# Anabrid: Shaping the Future of Analog Computing

Imagine facing a complicated computational problem, for instance, in engineering or chemistry. You could turn this problem into equations and derive instructions for your digital computers to run the numerics and compute a solution.

Or, you could build a physical model, typically by using electronic components governed by the same equations as your original computational problem, and read out the solution by observing the dynamics of the physical model. Such a physical model is called an analog computer.

Using continuous physical parameters, such as the current and voltage of an electric circuit, analog computers can solve computational problems much faster and more energy-efficiently than their digital counterparts. They were developed way before digital computers but became obsolete in the 1960s as the computing power of digital computers increased exponentially—what is known as Moore’s law. However, there might be a resurgence of analog computers owing to modern fabrication techniques that allow for miniaturization.

Founded by Bernd Ulmann, Lars Heimann, and Sven Köppel in the fall of 2020, Anabrid develops analog computers, aiming to fit them on a chip and thereby make analog computing ubiquitous.

Learn more about the future of analog computing from our interview with two of the co-founders of Anabrid, Lars Heimann and Sven Köppel:

## Why Did You start Anabrid?

As engineers and scientists, we saw the huge potential of analog computing: This could be the next big thing! We could see all the potential uses, from navigating an airplane to solving computational problems in chemistry, where an analog computer could really help solve important problems. Also, it’s great to work in a small team with great autonomy and do things no one has ever done. Very few companies currently explore analog computing, so we could be pioneers!

## How Does Analog Computing Work?

There are no algorithms, if/else branches, digits, or symbols in analog computing. It’s more like programming a dataflow machine. And speaking of flows, that is actually a great analogy! Imagine two rivers that merge into one: When their water flows combine, they add up their individual water flows! Basically, they are performing a summation operation.

Similarly, an analog computer solves a problem by implementing, for instance, an electrical circuit that models that problem. Just like water flowing in a river, applying a voltage makes electrons flow through a circuit, and by observing the resulting electric currents, one can read out the solution to the problem.

Analog computers are good at solving problems typically described by a differential equation. Instead of solving that differential equation using numbers and numerics, an analog computer is like an electronic circuit that obeys the same differential equation—and observing the circuit’s electric currents yields the solution to the differential equation.

The speed of an analog computer is set by a certain frequency cutoff that prevents damping, effectively acting as a low-pass filter. This means an analog computer could be about four to five orders of magnitudes faster and about a thousand times more energy efficient than a classical digital computer at solving certain computational problems. And it cannot be hacked!

We can implement basically any differential equation that people learn about during their studies—more complex equations we can transform into simpler differential equations. The more complex a problem and its differential equation, the larger the analog computer needed to implement it. That is, for every new computational problem, you need to rewire your analog computer. But this will be straightforward once we build a reconfigurable analog chip. And while digital computers need more execution time to solve more complex computational problems, an analog computer gets bigger and takes the same amount of time to run.

In practice, an analog computer could act as a co-processor for a digital computer, tackling the problems it is most suited for. Ideally, the users would not even notice that they’re using an analog computer.

## How Did You Evaluate Your Startup Idea?

We started simply by building and selling analog computers, box-sized machines purchased by universities and research centers and industry customers for all kinds of purposes. Since day one, customers have contacted us and told us that they would love to have an analog computer: someone even wanted to have an analog computer to simulate mining on the moon!

This way, and by selling small-scale analog computers for educational purposes through our own online shop, which has attracted great interest, especially in the U.S., we have already got quite some revenue. Venture capitalists tell you to focus on a single problem first. Yet, our goal, driven by customer demand, is to build a general-purpose analog chip that could be integrated into practically every device, from your mobile phone to your dishwasher.

## What Advice Would You Give Fellow Deep Tech Founders?

Raising venture capital is a lot about marketing and making your company look sexy. Engineers, especially in Germany, often only care about building the best possible product, but they should take inspiration from their U.S. counterparts and think about how they will sell the product. Yes, doing it professionally costs money, and the return on investment is not straightforward to measure. Still, it can help make customers buy your product and venture capitalists your company.