Arm is the most important chip company that doesn't make any chips. UK-based Arm produces low-power, high performance chip architectures that it licenses to other vendors. Customers include Apple, Samsung, and Qualcomm so Arm chips are found in most the mobile phones we use in the US. And the company is just getting started.
At CES 2019, I spoke with CEO Simon Segars for the latest episode of Fast Forward. He discussed the company's unique approach, its role in the move to 5G, and how secure the Internet of Things really is.
Dan Costa: So let's talk a little bit about Arm. PCMag users are going to be familiar with Arm, but for those who are not, talk a little bit about its unique business model.
Simon Segars: We're a chip company that doesn't make chips and we're often described as a semiconductor company, but in reality we serve the semiconductor industry. We design microprocessors. We design the compute and the intelligence that goes into so many of the products that you would see here at CES. Our business model from day one was all about 'let's not build chips, let's design IP.' Let's license that to semiconductor companies so that they can bring all their expertise, wrap that around a microprocessor, and increase the intelligence in embedded devices.
Dan Costa: When you say 'license to vendors,' it's big name vendors. You're licensing to Apple, you're licensing the Qualcomm, you're licensing to a variety of big chip players.
Simon Segars: Really pretty much every semiconductor company you can think of is using Arm technology in some way. As I say, they're using a microprocessor as a building block to bring all their own IP, bring all their own secret sauce to bear, in increasingly the intelligence in these chips.
Dan Costa: There's a variety of different trends at CES. Self-driving cars are one of them. And Arm has a role to play in that space too, right?
Simon Segars: As we're thinking about the future of automotive, clearly the semiconductor content of cars is going up and up and up. To have a self-driving car that can deal with every set of circumstances it's ever going to be in, that is a massively hard computational problem. You've got to take data off of cameras, data off of sensors, and you've got to crunch that down and instantaneously work out what the right thing to do is. That's a really hard problem. So people who are developing the algorithms for that are driving cars around which basically got a rack of servers in the back of it.
Now that's great for prototypes, but if you want to build millions of cars, you've got to turn that into a cost-effective and power-effective solution. So as we're developing our roadmap, we're trying to bring the right compute power, the right features in our processors, but deliver that in a way that enables this car, which is a safety-critical system, to be relied upon to be a safe environment in which you can sit and become a passenger, not a driver. So for that into our processors we're building in features that allow redundancy, features that allow safety-critical software to be deployed in the right way. So it's actually a really complex problem.
Dan Costa: And you're both using AI to help develop these chips but also build applications that can support other AI processes, right?
Simon Segars: AI and deep learning over the last couple of years has made huge strides and a lot of that kind of originated with complex algorithms running in the cloud where you've got effectively unlimited computing. Again, if you want to bring that to the edge, where the action is, then you've got to deliver it in a very cost-effective way. So to do that, we're building accelerators that can take the kind of maths that you run to do neural networks and implement those algorithms in a very cost- and power-efficient way in a tiny chip that might cost a dollar.
Dan Costa: I think it's a really interesting point that there are things you can do in software in the cloud with virtually unlimited compute that we would want to be able to do on the edge, in your car, in real time, at low power. And to do that, it really needs to be in silicon, and you're designing those chips for those solutions.
Simon Segars: Yeah, you can. I mean, the great thing about the growth of cloud is effectively you've got unlimited computing. So computer programming languages have evolved over the years to support very high levels of abstraction to allow you to code these algorithms. You can, in a few lines of Python, describe a neural network which, when I go back to, you know, I learnt in Assembler and in very primitive languages, the productivity that that's given is phenomenal. That's okay, if you're running in the cloud, but, again, if you want to put that in a cost- and power-efficient way, you've got to think about what are the underlying algorithms, what are the underlying maths that you have to do? And to do neural nets, you end up doing a lot of matrix multiplication. So to build an accelerator that's really good at that, to complement a general purpose microprocessor in one of these highly integrated chips. That's the way in which you deliver it in a scalable way.
Dan Costa: I think we have lot of the same challenges with 5G. We tend to think of it as wireless network that's going to make our data connections faster. We have to build out the network and the towers, but there's a lot of processing that needs to happen on the edge, too.
Simon Segars: In the handset itself, if you look at the complexity of the computer that is the modem, right? That deals with the protocols and the air interface. That's grown exponentially over the different generations of technology. 5G is no different. It's really complex, so you need a much more powerful computer to process that interface to the network. So again, we've been thinking about this for a long time, anticipating the rollout of 5G and developing our processors to enable the processing that a 5G modem needs.
Dan Costa: What do you think that 5G is going to mean in terms of beyond the handset, in terms of the environment that we live, IoT devices, and the connectivity of cities. What are some of the 5G applications you see in that space?
Simon Segars: When a lot of people are going to think about 5G, they're gonna think, oh, okay, we went from three to four and now this is five. It's the same again, only a little bit better. 5G is much more radically different than that because 5G does enable so many other things. It's not just about better handsets. Better handsets are cool, don't get me wrong.
Dan Costa: We'll take them.
Simon Segars: Yeah, I'll take it all day long, but it is those other areas. Two, three, 4G were basically designed around how many people are there carrying handsets? So when you think about the world of IoT, where you want to connect now billions more things. Old networks just don't support that. 5G is designed, you know, one of the design aspects of 5G is to support all those other connections and that allows you to drive cars around which can hand off between networks. It allows you to deploy IoT devices over cellular connectivity, which has got a huge number of benefits to it. The network's going to allow that, so it's going to allow for smart cities. I mean when we think about self-driving cars, actually the cars interacting with traffic lights and the infrastructure is an important thing. You want every lamp post to have a cellular connection and to be able to talk to cars as they're driving along. That's a lot more connected things. 5G is going to allow that.
Dan Costa: What is 5G going to enable that excites you personally? What are you looking forward to?
Simon Segars: This overall world of IoT, where data is being sensed, connected into the network, and then processed and turned into patterns. That is going to enable a whole load of efficiencies that we just don't have today and that frankly the world is going to rely on as population is continuing to grow, maybe not here in the US, but in other parts of the world where urbanization is a big trend. You know, you're dealing with megacities with a lot of people in them, managing the flow of people, managing everybody living on top of each other. That's going to require a lot more data and a lot more intelligence, and 5G with this ease of connectivity of all these things is going to allow that.
Dan Costa: A lot of people overlook embedded processing and how important it is to enabling a whole suite of applications. Arm is pretty key in the embedded processing market. What are some of the things you think consumers should appreciate more in the embedded space?
Simon Segars: Yeah, I mean, from a consumer point of view, I'm kind of torn because I'm really proud of what we do. I love it and I come to a show like this and you see it and people often ask me 'where's Arm's booth?' and I go, well, we're sort of everywhere.
Dan Costa: Inside of every booth, inside every device.
Simon Segars: On the other hand, from a consumer point of view, consumers shouldn't have to worry about what the technology is or how it works. They should just get the benefit from it. So I kind of like the fact and I talk about our technology being invisible and that's quite good.
And some people go, 'whoa, hang on a minute. You want to be invisible?' Don't get me wrong. It's the technology. As a consumer, you shouldn't have to be reading the manual. It should be intuitive, it should just do stuff for you. But, you know, what's driven that has been decades of progress in making computers smaller and cheaper and faster and lower power so that they can just be everywhere and everywhere is where we're getting to. IoT will take everywhere even further.
Dan Costa: I want to be respectful of your time. It's a busy show. I ask everybody these three questions. Is there a technology trend that concerns you? Something that keeps you up at night?
Simon Segars: When I think about all of these connected things, you have to worry about security. And for us security is a big part of how we're driving the company. It's a big thing for us. All these network-connected things are a potential way in for a hacker into the network. So we're working really hard to try and improve the underlying security features of our products and the building blocks that our partners have to put these chips together, but it's a never-ending challenge There is always new ways people are thinking to get around the security protections that people are putting in place. So that's just something that we just have to keep working on every day of the week.
Dan Costa: I think one of the problems with the security space is that it's hard to hold companies accountable for breaches and security flaws and as a result, you know, there can be no consequences. There are some companies that focus a lot on security, spend a lot of money on it. There are others that cut corners and I don't think there's any market consequences to a lot of that many times.
Simon Segars: Yes, and that is a really interesting challenge. I mean, the position that we're taking is, look, this isn't any one person's responsibility, it's everyone's and the tech industry as a whole. Whether it's us creating IP that's going to a chip company, the chip company, the OEM, the guy writing the software; everyone's got a role to play in making devices secure and manageable and updatable in the event of a security instant. The industry is going to come together. So we're trying to corral that, if you will. We're trying to, you know, we've put out a call to action amongst all of our partners to take this more seriously and to bring the industry together to really focus on this.
Dan Costa: Is there a technology or a service or a gadget that you use every day that inspires wonder?
Simon Segars: I drive a Tesla. I've got to say I get in that every day and I think this is really cool, and when there's a software update where I've suddenly got more features in it than I had before. One day we'll just take that for granted, but relative to every other car I've owned, that's pretty cool.
Dan Costa: You're the second person at CES that has told me that and with the same enthusiasm. Tesla is clearly is a different type of product experience.
Simon Segars: Absolutely. People talk about, you know, and say it's an iPad on wheels. There's a lot of kind of similarities between those kind of consumer products where it's a vehicle, pardon the pun, for software, and it's got a wheel in each corner and a steering wheel today. But it is this software platform that keeps getting upgraded.
Dan Costa: Very cool. So if people want to know more about Arm, follow what you're doing, how can they find you online?
Simon Segars: So arm.com is where people come to get information about what we're doing. We keep that live, we keep that updated with new announcements. My own Twitter handle, if people want to follow that is @SimonSegars.
For more Fast Forward with Dan Costa, subscribe to the podcast. On iOS, download Apple's Podcasts app, search for Fast Forward and subscribe. On Android, download the Stitcher Radio for Podcasts app via Google Play.
Get Our Best Stories!
This newsletter may contain advertising, deals, or affiliate links. Subscribing to a newsletter indicates your consent to our Terms of Use and Privacy Policy. You may unsubscribe from the newsletters at any time.
Thanks for signing up!
Your subscription has been confirmed. Keep an eye on your inbox!
Sign up for other newsletters