The CPU Returns as a Protagonist: $400 Million for RISC-V to Challenge Traditional Architectures

SiFive has announced the closing of a $400 million Series G funding round, bringing its total valuation to $3.65 billion. The operation, led by Atreides Management, saw participation from major investors including NVIDIA, Apollo Global Management, and T. Rowe Price. The declared objective is to accelerate the development of high-performance RISC-V solutions for the datacenter, focusing specifically on workloads related to so-called "agentic AI," that is, systems capable of orchestrating complex processes autonomously.

According to the company, the growing demand from hyperscalers is highlighting the need for open and highly customizable alternatives to proprietary architectures. In this context, RISC-V is presented as a flexible platform, capable of offering custom-designed CPUs in the form of IP.

The funding will be utilized to expand the technology roadmap in three main directions:

• Development of RISC-V CPUs with advanced scalar, vector, and matrix capabilities
• Strengthening the software ecosystem, with support for platforms such as CUDA, Red Hat Enterprise Linux, and Ubuntu
• Collaboration with industrial partners to facilitate adoption, including through technologies such as NVLink

One of the central theses supported by SiFive is the return of the CPU's centrality in AI systems. While GPUs and accelerators remain essential for massive processing, CPUs are considered crucial for managing and coordinating complex workflows. In advanced AI scenarios, energy efficiency and orchestration capability become critical factors. RISC-V architectures, thanks to their modular and open nature, promise a better performance-to-watt ratio compared to legacy solutions.

Several analysts highlight how the growth of AI workloads is challenging the limits of traditional CPU architectures. The open standard approach of RISC-V allows for greater design freedom, fostering innovation especially in hyperscale contexts. The direct involvement of semiconductor and cloud companies suggests a growing interest in highly customizable CPUs that can quickly adapt to the evolution of AI models.

If this trend solidifies, SiFive could carve out a significant role in a potential market estimated at over $100 billion, related to next-generation infrastructure for AI and datacenters.