SPHERICAL: Shaping the Future of Chip Design for Space
With satellite launches more frequent than ever before, we are witnessing the dawn of the space economy.
The proliferation of satellites is a testament to our technological progress, but it brings with it a critical challenge: the need for custom-designed chips that can endure the unforgiving conditions of outer space. Yet, chip design is stuck with decade-old design paradigms and software tools using principles that are just as old, which is holding back innovation and the progress of space systems.
Enter SPHERICAL. The startup is reimagining space systems from the very core, the semiconductor level, delivering chips that are smaller, more reliable, and incredibly adaptable through software configuration. Founded by Thomas Parry in the fall of 2022 and joined by co-founder Bastiaan Bom, they have gone through the Antler startup program and gotten into the ESA BIC startup incubator in Noordwijk, Netherlands.
Learn more about the future of chip design for space from our interview with the co-founder and CEO, Thomas Parry:
Why Did You Start SPHERICAL?
Before SPHERICAL, I worked at Clyde Space designing hardware used by several small satellite startups and national space agencies and control electronics that had to be small, reliable, and still flexibly configurable. I then moved to the semiconductor design industry, where I learned about the needs of different customers and saw how semiconductor approaches were quickly changing—going from off-the-shelf microprocessors increasingly to chips specialized for specific purposes—and receiving increasing national support through, for instance, the European Chips Act or open-source silicon.
That’s when everything came full circuit, and I decided to found a startup to solve the problems I had experienced for years around chip design specifically for space applications. The semiconductor industry is stuck with electronic design tools using old methods and approaches. I thought there could be a new and better way to design specialized chips.
How Does Chip Design Work?
For a company like Apple to produce top-notch phones, it has to design chips that are powerful, compact, and lightweight. Our objective is similar: to make satellite systems more performant with better chips that are perfect for the system’s needs.
Chip design used to work something like this: first is the customer’s need, then an engineer figures out how to fulfill that need, what chip IP can be off-the-shelf or have to be custom-built, and then those circuits are designed and a physical layout made. However, finding the best design and layout is a major challenge because it requires expensive electronic design automation (EDA) tools, tools that were themselves developed in the late 1980s and also look like that.
Not only are they user-unfriendly, but their chip designs are also very static and fragile. Once you find a solution, you don’t want to go back and change even a single thing unless you want to start over again and find a new solution.
There’s an opportunity to make chip design more dynamic and generative. Driven by the high-level input of the system designer, a machine learning model can take care of generating low-level functions. Chips have many sub-functions, like analog-to-digital conversions or operational amplifiers, to increase signal strength. Instead of designing an operational amplifier by hand, one could run a machine learning simulation overnight and get a template the next morning.
Another important point to consider, especially for space applications, is protection against radiation damage. Radiation can damage chips in two ways. One, by its total ionizing dose, which accumulates charge in insulators within the chips over time. This leads to charges appearing in places you don’t want them, and transistors behaving strangely, leaking, and current paths deviating from the chip’s intended layout. The other way is through single-event strikes of high-energy particles that also insert charges in unwanted places, such as flipping a zero to a one in your memory or shorting circuits. So, you need your circuits to be resilient to such impacts and use memory-correcting codes to detect unwanted alterations.
Our first product will be to design the best possible satellite power system. We’ll have the first chips manufactured in a few months, and over the next few years, we’ll demonstrate our first product for satellite power management and expand from there to enable other functionalities, like communications. Besides space systems, we will also be able to address other use cases for different industries. It is not for nothing that large companies develop chips in-house, as this is an important part of the value chain.
How Did You Evaluate Your Startup Idea?
First, I was solving my own problems, so I knew about it fairly well. I also did some bottom-up market sizing—considering every announced satellite for the next ten years. Once I was convinced that there could be a sizable market, I talked to many potential customers, understanding their needs and finding potential partnerships. Ultimately, it’s a small industry; everyone knows everyone through conferences.
The space industry has started to launch more CubeSats, tiny satellites expected to last only three to five years. But as the industry grows, more satellites will become larger again and go beyond low Earth orbits into medium Earth orbits and geostationary orbits, where radiation is more of a concern.
We had to deeply understand customers’ needs for all these different use cases and demonstrate that our tools can provide more value than existing solutions. Our initial focus is on customers who require greater radiation reliability as they target higher orbits. There, we have a competitive advantage in designing chips.
What Advice Would You Give Fellow Deep Tech Founders?
Start building a team sooner than you think you should. I started all by myself, but now that I have found the right co-founder and received some initial funding, we can now make much more progress. Finding a co-founder is really about speaking to as many people as possible. Start filling your pipeline with people excited about your mission, and then talk to them!
I knew Bastiaan through the local Dutch space ecosystem, where we met at one of the Space Netherlands networking events. We spent a lot of time talking about the industry, our mission, and what we’re going to build, and we formed a deep relationship over months.