本文探讨了RISC-V开源芯片架构在全球的兴起,特别是在中美科技竞争背景下的发展。文章分析了RISC-V如何成为中美半导体设计竞争的焦点,以及其开放标准面临的挑战和地缘政治影响。
The U.S. and China have been trying to dominate semiconductor design and manufacture. But the creators of one type of semiconductor design wanted to make it free for anyone in the world to build upon. Today, when we talk about semiconductor chips, we often talk about competition between the U.S. and China and how the U.S. is trying to protect that technology through export controls. The creators behind one category of semiconductor design have a different approach. They wanted to make it open standard, free for everyone anywhere in the world to build upon. But as NPR's Emily Feng reports, this vision is increasingly at risk. Back in 2010, Professor Krste Asanovic and his students at the University of California, Berkeley set out to make a new instruction set for chip architecture, a set of broad standards and design principles kind of like a blueprint for a semiconductor. There was no grand vision. It was only for ourselves and our friends. But fast forward 15 years. That architecture, known by its acronym RISC-V, has become globally adopted, in part because RISC-V is cheap. To use most other commercial chip architecture, like from Intel or U.K.-based Arm, you'd have to pay a royalty, but Asanovic says anyone can innovate on RISC-V, modifying RISC-V's blueprint for free to make their own products, like processors and software. This is going to enable more competition, more innovation because you're not relying on one vendor. One of the big adopters of RISC-V have been semiconductor design companies in China. And here, at one of the most recent RISC-V meetups in Beijing, full of coders and entrepreneurs... People shared ideas and the latest products being designed by software teams in China, all based on RISC-V. Software entrepreneur Wu Wei is the head of this Beijing chapter of RISC-V... And tells NPR in an interview later that while there has been huge interest from Chinese companies, many are immature, disappearing once their investment funding fizzles out. But given the ever more complex hardware and software semiconductor needs, Wu says it is imperative that experienced tech workers, including those in China, collaborate with people worldwide. Yet RISC-V's idealistic vision of global research on chips, unhindered by questions of sovereignty or geopolitics, is butting up against technology protectionism, especially in the U.S., where American lawmakers have questioned whether RISC-V is helping China innovate faster. I would argue it is helping them innovate more quickly. This is Jacob Feldgoise, a fellow at the Washington-based think tank CSET, which focuses on technology research. But it's not handing them a finished product or a finished design. Plus, Feldgoise argues, American companies on the whole benefit from being able to access a global tech standard, like they do with Wi-Fi or Bluetooth or cellular 5G, all of which are global standards. So there's a lot more transparency into the development process, into the details. There's also more opportunity for companies to take action to become involved in the development of this technology. So the solution, Feldgoise argues, is not to ban American or Chinese firms from selling RISC-V products, but to encourage more American development of RISC-V. But the shadow of the U.S.-China rivalry is long, and it has complicated RISC-V's global mission. The RISC-V Foundation ended up moving their legal registration to Switzerland a few years ago, Asanovic says, to signify their neutrality in the U.S.-China tech rivalry that started heating up in 2020. So India, Europe, Japan as well as China were very worried about U.S. controlling the technology. Asanovic says the U.S. could not effectively export-control the RISC-V standard anyways. It is online and free to use. It just doesn't make any sense, so it's just illogical. And he argues, even if the U.S. could, China would just come up with its own open-standards architecture. Building an instruction set and building an ecosystem around it takes some work, but it's been done many times in the past. But that, he says, risks the world fragmenting even more, this time along the lines of competing technology standards and designs.