Last week, in writing about the competition between x86, Arm and RISC V, we said that RISC V would win the market for “IoT” because it is a greenfield market. In the comments, none other than luminary software expert Tim Bray weighed in – pointing out that IoT is not new by any measure. Mr. Bray is one of the most gracious people we have ever interacted with on the Internet, full of knowledge and a great writer. But if you are going to disagree with him, you had better come with more than a half-baked, poorly worded argument. So here, we want to clarify our point on IoT.
To be clear, he is entirely correct in that the Internet of Things (IoT) is an idea that has been around for a very long time. We think Cisco coined the term about 15 years ago, but that was merely a rebranding of what we used to call Machine to Machine (M2M) communications, which itself was just a new name for what we used to call “Telematics”. So much of the discussion about IoT (and M2M) is fluffy marketing speak. Cisco and Ericsson fluffed these terms up considerably – putting out estimates of hundreds of billions of connected devices by some future arbitrary date. This was great marketing work on their part, which we imagine helped them sell a fair chunk of gear.
But there really is no “Internet of Things”. There is no set of devices that operates on its own special Internet, as Bray pointed out its all just TCP, HTTP and IP all the way down. Moreover, all those connected devices are entirely heterogenous spread across 50,000 uses cases over 30 industries. Like we said back in 2018 – IoT is a Mess.
That being said, the number of connected devices is growing rapidly. By most estimates, the amount of data passed between machines with no human interaction has surpassed the amount of traffic in which humans are involved. We believe firmly that this process is going to accelerate further. The cost of connecting things continues to fall. You can buy Bluetooth radios that cost less than $1 today, and will at some point likely reach the point of disposability – a Bluetooth sticker. We are also seeing a growing number of things come standard with connectivity – power pools, cabling, espresso makers, and crucially here for our purposes industrial machinery. Manufacturers have been building some form of connectivity into machinery for a long time, but the process is advancing just as more machines (aka robots) seem to be enjoying new prominence on factory and warehouse floors. So that part of the story holds up – there are going to be a lot of things connected to the Internet, measured not in billions but trillions.
And while there is never going to be a single operating system or software stack that runs on top of those, all of these machines will need chips to power them. Today, most things classified as IoT devices need little in the way of digital logic – a microcontroller is more than enough, but the cost of putting real compute in those things seems close to reaching a tipping point. True, there is no point in putting a CPU into most connectivity modules for industrial sensors and machines, but adding a smaller processor inside some sort of system on a chip (SoC) that handles multiple functions will become so cheap as to merit a “sure, why not” response. (Of course, this brings all sorts of other problems, but we will save the topic of security for another day.) So there is going to be considerable growth in these areas, into products which previously never would have needed serious semi content, and that, to us, is a greenfield opportunity.
Another important, “greenfield” aspect to all this, is that many of these future connected devices will be built by new Chinese manufacturers. And so the question becomes – who is going to build all those chips? In the US, this will likely be the usual suspects – Qualcomm and Broadcom in the home, Texas Instruments and ADI for industrial. And do not sleep on Silicon Labs. All of these are Arm licensees, but many have RISC V projects in the works. But the even bigger opportunity will come from Chinese chip vendors – who are plugging away at industrial and analog markets, with a new chip company getting funded every day. From Arm and RISC V’s perspective this is greenfield opportunity, and it is going to swing hard towards RISC V.