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

Nvidia-backed SiFive hits $3.65 billion valuation for open AI chips | TechCrunch

SiFive, a company founded in 2015 by the UC Berkeley engineers who created an open source chip design, has landed a $400 million oversubscribed round that values the company at $3.65 billion. This deal is interesting for a bunch of reasons. For one, SiFive's RISC-V open chip design is based on the RISC processor, not Intel's x86 or ARM, the two major types of CPUs that currently feed Nvidia's GPU computer system AI empire. Also, Nvidia was investor in this round, alongside a long list of VCs, private equity, and hedge funds. The round was led by Atreides Management, founded by former Fidelity investor bigwig Gavin Baker. (Atreides was also an investor in Cerebras Systems $1 billion round). Other investors in the round include Apollo Global Management, D1 Capital Partners, Point72 Turion, T. Rowe Price Sutter Hill Ventures, and others. SiFive's business model is like Arm's was in years gone by -- it licenses its chip designs to those who modify them for their own needs and does not sell the chips themselves. (In March, Arm changed its model when it launched the first-ever chip it manufactured, an AI chip, developed with Meta with customers including OpenAI, Cerebras, and Cloudflare.) SiFive stands in rarified air with chip designs that are open, not proprietary, as well as neutral, not reliant on specific customers. In fact, SiFive hasn't raised since March 2022, Pitchbook estimates, when it brought in $175 million led by Coatue Management at a pre-money valuation of $2.33 billion. Intel Capital, Qualcomm Ventures, Aramco Ventures, were part of that round. RISC-V has been, until recently, better known as a chip for smaller uses, like embedded systems. But with this cash and Nvidia's attention, SiFive is moving into CPUs for AI data centers. 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 factory." In other words, 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.

SiFive to Fuel Data Center Push With $400 Million Funding Round

Chip startup SiFive Inc. received $400 million in funding from a group of investors led by Atreides Management, money the company plans to use to get a bigger foothold in AI data centers. Nvidia Corp., Apollo Global Management and Point72 also participated in the fundraising, which valued SiFive at $3.65 billion, according to a statement Thursday. The round was oversubscribed, meaning the startup was offered more money than it ultimately raised. SiFive, which uses a chip technology standard called RISC-V, is part of a group of upstarts trying to break into a lucrative market dominated by Nvidia. Data centers rely on that company's AI accelerators to develop and run artificial intelligence models. But there's a growing demand for alternatives. SiFive aims to sell technology that can be rapidly turned into microprocessors by companies trying to build their own components. So-called hyperscalers -- the world's largest data center operators -- currently develop in-house processors using designs from Arm Holdings Plc. The RISC-V standard would offer a rival option. But so far, this technology is mainly used in less sophisticated components, rather than powerful microprocessors or AI accelerators. SiFive sees that changing. "Hyperscale customers have made it very clear that it is time to accelerate the availability of open standard alternatives for the data center," Chief Executive Officer Patrick Little said in the statement. That company is hoping to exploit a growing trend: While Nvidia's accelerators are still at the center of most AI machinery, there's an increasing interest in using more generalist microprocessors. The idea is to tap central processing units, or CPUs, to handle more AI tasks -- especially ones occurring after the complex models have been trained. Nvidia is offering a new version of its CPU as a standalone product for the first time. And Arm is rolling out this kind of chip as well. Veteran CPU sellers Intel Corp. and Advanced Micro Devices Inc., meanwhile, have pointed to strong demand for their server chips. SiFive will use the money to create new products and speed up work on making compatible software, according to the statement. Still, the Santa Clara, California-based company is facing a crowded field. According to Jon Peddie Research, there are 135 companies actively creating or planning to make AI processors. Investors have given startups in this area $28.8 billion since 2000. By 2030, about 25 specialized AI processor companies will have survived, the research company projects.

RISC-V chip design startup SiFive nabs $400M investment - SiliconANGLE

SiFive Inc., a startup that sells chip designs based on the open-source RISC-V architecture, has raised $400 million in funding at a $3.65 billion valuation. Atreides Management led the Series G round. SiFive stated in today's funding announcement that Nvidia Corp., enterprise technology investor Apollo and other institutional backers contributed as well. The company's total outside funding now exceeds $760 million. Central processing units crunch data using a library of pre-packaged computing operations. Some of those operations perform simple mathematical calculations. Others carry out more specialized tasks such as moving data between different memory addresses. A CPU's library of computing operations is known as its instruction set architecture, or ISA. SiFive was founded in 2015 by the creators of RISC-V, a popular open-source ISA. It includes not only a library of pre-packed computing operations, or instructions, but also supporting components such as memory management modules. RSIC-V can significantly reduce the amount of time needed to develop a custom CPU. However, the task is still highly resource-intensive and requires specialized skills. SiFive sells ready-to-use RISC-V processor designs that customers can use to speed up their semiconductor projects. The company's product portfolio includes more than a dozen CPU blueprints. Some include reliability features that make them suitable for use in vehicles, while others are geared towards battery-powered connected devices. SiFive also offers a growing lineup of data center chips. The company's most advanced design, the Performance P870-D, makes it possible to build server CPUs with up to 256 cores. It includes a module called an interrupt controller that optimizes the order in which calculations are carried out based on their importance. Data protection accelerators speed up tasks such as encrypting network traffic. The Performance P870-D includes a so-called cluster accelerator port that can be used to connect it to another processor. A cloud provider, for example, could integrate the CPU with a custom graphics card. The Performance P870-D also works with SiFive's own SiFive Intelligence line of machine learning accelerators. The company introduced its flagship AI chip last September. The XM Gen 2 is optimized to process matrices, the mathematical objects in which AI models keep information. XM Gen 2 makes it possible to assemble an accelerator from multiple 4-core processing clusters that can each perform 16 trillion calculations per second. SiFive's customer base reportedly includes several of the world's largest tech firms. Those customers have used its blueprints to create more than 500 chip designs. "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 chief executive officer Patrick Little. "Their consistent ask is for customizable CPU solutions in IP form, that will enable them to meaningfully differentiate their data center compute solutions."