Chip design is becoming a viable tool to solve problems, even for companies who are not chip design specialists. But for us to profit from these changes, we have a lot of hard work to do. If you are interested in why we believe in democratized chip design and how we want to be part of it, well, this is our secret plan:
What Is Going On in the Semiconductor Business?
Last year’s chip shortage taught many companies that they should invest more in this part of their supply chain and get more independence and control over the chips they use. But there are also many positive industry trends that make this a good idea:
Design vs. Fabrication
Getting involved in chip design doesn’t mean you have to manufacture them yourself – and vice versa. For example, Apple and AMD currently have the best designs for processors, but neither of them have factories. TSMC does not design processors, but they have the best factories. AMD and Apple teamed up with TSMC and now they have the best products on the market. Factories are incredibly expensive to build and this used to be a big barrier for entry into the chip market. Nowadays, you can design new chips and have them made by a suitable factory, without having to invest billions upfront. In an ideal future, designing custom chips and ordering them from a manufacturer will become as easy as the process is for PCBs today.
Better Design
In the early years, improvements in manufacturing (more and smaller transistors) were providing better performance faster than the designers could optimize their chips. This is not the case anymore, so optimizing the designs becomes more important. This also means that chips stay relevant for a longer time, giving the developer more time to earn money from it, which in turn means that serving smaller niches becomes economically viable.
FPGAs (Field Programmable Gate Arrays)
FPGAs are special chips that can be configured to implement a digital circuit. They provide an intermediate, much easier step before actually manufacturing custom chips, while still provide many of the benefits. FPGAs are not new, but they have become much more approachable in the recent years.
RISC-V
RISC-V is an open standard for making processors. It is similar to ARM (e.g. aarch64), but since it is open, similar to open-source software, the freedom to use it and modify it in whichever way you want, opens up lots of possibilities.
Why Do We Want to Get Into the Chip Game?
First of all: it’s fun. But also business. We outline some of our thoughts here, which inform our motivation, but each and every company out there has to evaluate this for themselves.
Making Custom Chips Will Be Like Making Custom Pcbs in the Future
It might not be as easy, but it will get a lot closer. The push for smaller nodes on the top end will free up more production capacity on the “older” nodes, which are still extremely beneficial for specialized chips. For example, 22nm is old news for a smartphone processor, but is perfect for a low-power micro-controller with a specialized AI accelerator.
The current chip shortage also spurs a lot investment in production, especially for the top end of the market. Building these factories takes a while, but once they are online, the world’s silicon production capabilities will have grown quite a bit. We think that in the near future, everyone with a good design for any of the older production nodes will be able to cheaply produce it.
Value Added to Counter Value Lost
The big car manufacturers sell about 2 (BMW, Mercedes) to 5 (VW, Toyota) million cars per year. We guesstimate that each car contains about 100 micro-controllers. Apple sells about 200 million iPhones per year in comparison. If a car-maker standardized the controllers they use in a car (use the same chip over and over again instead of different ones), they would reach 200 to 500 million chips, using maybe 2 or 3 different designs. This translates to a revenue of roughly 1 to 2 billion euros per year, the profits of which would not need to be shared with a supplier, but only the chip manufacturer.
For traditional automakers the move to electric vehicles means that they lose the value-add of producing engines and transmissions. Capturing the value add of making batteries – and maybe chips – will make the differences between being a stylist for cars or an industrial power house.
Owning the Stack Means Freedom to Innovate
Chips and software are the beating heart of the economy. Someone who is dependent on chips and software designed by others will be in a weaker position to innovate.
Apple recently reached the culmination of a more than 10 year long process that lead to them making their own processors. These chips are now the best in the market by most metrics, and the products Apple builds around them therefore also beat the competition. But here is the problem: Apple’s awesome hardware does not help anyone else. Nintendo could build an incredible new version of the Switch using Apple Silicon, except they can’t, because Apple won’t sell them the chips. You can’t build innovative new customer experiences if you are the last in line to get the parts to build it.
What Is Motius Doing to Get In?
Now that we know that we want to get into the market – what is our plan of action? We are starting an internal project to evaluate our ideas in this space, focusing especially on RISC-V and the opportunities it provides. We will try to answer these questions:
- Where does Motius fit into this ecosystem?
- How can we provide value?
- How can we turn it into a business?
This internal project will be run and treated the same way we run and treat our projects with customers. It will be staffed with a small team, but we want everyone at Motius to be part of the bigger family. In fact, we are publishing this very article to the public, because we also want our existing and potential customers to be part of this. And for the external world, we think this will provide valuable insights into how Motius thinks about new topics like this and how we approach learning new technologies.
To bring everyone at Motius along for the ride, our Sprint Reviews will be called “Motius Silicon Seminar” and be public to the whole company. Why?
We want everyone in the company to be informed, engaged, and enthusiastic. By doing this and by revealing our intentions to the public we want to spread the idea and find suitable projects to apply our work.
Finding suitable projects is the key to success for everything we do.
This allows us to also practice how to communicate about this topic. If you read this article and are slightly confused – we understand. It’s a complex and abstract topic and it is not easy to explain and talk about. But the more we talk about it, the better we will get at talking about it.
This is our Tech Roadmap
- Research different open-source cores and write a report about the ecosystem.
- Research the tools in the space and write a report about the ecosystem.
- Pick a 32bit core, a set of tools, and all parts needed for a microcontroller design and put it on an FPGA.
- Extend that design with a custom hardware accelerator for a highly specialized use-case.
- Review, retrospect, and design next steps.
Following This Roadmap Will Have Four Outcomes:
- We learned a lot about the RISC-V ecosystem
- We learned a lot about FPGA programming, which will open up new opportunities for Motius independently from the success of RISC-V (even if we won’t end up making our own chips, we can do FPGA project work in the future)
- We have a demonstrator showing one of the key arguments of RISC-V: custom “silicon” paired with standard modules
- We will have a clear picture of how Motius can provide value to the RISC-V ecosystem
This is Our Business Roadmap
- Find a market for FPGA development services
- Learn about the IP block market
- Figure out who would be interested in doing custom silicon with us
- Find the weak-spots in the process – which tools, services, consulting, IP-blocks could we provide to someone who wants to create custom silicon?
We are very excited to start this new endeavor and are curious on where it will lead us. If you are curious as well, get in touch with us.
