The Exchanges

Q Well, yeah. Dirk, give me an example of that. So a power plant, how long does it take them to solve one of their big, what's the, what's the problem a power plant has and how long does that typically take them without you?

A Yeah. For example, in Germany, uh, all the energy is traded within windows of 15 minutes. So a big, Energy companies, they make every 15 minutes a decision. Should we buy energy? Should we produce energy? Do we get enough solar energy for the next 15 minutes or wind energy? And so they have a time window of these 15 minutes is five minutes data in, five minutes making the next, the right decision, and then five minutes data out. And if they could develop, if they could solve in these five minutes a more complex problem, that's a huge business impact for them, because then it's like a trading in finance, yeah? And so that's the benchmark. How much could we calculate in these five minutes? If we could do it faster, if we could calculate it in three minutes, then they could put more.

AI assessment note: “all the energy is traded within windows of 15 minutes.”

Q Okay. And how are you able to recruit these engineers, right? I imagine this requires like top, top tier talent.

A Yeah, we have a very strong network, for example, two of our founders, Sebastian, he's leading, or he's vice president of the SUSE Institute in Berlin, which is one of the top HPC Institute's centers here in Europe. Um, so he has also a professorship for optimization and machine learning, so we could attract through his network a lot of people. We have one of our co-founders, Sabina Jeschke. She's also professor, two professorships at TU Berlin and in Aachen. And she is a former, uh, board member of the Deutsche Bahn. That's a German railways here in Germany's top company belongs to the top 50 companies here. So through that network, we could, uh, hire a lot of, hire and attract a lot of people. And, um, yeah, myself, I'm in since 25 years in this business, uh, of optimization simulations and machine learning. And yeah, basically mainly through our network, we are able to get the people.

AI assessment note: “we have a very strong network, for example, two of our founders”

Q So your title is CEO, but just to be curious, I mean, you two are basically co-founders, right?

A Yes, exactly. So my career is I'm a computer scientist. I, um, started in, uh, technical positions, but then I moved pretty fast to sales, uh, positions, leading, uh, whole sales teams. And then I, uh, was the co-founder of a company, GOROBI GmbH in Germany. Uh, I was building up the whole operations, uh, here for the German market. They, they are the market leader for, um, optimization and software based, uh, that company got sold in 2017 to a private equity firm. Um, and I also founded a company doing sports scheduling. So, which is also related to optimization because you cannot imagine, but these problems are also really complicated in terms of how many solutions you could find and getting the best schedule out of it. Yeah. Like the NFL, they are running several thousand notes on Amazon for a few months just to get the best NFL schedule in the US.

AI assessment note: “Yes, exactly. So my career is I'm a computer scientist.”

Q So how do you do that? I mean, do you guys own a bunch of quantum computers? You rent out space or you're using other pieces, hardware and people's hardware and processing power.

A Yeah, so far we have access to quantum computers, but all these quantum computers are more in research state, because there are currently no real industry, um, usable quantum computers out there, but we are currently, so that's our hybrid, uh, classical compute, um, um, um, uh, way we are going, because currently we are able to use these, um, problems and put them on classical hardware, and this classical hardware could be CPUs or GPUs, FPGAs, So that these companies already get to speed up. And later on, when quantum computers are available, we can internally move these programs on a quantum computer. And basically the customer will not see anything, but he will get, get an immense speed up out of that.

AI assessment note: “we have access to quantum computers, but all these quantum computers are more in research state”

Q Ah, very interesting. Okay. That's very helpful. And then, okay. So next step is how do you think about pricing? Let's say you have successful LOIs, right? How do you decide what to price the thing at?

A Yeah. So that's what we're currently working on. So it depends. So because there are customers, they just run a problem twice a year because it's a strategic decision. Yeah. How do we build up our supply chain network? So that's a decision you make once every few years. Yeah. Or like if you do parcel delivery, a big company, for example, like UPS, so they make a decision for their network once a year. So then it might be more expensive to create this one, solve this one, uh, once in a time, Uh, solution instead of companies, like I said, the electricity power plant companies doing that every 15 minutes. So they might buy more, uh, subscription, which allows them to do, yeah, calculations all the time, for example. And, um, that's where we are currently working on to make a price model, which adapts to, to everybody. So we have benchmarks because from our previous experiences, how, how much we could charge for that. Um, and they, as I said, they are saving a lot of money if they get to the best solution, and they're easily willing to pay in the six digits, um, um, US dollar space per year, for example.

AI assessment note: “they're easily willing to pay in the six digits, um, um, US dollar space”

Q So that's hard, right? Because then you have to prove attribution, right? So that, that is basically custom pricing every single time. It's, it's very, very high touch. It's a, you know, you need to hire a bunch of AEs, et cetera.

A We will, we will have, uh, in the, in the future, some standard pricing, for example, if you, if you utilize our, um, service for, for the, like, for the whole time, you need that kind of speed up, you need that kind of, you have that kind of, uh, model size, then we could, uh, put a price tag for it. Uh, like I said, something like a hundred K a year, if you have a PowerPoint, a power, um, plan scheduling, for example, but, uh, Yeah. If you have just, if you need to, because you cannot make a pricing based on, um, on time, as I said, because one customer is running it for 10 minutes once a year, having high value and others needs to run it the whole year and having low value and a high value, but low value per time. So that's not comparable.

AI assessment note: “We will, we will have, uh, in the, in the future, some standard pricing”