When the umpire chooses a team, maybe it's time to look for a new umpire.
After Nvidia announced last week its plans to acquire Arm, the chip industry was turned upside down. Most large chip companies — save for Intel — license Arm's chip architecture in order to build their own hardware. Arm co-founder Hermann Hauser has an explanation for the company's runaway success: neutrality. Arm doesn't make its own chips, so it doesn't compete with its customers.
But the Nvidia deal, if it goes through, could change that. For many chip companies, licensing Arm's technology could mean giving money not to a neutral company, but to a rival — a deeply uncomfortable position for any company to find itself in.
So the acquisition was assumed by many to be a huge boost to RISC-V, the only viable Arm competitor. RISC-V is an open-source instruction set architecture, backed by tech companies including Huawei and Google, which lets chip companies design RISC-based chips without paying any licensing fees. Since the project began at UC Berkeley in 2010, an ecosystem has flourished around it: Qualcomm, Samsung, Alibaba and even Nvidia use RISC-V in their chips now, while startups such as SiFive are trying to design full-fledged Arm processor competitors using the technology.
Calista Redmond, CEO of RISC-V international, said the Nvidia-Arm deal hasn't changed anything just yet. "We've been growing so fast over the last five years that we're not noticing a significant swing at any one of these inflection points across the industry," Redmond told Protocol. But, she added, the acquisition "may be a call to evaluate choices for those who have not already adopted RISC-V."
In a conversation with Protocol, Redmond outlined RISC-V's priorities, described the project's hopes for Android support, and predicted that the organization is a couple of years away from its version of Arm's iPhone moment.
This interview has been edited for clarity and length.
How does the proposed Nvidia deal change things for RISC-V?
There are a couple of things. First, diversity and choice is what open source is all about, and diversity and choice is something that has already been exercised by nearly all of our members. When you build a business, you fundamentally do not want to have all of your dependencies on any one other company. You don't want one client, you don't want one innovation partner. You want to have that diversity and choice, and so many of our members already have that — they're already working on multiple architectures.
Now many of our entrepreneurs and new entrants to RISC-V have really designated RISC-V as their base building block. But RISC-V is not like having a dependency on one company. RISC-V is an open and shared building block, so in fact it swings things wildly in the other direction, in that your choice is unlimited. Your choice of innovation partners, of what direction you want to take the technology [in] is fundamentally wide open, from the smallest microprocessor to the largest SoC.
And so when it comes to proposed movements or reorchestration of the players in the industry, our membership is somewhat insulated already, when they have already got a diversity of choice going on here. Now, it may be a call to evaluate choices for those who have not already adopted RISC-V. So there may be folks that say "maybe we need to look at multiple architectures a little more closely." We've seen some of that activity happening, but … we're probably 12 to 18 months out from even knowing whether or not a potential [Nvidia-Arm] transaction will take place.
The other piece of the equation is that RISC-V fundamentally opens up the technology. I've been talking about the business model and freedom of choice. But the other side of that is freedom of opportunity. You can take RISC-V into any geography, into any domain, into adjacent spaces where you've already got a business built, and that expands opportunity, at the same time that you've expanded what goes into your product or offering. So you have all of these partners you can work with, you now have a broad set of clients that you can pursue as well.
For us, a transaction in the industry such as the proposed [Nvidia-Arm] acquisition doesn't change our equation at all. It may inspire more interest. Honestly, we've been growing so fast over the last five years that we're not noticing a significant swing at any one of these inflection points across the industry. You could have asked me a week ago or a month ago "has COVID impacted RISC-V?" and I would say: Lucky for us, the technology industry has already been in a virtual space, and we're continuing a lot of the programming work that's been going on and we haven't noticed any big swings there either. So we haven't noticed slowdowns, we haven't noticed any substantial uptakes. Many folks are waiting to see where this transaction lands.
You have lots of Chinese companies among your members. Have you been seeing increased interest from them, given the current geopolitical climate between the U.S. and China, and the question of if restrictions come on Arm, many of them might not have an architecture?
Most of the companies that you would consider in the boat of concern there have already been engaged in RISC-V. There's been no change in that: All the big fish are already working on RISC-V. Other organizations in China have been steadily coming on board and growing pretty rapidly over time, but there hasn't been a swing relative to the proposed transaction.
There was just as much concern about geopolitical [risks] in the last Arm transaction or the MIPS transactions. When players move around, there is an opportunity to take a second look at: Where are the barriers for me? Are they geopolitical, are they company related, is it technology related, is it a business model relation? If you're operating in one of these multinationals, you're looking at the entire globe, you're not just looking at what's going on between the U.S. and China.
Important to consider is our membership today is very solidly a third North America, a third Europe and a third China. If you look at some of the news coming out of Europe, they're equally concerned about having technical independence and technical sovereignty. You see the European Commission doing things like the European Processor Initiative, where they've designated RISC-V as a technical building block because it is not tied to any one country.
Calista Redmond, CEO of RISC-V international, says here organization's open-source technology should have its iPhone moment in the next couple of years.Image: RISC-V
One last thing on the Nvidia deal: As you mentioned, they are a member of the foundation, and they use RISC-V technology. Do you expect them to leave the foundation now? Is that a conflict of interest?
They have been a leader in RISC-V from the beginning. They have a seat on the board of directors, and Jensen [Huang, its CEO] has publicly stated that they are continuing with no pivot in their investment in RISC-V. You look at any large company, and they have multiple strategies across multiple parts of the organization, and where it makes sense to go external to get from point A to B, that's what they do. You can go external through partnerships, through acquisitions, through divestiture, through open source, through one-to-one relationships or other types of consortiums, through joining standards bodies. There are many ways to engage externally as a company, and Nvidia has publicly stated that they have an ongoing commitment to RISC-V. Take their words, not mine — I of course wouldn't speak for them.
By the way, Nvidia has been a great open-source citizen for a very long time. They were a board member, as well as a founder, in OpenPOWER, same thing with RISC-V. Many aspects of their portfolio lean heavily on open source, so this is not a new frontier for them, and I do not think it would indicate a wild pivot in their strategy.
What's the focus area for RISC-V at the moment? Do you have a particular end-market priority, be that phones, servers, IoT?
The early entrants into RISC-V have really kind of circled around embedded or microprocessors within other designs. Nvidia has been shipping RISC-V in their GPUs for a while now. Western Digital has pivoted in memory controllers, and [is] bringing [its] whole portfolio over to RISC-V. Alibaba has announced and is doing AI in the cloud for RISC-V. That's where the early traction has been. We are now seeing that branch off into many different domains and industries. So I mentioned earlier HPC in Europe. We have a lot of additional scale out and hyperscale type interest around specific new workloads, so AI or AR/VR — Facebook's Oculus team has been involved for a long time.
We see things moving into automotive. One of the great things about open source is our religious adherence to transparency, and transparency equals a lot more security — you can see everything that's going on. And so we see it moving into automotive very rapidly, around aspects of safety.
We see just in the last 10 days, at the Linley Processor Conference, SiFive bringing out a PC. We have Qualcomm on the board of directors as well, they've got a great investment in RISC-V. We see lots of adoption going on, and moving into some of those traditional spaces that have relied on battery power, rather than a plug in the wall, whether it's the scale out in the server space, or mobile or other battery powered things. You see it in security cameras and other things that leverage compute right at the point of ingest. We see a lot of things starting to bring us across the full spectrum of computing.
If you say that the iPhone was kind of a step change in adoption of ARM processors, how far away do you think we are for something similar happening with RISC-V?
I think you're going to see it coming about far more rapidly than the onramp that prior architectures had seen. I think our onramp is far faster. I think you're going to see it proliferate within some industries within the next couple of years.
There are more design companies focused on RISC-V designs now than any other architecture. Part of that was this surge in entrepreneurs that are embarking on this. So you're going to see very rapidly, in the next couple of years, some of those really come to prominence and to gain traction with volumes.
On the mobile side of things, are you working at all to get Android running on RISC-V?
It's not something that we have solidified just yet. We have a lot of companies with very strong interest in participating in that. We need to show the appropriate levels of volume prior to getting everyone on board. At the end of the day, it's a business case for everyone.
What do you see as the biggest challenges for RISC-V in the next six months? And what are your biggest priorities?
We continue to build out our ecosystem, both in engaging thought leaders around the industry, and we need to catch up with the demand.
The demand is enormous. To catch up to the demand, we are working hard to engage our members to hands-on code: We need to get a lot of our technical deliverables out the door. That's where, as in any open-source community, we rely … on the community to get the work done. RISC-V is not like some of the other hardware plays — we don't have a development team. We're more like open source in that community sense, where we rely on the community to all contribute together to refine, ratify and vote to approve these extensions, software tools and other resources. That's something that we are working hard on. We have 43 workgroups running today on various pieces of that puzzle, as well as numerous highly invested members who are already building, already going to market, and they're leveraging the tools and resources that they can.
When you've got so many different groups working on this at the same time, is it difficult to keep everybody aligned and working toward the same thing?
It can be, if you don't have a good governance structure in place. But that's one of the things that we did earlier this year: We refined how we govern ourselves. We don't want that governing body to be [heavily] weighted by any one particular voice or member. So we've set up a structure of authority to approve and move things forward, as well as set up intentional interconnects between the various work groups that have codependencies upon one another. So we do work very much in concert.
I've also brought in a CTO, Mark Himelstein, to join us and to help guide some of that thinking. And so we're bringing in party-neutral leadership by having him on staff at RISC-V, as well as instituting a very clean, very fair and equitable governance structure where we are able to progress the state of the art across all interests at once.