Quantum CEO Series: Alan Baratz

Thank you, Eric. It's a pleasure for me to be here.

Sure, happy to do it. There are a lot of different technologies that are being used to build quantum computers. Everything from superconducting to trapped ion to neutral atoms to photonics to spin qubits. And each of these technologies have different advantages and disadvantages when it comes to building quantum computers. But at a higher level, there are two main architectural approaches to building quantum computers. One is called annealing and the other is called gate. And what's interesting about these two architectural approaches is that they are each good at solving different classes of problems. So, annealing quantum computers are very good at solving business optimization problems. Things like workforce scheduling or manufacturing plant floor optimization or autonomous vehicle routing. But annealing quantum computers are not very good at quantum chemistry or drug discovery or designing new materials. Gate model systems, on the other hand, are very good at quantum chemistry and materials design, but they are not very good at solving business optimization problems. And so, as a result, both of these approaches to quantum computing are important in the marketplace. D-Wave is a unique company in this space in the sense that we're actually the only dual platform quantum computing company. And what I mean by that is that we're the only quantum computing company that is developing and bringing to market both annealing and gate model systems to be essentially the one-stop shop for quantum so that we can solve whatever hard computational problem you have. If you need annealing, we've got it. If you need gate model, we've got it.

It absolutely is. So, D-Wave has always been the leader in annealing quantum computing. That's where we started. Our annealing quantum computers are in their sixth generation. These are the largest and most powerful quantum computers in the world. They're the only quantum computers on which true quantum supremacy has been demonstrated. Specifically, we can compute properties of magnetic materials in minutes that would take millions of years to compute on the fastest supercomputers. And they're the only commercial quantum computers today actually being used by customers as part of their business operations. However, we have not been at the forefront when it comes to gate model quantum computing. But, at the beginning of this year, we announced the acquisition of Quantum Circuits. This was a spin-out from Yale University that has developed some revolutionary gate model quantum computing technology that basically has catapulted us to the forefront of gate model quantum computing as well. So, now we believe that we are world leaders in both annealing and gate model quantum computing.

So, yeah, so again, this goes back to whether we're talking about annealing or gate. If we're talking about gate model, those systems are not yet commercially viable. They're still very much in the R&D phase, and there are still some very challenging problems that need to be addressed before we will see gate model final computers large enough and powerful enough to solve real-world problems. However, in the annealing space, we are there today. As I said, we demonstrated quantum supremacy. We're the only company in the world that's demonstrated quantum supremacy on a useful real-world problem. And we have customers that are using us as a part of their business operations. For example, we have one of the world's largest airlines. We have one of the world's largest healthcare companies. We have one of the world's largest aerospace companies. We have one of the world's largest mobile carriers. We basically have a number of very significant companies that are leveraging us for a variety of different types of applications. Let's talk about a chemical company. This is BASF, one of the largest chemical companies in the world. So, one of the problems that they were grappling with was how to efficiently fill orders. So, they get orders for different chemicals. They've got a production facility that fulfills those orders by mixing different components together to create the chemicals to fulfill the orders. What they found is that by using our quantum computer, they can reduce the time to compute the scheduling for fulfilling those orders from what was taking them 10 hours to compute the schedule down to just seconds to compute the schedule. Now, why is this so important? The reason why this is so important is because things happen. A supply truck might be late, and the production schedule that they had computed might need to be modified. Well, if it takes 10 hours to recompute the schedule, you know, you're going to have to wing it. Whereas, if it takes seconds, you can immediately recompute the schedule. Another example, we've been doing some work with Anduril and Davidson Technologies, two government contractors. We basically did work on missile defense, very timely application area. So, essentially, what we did was we demonstrated that on a 500 incoming missile simulation, we can compute the solution to neutralize that incoming threat in 1/10 the time. A factor of 10 improvement in the time to compute the solution. Obviously, when you're trying to take out incoming threats, time matters.

And our solutions were able to take out 10% more threats. So, on the order of 50 more threats were able to be taken out with the solution being computed in 1/10 the time. Here are just a couple of examples. Maybe I'll just give you one more. We are also the only company with a quantum computer that's been able to essentially address large language model training in a real commercial application area. So, this is work we did with Shionogi, a drug company in Japan. And essentially, they had developed a large language model for generating new molecular structures for human drugs. They then retrained that model using our quantum computer. And what they found was that the molecular structures that were being generated were far more well-suited to human drugs. So, they basically got more accurate models that they were able to train faster with lower power consumption. And of course, power consumption is a really important issue when it comes to AI and model training.

Yeah, what most people don't actually recognize is that most of the problems that businesses need to solve, the computational problems that businesses need to solve are what we call exponentially hard problems. So, what this means is that the time to compute the optimal solution grows exponentially in the size of the problem. So, when you get to even medium-size problems, let alone large problems that businesses need to deal with, that's outside of the reach of classical computers to compute the optimal solution. They can't do it. There's not enough time. There's not enough time in the universe, right? Or in your lifetime. However, what they end up doing is developing heuristics, simplifications to the problem that allow them to get solutions that they hope are good enough. And that's pretty much how the world works today. It runs on heuristics where you simplify the problem and hope that you're getting good enough solutions. Well, quantum computing is so much faster and so much more powerful that in typically an even shorter period of time, we're able to compute much better, if not optimal solutions. So, it's not that these companies weren't computing solutions before. BASF was computing solutions and they were probably good enough solutions for what they needed, but with our quantum computer, we're able to get them even better solutions much faster.

It's a combination. It is better solutions faster. We want to tackle the whole problem, not a simplification of the problem, to allow for better solutions and we want to compute those better solutions much faster.