The semiconductor industry has been a crucial part of the modern world, with applications from computing and communication to healthcare and transportation. Semiconductors, also known as microchips, are electronic devices made from materials such as silicon using microscopic switches called transistors that are used to control the flow of electricity in various types of electronic equipment.

The semiconductor industry is comprised of IDMs (integrated Device Manufacturers), fabless semiconductor companies that design and engineer the chips, and the so-called fabs that fabricate the chips. In total, it is responsible for designing, manufacturing, and selling these microchips, which are the building blocks of many electronic devices we use daily. The industry is constantly evolving, with innovations and technologies being developed all the time, and it plays a vital role in driving the global economy.

Abacus Semiconductor Corporation is one such organization involved in developing several processors and memory subsystems that are intended to reduce complexity and cost while maintaining a performance lead over traditional solutions. Axel Kloth, Founder, President, and CEO of Abacus Semiconductor Corporation (ASC), was with the CEOViews team to share the story of their entrepreneurial journey.

Inspiration and mission

Axel has been not only involved in developing components that make up supercomputers, but he has been using them for a few decades now. He felt that the user experience did not always mirror the theoretical peak performance of these systems. As such, he had incentives to improve the real-world performance of these systems in the hands of real-world users running real-world applications. He wanted to ensure that the functional performance of supercomputers running applications that typical users deploy matches the theoretical peak performance benchmarks indicated as closely as possible.

He then set out to solve the scale-out problem, and in the beginning, only a small number of scientists and users confirmed their findings. Over time, that number has grown substantially, encouraging him to continue. Engineers and marketing as well as sales at ASC received excellent feedback on where the problems of current systems were and what users looked forward to solving in next-generation designs. In 2022, they saw that more and more articles were being published that made the lack of scale-out public, and an ever-increasing number of users and scientists confirmed that they were on the right path. The Department of Energy also invited them to partake in finding new ways to improve supercomputer performance with a focus on integrating accelerators into the system in a useful and efficient manner.

Role of innovation

A novel idea often requires a novel toolset or some other fresh thinking around it to make it work and implement it. Luckily, they early on decided to use a new processor Instruction Set Architecture (ISA) – RISC-V – as the programmable element in their technology, and with it came a novel design tool. As a result, they are implementing a new kind of processor with a new kind of toolset to build a novel system architecture for accelerated supercomputers that overcome the prior scale-out limitations.

Most often, we as humans go down the same path despite the fact that circumstances have changed, and those changed circumstances would require a different set of actions. Overcoming this momentum is paramount to making novel ideas work and ensuring a successful implementation. As a result, everyone at ASC uses new technology internally, including design tools. Particularly engineering still uses the established tools for design verification and other tasks, such as those that are not yet covered by novel tools, but they monitor and are open to any changes.

Innovation is also a management challenge, and Axel feels there is a need to strike a delicate balance between being focused on getting things done and pure experimentation. Just focusing on getting things done will revert the company to old thinking. However, experimentation alone will not produce a product on time. So one has to find out where they need to stir up things and change and where they have to follow normal procedures.

Providing effective solutions

ASC provides the processors, I/O frontends, memories, and accelerators that make up the supercomputers of the future, and they work with all interested parties that provide operating systems, compilers, and applications so that their common customers have a solution that fulfils their computational needs. Adding more to it, Axel said, “We listen a lot, and we take feedback seriously. We also work with others, even if they can be considered competitors. If company A has an existing product that customers love, and that product could be even better if it were integrated with our technology, we do not hesitate to license out whatever it takes to make the customer even happier with a solution that is even better than what they have today. In most cases, our customers will not be the end users of our technology. As a result, we listen to our customers and our end customers and end users”.

THE LARGEST TRENDS IN THIS DOMAIN ARE THE USE OF ACCELERATORS IN HPC, PREMIERED BY NVIDIA WITH ITS CUDA, AND THE NOTION THAT QUANTUM COMPUTING WILL BE ABLE TO SOLVE A NEW CLASS OF PROBLEMS ~25 YEARS FROM NOW.

 Staying ahead of peers

Competition in the semiconductor industry has always been fierce, and it is very important to stay ahead of the competition. As a startup, ASC is very agile and nimble and must react quickly. Further adding to this, Axel said, “If we see a $100M opportunity, that’s huge for us. We will jump on it. If a large semiconductor company with an annual revenue of $70B sees a $100M opportunity, it will pass. Why? We can focus 100% on that opportunity. It takes 700 projects with $100M each to get to $70B annual revenue. That’s 700 project managers on a good day. They simply can’t and won’t take on another PM for this. On the other hand, we are 100% focused on that opportunity. It is just a numbers game”.

Also, Axel feels that funds and employees are crucial to staying ahead of the competition. It is hard to find excellent people, and once you have them on board, you must make sure they never get bored. If they get bored and stop learning, they’ll leave, and you’ll need to find another employee, get them trained, and keep them busy. That costs time, money, and energy.

Future technology

Semiconductors are considered crucial for the days ahead. The largest trends in this domain are the use of accelerators in HPC, premiered by NVIDIA with its CUDA, and the notion that quantum computing will be able to solve a new class of problems ~25 years from now. In other words, everyone is sandwiched between the technology that is available today and inadequate and the quantum computing revolution beginning in 2040 to 2050 for commercial deployments. In between, there are lots of challenges one has to solve. With its able team and decades of experience, ASC is set on a mission to solve these and become a leader in this space.