SiFive Secures $400 Million in Oversubscribed Funding Round
SiFive, a company founded in 2015 by engineers from UC Berkeley known for their contributions to open-source chip design, has successfully raised $400 million in an oversubscribed funding round, bringing its valuation to $3.65 billion.
Significance of the Funding Round
This funding round is noteworthy for several reasons. Notably, SiFive’s RISC-V open chip architecture deviates from the dominant Intel x86 and ARM architectures, which currently support Nvidia’s AI-driven GPU computing ecosystem.
Key Investors in the Round
Nvidia participated as an investor in this latest funding round, which was primarily led by Atreides Management, founded by Gavin Baker, a former Fidelity executive. Other significant investors include Apollo Global Management, D1 Capital Partners, Point72 Turion, and T. Rowe Price, among others. Notably, Atreides Management was also involved in Cerebras Systems’ recent $1 billion funding round.
SiFive’s Business Model Explained
SiFive operates on a business model reminiscent of Arm’s earlier strategies — it licenses its chip designs to clients who then customize them rather than selling the chips directly. This contrasts sharply with Arm’s recent pivot to manufacturing its first proprietary AI chip in collaboration with Meta, which caters to an array of clients, including OpenAI and Cerebras.
A Unique Position in the Market
SiFive occupies a unique niche in the semiconductor landscape, offering open chip designs that are neutral and not tied to specific customers. According to PitchBook estimates, the company has not sought additional funding since raising $175 million in March 2022, a round that valued it at $2.33 billion and included investment from notable players like Intel Capital and Qualcomm Ventures.
Expansion Into AI Data Centers
Initially known for applications in smaller-scale embedded systems, RISC-V is gaining traction in the realm of AI data centers due to this recent funding and Nvidia’s interest. SiFive’s RISC-V designs are set to integrate with Nvidia’s CUDA software and its NVLink Fusion server architecture, enabling seamless connectivity between various CPUs within Nvidia’s AI ecosystems.
Nvidia’s Strategic Support for SiFive
In a competitive landscape where rivals such as Intel and AMD strive to match Nvidia’s GPU capabilities, Nvidia’s backing of the 11-year-old startup signifies a strategic move towards supporting open and innovative CPU design alternatives.