GPU platform’s integration with the surrounding CPU ecosystem.
According to Pitchbook, SiFive’s previous funding round occurred in March 2022, when it raised $175 million led by Coatue Management at a pre-money valuation of $2.33 billion. That round included investors such as Intel Capital, Qualcomm Ventures, and Aramco Ventures.
RISC-V architecture: an alternative to x86 and ARM
SiFive’s chip designs are based on RISC-V, which differs from x86 and ARM, the two major CPU types that currently supply data center systems. The funding represents a push to establish RISC-V as a viable architecture for a market historically centered on x86 and ARM-based servers.
RISC-V has been primarily used for smaller applications, such as embedded systems. With this funding and Nvidia’s participation, SiFive is targeting CPUs for AI data centers. SiFive’s designs are described as both open and neutral, not reliant on specific customers. This neutrality is reinforced by SiFive’s licensing model, which operates similarly to Arm’s historical approach.
Licensing model and Nvidia integration
SiFive’s business model centers on licensing: the company licenses its chip designs to companies that modify them for their own needs and does not sell the chips themselves. This contrasts with Arm’s recent shift. In March, Arm launched its first-ever manufactured chip, an AI chip developed with Meta that serves customers including OpenAI, Cerebras, and Cloudflare.
SiFive is moving into the data-center CPU layer for AI systems. The integration story centers on Nvidia’s platform components. SiFive’s designs will work with Nvidia’s CUDA software and its NVLink Fusion, a rack server system that enables different CPUs to integrate into Nvidia’s GPU-based infrastructure. This engineering approach ensures that an alternate CPU architecture can participate in the host-and-rack topology that Nvidia supports for its accelerators.
From a systems perspective, AI training and inference deployments are typically engineered around a stack: CPU for host control and data movement, GPU for compute, and interconnects and software layers that coordinate the two. SiFive’s alignment with CUDA and NVLink Fusion indicates the company is building its CPU roadmap around established Nvidia software expectations rather than requiring a separate software ecosystem.
Competitive context and industry implications
The investment occurs as Intel and AMD compete with Nvidia’s GPU dominance. Nvidia’s backing of SiFive suggests the company is supporting an alternative CPU architecture path that remains compatible with its platform ecosystem.
The setup raises several industry questions. If SiFive’s licensing model scales and data-center integrators adopt RISC-V-based CPU designs for AI racks, the practical effect could be a broader CPU supply chain around Nvidia’s GPU systems. Alternatively, the impact could remain limited if customization and validation cycles slow adoption. SiFive’s emphasis on CUDA and NVLink Fusion compatibility points to a potential path for interoperability.
For the broader CPU landscape, this development highlights a shift toward open architectures. SiFive’s chip designs are open and neutral, contrasting with the proprietary CPU families that currently dominate data center deployments. If the market adopts open CPU designs for AI data centers, it could influence how vendors approach licensing, integration, and platform compatibility.
Source: TechCrunch