Codasip: Shaping the Future of RISC-V Processors
It’s like magic when microprocessors convert lines of code into movements of electrons, leading transistors to change their state and controlling everything from smartphones to satellites.
This conversion is orchestrated by so-called instruction set architectures. While proprietary instruction set architectures have been used by semiconductor giants for a long time, a new and open instruction set architecture called RISC-V is poised to level the playing field, allowing also smaller companies to design chips more flexibly and innovatively.
Codasip delivers IP and design tools to help chip designers build actual microprocessors based on RISC-V and automatically optimize their processor designs. Founded by Karel Masarik in 2014, the company secured several funding rounds and became a leader in custom compute.
Learn more about the future of RISC-V processors from our interview with the founder and president, Karel Masarik:
Why Did You Start Codasip?
When I was in my twenties, I moved from the Czech Republic to Germany for an internship in the automotive industry as part of the European Leonardo da Vinci program. That way, I got my first hands-on experience with embedded systems—specifically, microprocessors that take care of specific tasks within a larger system, such as a car.
It led me to pursue a Master’s in embedded systems at the Fernuniversität Hagen and then work for some time in the industry before doing a computer science PhD at the University of Brno. While working in academia, I secured several grants for microprocessor design EDA tool projects with academic and industrial partners.
In 2013, we had several industrial customers for this tool, and with my academic supervisor, we decided to create a spin-out. He was in contact with Credo Ventures, a Czech venture capital fund, and following a first round of funding, we established Codasip as a startup in 2014 to develop and boost the commercialization of the EDA tool and processor IP.
How Do You Help Design RISC-V Microprocessors?
RISC-V is an independent instruction set architecture to build microprocessors. An instruction set architecture is a vital part of computer architecture that defines, as the name says, which instructions a microprocessor can execute. For example, it defines whether a computer can perform multiplication directly or has to resort to repeated additions. It’s the interface between the computer’s hardware and software, dictating how software controls the hardware to perform tasks.
There are two major instruction set architectures, RISC-V and Arm, that are based on the RISC (Reduced Instruction Set Computing) principles, emphasizing simplicity and efficiency in the instruction set design.
Arm is proprietary, licensed by Arm Holdings, and popular in mobile devices, embedded systems, and increasingly in servers and desktop computing. It is well-established and used by many companies, so if you want to differentiate, you need to adapt it to your needs. You can be successful that way, as Apple has demonstrated with the M1, M2, and, lately, the M3 chips series. But it requires large teams and a lot of overhead.
With RISC-V, we want to make it easier for people to innovate. RISC-V is an open standard instruction set architecture that originated from the University of California, Berkeley, and is maintained by the non-profit RISC-V International. Unlike Arm, RISC-V’s open and royalty-free licensing model allows anyone to design, manufacture, and sell RISC-V chips without paying licensing fees.
Using an open instruction set architecture allows different vendors to work together and ensures you have the freedom to choose with whom you’re partnering. You don’t want a standard to come from a single supplier, as you would then be forced to buy from them. That’s the issue with Arm—you get locked in. Being open also allows for innovation—if you have a good idea to improve a certain aspect, you can do so and change the instruction set architecture straightforwardly.
An instruction set architecture is like defining the components you have for building a house—walls, doors, toilets, whatever you may need. Yet, you still need to arrange those in a way that will allow you to build a stable and nice house, making sure the walls support each other and that the toilet remains accessible. You can start with blueprints that worked well in the past for other houses, but you still need to adapt them for your own design.
It’s very much the same for microprocessor design. We’re providing both IP, i.e., blueprints for building a house, and software tools for adapting your designs, i.e., helping you to move a wall or add more toilets and ensuring the house remains stable. Our software is like an ‘AutoCAD for microprocessor design,’ way more flexible than pen and paper, and we have built an entire ecosystem around it, ensuring the processor designs are secure, performant, and can address concrete industry applications.
How Did You Evaluate Your Startup Idea?
Soon after founding Codasip, we got customers beyond the Czech Republic in Germany, Japan, and the US. Working with real, paying customers allowed us to get immediate feedback, develop our software further, and attract more investors. We initially focussed a lot on software for microprocessor design; in 2018, we flipped over to Germany, making it easier for us to raise capital, expand internationally, and focus more on the design of microprocessor IP.
We are a founding member of RISC-V International and have established a design center in the UK, France, Greece, and Spain, where we attracted a lot of semiconductor talent, among others, from Arm. We had grown fast and learned a lot when others realized the opportunity around RISC-V and founded competitors.
In 2021, Ron Black became the new CEO of Codasip as we were talking about future expansions of Codasip with investors. He has more experience than me with leading large companies with hundreds of employees, and I transitioned to the president role, focusing on building new cutting-edge technologies and working with the European Commission and national governments to ensure we have the right environment to be innovative with balanced risks.
Triggered by COVID-19, we had a huge chip shortage, and this crisis showed that our semiconductor supply chains were too fragile. We need to prop up our own semiconductor industry, which is tricky when Chinese companies get a lot of free money. Yet, the EU Chips Act provided certainty and helped to mitigate the risks around investments in hardware, matching VC funding with money from the European Commission. This not only helps them grow faster but also allows them to undertake riskier projects.
What Advice Would You Give Fellow Deep Tech Founders?
Nothing in startup land is stable. Treat your beliefs as assumptions that you need to validate with customers; otherwise, they’re just dreams. You have to validate your technology regularly—there’s a reason there are no dinosaurs anymore; only those who adapt survive.
Having a technology doesn’t mean you have a product—there is a big leap in between, and you need a product that people pay for to survive as a company. Especially in cash-intensive domains like the semiconductor industry, it takes a couple of years to build a reputation and convince investors and customers of your propositions.