SiFive raises $400 million to bring open RISC-V chip designs to AI data centers

SiFive Secures $400 Million Funding Round with Nvidia Backing

SiFive, the chip design startup founded in 2015 by UC Berkeley engineers who created RISC-V, has closed an oversubscribed $400 million funding round that values the company at $3.65 billion 1

1

. The Series G round was led by Atreides Management, founded by former Fidelity investor Gavin Baker, with participation from Nvidia, Apollo Global Management, D1 Capital Partners, Point72 Turion, and T. Rowe Price Sutter Hill Ventures 1

1

. This marks SiFive's first fundraising since March 2022, when it raised $175 million led by Coatue Management at a pre-money valuation of $2.33 billion, bringing total outside funding to over $760 million 1

1

3

3

.

RISC-V Chip Technology Challenges Arm and Intel Dominance

SiFive's business model centers on licensing open-source RISC-V chip design rather than selling chips directly, similar to Arm's historical approach 1

1

. The open-source architecture provides a significant advantage by reducing development time for custom CPUs while remaining neutral and not reliant on specific customers 1

1

. An instruction set architecture like RISC-V includes a library of pre-packaged computing operations and supporting components such as memory management modules 3

3

. SiFive's customer base reportedly includes several of the world's largest tech firms, who have used its blueprints to create more than 500 chip designs 3

3

.

Expanding into the AI Data Center Market

The company plans to use the capital to accelerate its push into CPUs for AI data centers, a market currently dominated by Nvidia's GPU systems and Arm-based processors 2

2

. "Hyperscale customers have made it very clear that it is time to accelerate the availability of open standard alternatives for the data center," said SiFive CEO Patrick Little 3

3

. SiFive's designs will work with Nvidia's CUDA software and its NVLink Fusion, a rack server system that lets different CPUs plug into Nvidia's AI infrastructure 1

1

. The funding will support creating new products and speeding up work on compatible software for data center compute solutions 2

2

.

Advanced AI Chips and Machine Learning Accelerators

SiFive's product portfolio includes more than a dozen CPU blueprints tailored for various applications 3

3

. The company's most advanced design, the Performance P870-D, enables building server CPUs with up to 256 cores and includes data protection accelerators for tasks like encrypting network traffic 3

3

. The Performance P870-D features a cluster accelerator port for connecting to other processors and works with SiFive Intelligence machine learning accelerators 3

3

. The company introduced its flagship AI chip, the XM Gen 2, last September, optimized to process matrices where AI models store information, with 4-core processing clusters capable of 16 trillion calculations per second 3

3

.

Growing Demand for Alternative Microprocessor Designs

The investment reflects an increasing interest in using generalist microprocessors to handle AI tasks, especially those occurring after complex models have been trained 2

2

. While Nvidia's AI accelerators remain central to most AI infrastructure, hyperscalers are seeking customizable CPU solutions in IP form to differentiate their offerings 3

3

. The move is notable because Nvidia, which competes with Intel and AMD in the GPU market, is backing an 11-year-old startup designing CPUs on an open and completely alternate technology 1

1

. However, SiFive faces stiff competition in a crowded field. According to Jon Peddie Research, 135 companies are actively creating or planning AI chips, with investors providing $28.8 billion to startups in this area since 2000, though only about 25 specialized AI processor companies are projected to survive by 2030 2

2

.