How Quantum Chemistry Could Become the Next Engine of Industrial Innovation

Hello and welcome to the latest edition of the digital CXO Leadership Insight series. I'm your host, Mike Fazar. Today we're with Loon Wang, who's the CEO of QDX technologies, and we're having a chat about how quantum computing will be applied to chemistry 'cause well, it might be one of the killer apps out there for it.

Hey Lou, welcome to show. Thanks for having me, Mike. So I think people, you know, at least they're aware of quantum computing.

I don't think most people truly understand how it works, but there are a lot of use cases that have historically been intractable for conventional computing, including the infamous, uh, Schrodinger's cats and associated equations. Correct. Um, so from your perspective though, what's exciting here?

What are you looking forward to and, and what can we do today versus maybe tomorrow? Yeah. Um, I think like it's important to understand that quantum computing basically has sort of like two ways of thinking about it.

There's using these new quantum computers that are getting built to try and do computing in a, in a fundamentally new way. And then there's like using conventional computers, classical machines that we've had for decades to try and predict quantum phenomena. And they're sort of like two sides of the same coin.

Right. I like to frame it as either you use quantum phenomena to try and do computing or you use computing to try and predict quantum phenomena. Um, I think today we are yet to see, um, large scale applications of quantum computing.

So like using these phenomena to, to do computing. But what we're increasingly seeing that we're capable of doing, um, and sort of what we're focused on as a company is how do you use conventional computers to do a lot of the things that we thought we would need quantum computers to do. But it turns out we actually don't.

And it turns out that you can make qu uh, you can make conventional computers fast enough that you can still solve many of these challenges. Hmm. Well how do you achieve that?

Exactly. 'cause a lot of folks have at least been convinced that we need, you know, massive data centers to run quantum computers and, uh, we may not see the benefits of these things from the end of the decade. And you seem to be saying we can do a lot with what we already have.

Yeah, I think, um, what it, what requires is going back to the drawing board a little bit with some of the algorithms that have existed for, you know, decades. Uh, I think we're approaching the hundredth, uh, anniversary of Schrodinger publishing his equation, uh, and thinking about how to rewrite them and redesign them for modern computers. A lot of these algorithms would've been written well before the burst of, you know, the GPU industry.

And right now that's going, you know, like crazy because of ai and there's a lot of these massively parallel accelerated pieces of hardware that you can actually take advantage of, um, and upgrade all of the old algorithms that we used to use. Uh, and that's kind of at a very high level how one unlocks a lot of these use cases that, you know, we thought we might need quantum computers for. There's still a whole section of problems that you definitely need quantum computers for, but in chemistry at least, there's a lot of things that look start to look like they're, they're attractable either now or in the next couple of years.

A lot of those algorithms to your point, were written, I don't know, as far back sometimes as the 1950s, um, who's gonna go in and kinda rewrite those algorithms. Is that something that you guys are doing or is that something a community of researchers needs to do? Or how does it come about?

Um, it's a mix of both. I mean, there's lots of different people trying to do this, but it is one of the things that our company has done. Um, and so the way that we were able to get the kinds of speedups that we have been able to get was by going, you know, going back to the drawing board and rebuilding all of this technology essentially from scratch for the modern uta.

Are there things that you expect the chemistry sector to be able to do with conventional computers either in the next few months or years that, you know, people would be amazed by what's, what's on your to-do list? One of the things that um, is particularly exciting for quantum chemistry is the ability to look at reactions. So typically, you know, if we could, we probably would simulate everything quantum mechanically.

The only reason that we don't is it's like intractably hard, uh, requires a huge amount of compute power, but as you start making your computers better and as you start making algorithms that are far more efficient at leveraging these new computers, you do unlock a lot of these capabilities that maybe even a couple years ago we would've thought we couldn't, couldn't achieve. So one of the ones that categorically is really interesting is like chemical reactions. So how do you model chemical reactions?

And when you're using quantum techniques to do this, typically we focused on really small systems, really simple or simple in terms of like how long it takes and, and um, how many atoms we're looking at. What we're starting to be able to do now is look at chemical reactions that involve entire proteins. So you're looking at like enzymes or cobell on binders, things that, you know, involve tens of thousands of atoms, 20,000 atoms plus, uh, and you can run that whole calculation quantum mechanically and that's just something that, you know, wouldn't have been possible even a couple of years ago.

Will this kinda drive innovations downstream because, well let's take for example, healthcare. There is a lot of research dependent upon what's going on in the chemistry world. So, um, is this bigger than just the fact that I can track some interactions, uh, for chemicals?

It just has implications for all kinds of things downstream. That's exactly right. I mean, at the end of the day, drugs are just chemicals interacting with proteins in your body.

And so if you can super accurately simulate the interactions between those chemicals and, and various proteins of interest, then you can massively improve the way that we drew drug dis do drug discovery. Not only can you make, um, the existing types of drugs that we're trying to develop easier to develop, cheaper to develop faster to develop, you can also start thinking about building new types of drugs that traditionally have been, uh, sort of avoided because of how complex they are to make and how complex they are to get right. But if you can model the chemistry really well, then suddenly these types of modalities become much less scary.

Um, and you also paved the way for just a lot of like maybe basic infrastructure. So one of the challenges that that can occur in the space is when you design a, uh, interesting drug or an interesting compound that you think might be effective, there's this follow on question of how do I make this thing, what's the series of chemical reactions that I need to do to bring this design into the world physically? Um, and that's kind of an unsolved problem in computing right now.

You know, give me a compound and I'll tell you exactly what series of, uh, reactions you need to do to make it. That's, that's not something that computers can do presently, but it's absolutely something that you can do with quantum chemistry. How do you perceive AI today?

And within that context? 'cause some people sometimes talk about AI and quantum as if they're two completely different things, but I wonder if as we go forward, are we just gonna wind up seeing these two things kind of meld together in a way that, um, gives us some greater outcome because one makes the other more accessible? I mean, that's exactly right.

So, um, we're already seeing these two things come together in our company today where, uh, on one hand you can use artificial intelligence systems to make quantum chemistry more approachable to the average person, right? Right. Now you have to be not just a computational chemist, but a very specific type of computational chemist to have a chance of correctly using quantum chemistry technology.

But with the advent of LLMs, um, we're seeing it's increasingly possible to have these AI systems design their own algorithms or to be more approachable to computational chemists as opposed to specialized quantum computational chemists or even medicinal chemists instead of computational chemists. And so the more that you can broaden access to the technology, the more that technology is gonna proliferate and the more of those sort of problems that quantum chemistry is useful for, uh, will be, will be solved. And then secondly, um, it's becoming, you know, pretty common to use AI systems to do design in like most spaces, whether that's drug design, material sciences, et cetera.

And at the end of the day, the artificial intelligence puts forward the design and it needs feedback on how good that design is. If you don't have super high accuracy simulations, your only option is to go into the lab, make the thing, test the thing, and give that data back to the ai. And that's a really slow loop 'cause it takes weeks to, to do that.

Um, and you'd like your AI ideally to iterate hundreds if not thousands of times on its own designs. But if you have simulations that are sufficiently accurate and sufficiently fast, then you don't need to go to the lab straightaway. You can do several iterations with the ai, you know, purely on a machine before going into the lab and spot checking your work.

Um, and if you want things that are sort of, if you, if you wanna be able to do that, you need simulations that are super accurate, super fast. And at the end of the day, the best thing there is is quantum chemistry. Where is the funding for this coming from?

Because, you know, we see massive amounts of dollars being poured into AI and we'll see massive amounts of money lining up to be poured into quantum. But um, are people looking at quantum chemistry and and and kind of equating the two or does it kind of fall between the two sometimes and, uh, and, and everybody else thinks somebody else is gonna do something about it? I think it integrates nicely with the two.

There's a lot of complementarity between sort of the, the typical approach to quantum computing and how you accelerate quantum chemistry methods. And there's a, a lot of overlap between how you use artificial intelligence to sufficiently advance and accelerate these quantum chemistry methods. But it is sort of like this in-between stage.

And I would say it's sort of like an emerging field. There are certainly not, you know, VCs that are specialized in quantum chemistry, whereas there are VCs that, um, you know, specialized in AI for example. Um, but I think like the deep tech investors typically have a really broad, uh, range of capabilities, uh, and they're typically able to get their heads around this kind of technology, uh, and see the value in it.

What would you like to see organizations do to help us get to this goal? Is there something that the federal government should be doing or is there something that research labs should be doing? What's kind of missing from your perspective?

I think the change in the scale at which you can use quantum chemistry has changed really rapidly over the last even two to three years. And I think a lot of people haven't caught up with that. And when they think about using quantum chemistry, they still think in a framework of five to 10 years ago where they say, okay, well we can do quantum chemistry, but it's really restricted.

We can only look at systems of, you know, that are very small of a very small timeframe. Um, so even though we get really accurate answers, we can't look at really big systems that are, that are interesting to us. Um, and I think people just need to sort of look around and realize that that's actually not true anymore.

And if you wanted to do thousands of quantum chemistry calculations on protein size systems, you know, that's something that is actually possible today cheaply and and efficiently. Um, and the kinds of research that you might do, the kinds of experiments that you might run dramatically change when you sort of reach that sort of thousand fold scale up, uh, things become sort of quantitatively different and take on an entirely new quality. Are there things that you're hoping that maybe will solve or issues that we've been unable to kind of wrap our heads around, whether it's, I don't know, some sort of disease that we can't really understand yet, or some interaction between, I don't know, energy and humanity and the planet, but what things do you think might get solved in the next half a decade or so?

Because we're investing in quantum chemistry, I'm really interested in two problems. Um, one that I know a lot more about and one that I know a lot less about. Um, the one that I know a lot less about is plasma facing materials.

A lot of really complex quantum chemistry calculations at a, require a very, very high level of theory in order to design the kinds of materials that you can use in modern and emerging fusion reactors. Um, and that I think would be, you know, I, I think the idea of helping advance materials in the direction of, of new types of energy is just incredibly important for humanity. Um, the other area that I'm really excited by is simulating something called the SIP family of enzymes.

So it's this sort of series of enzymes that are in your liver and they're responsible for metabolizing 80 to 90% of all the drugs that we produce. Um, but because that metabolic process is a really complicated chemical reaction are probably one of the most complex chemical reactions that happens in your body, no one has ever simulated this thing end to end. So there's very little information on how it actually works on any given drug.

Uh, and it's very common to see toxicity show up in the clinic when you put a compound into humans for the first time. That is a consequence of us not really understanding how the SIP enzyme was going to interact with that drug. And I think in the next year or two we'll be able to simulate enzymes like that and completely change how med chemists think about toxicity.

So what's that one thing you kinda see us doing today that makes you shake your head a little bit, go and say, you know, folks, maybe we're paying too much attention to that and not enough to this and, you know, is there something we should be doing in at the university level now to kind of change everybody's mindsets? That's a really good question. I think hitting the same benchmarks over and over again.

So what's pretty common in this space is that you have a series of benchmarks that sort of sit there for many years and people develop algorithms against those benchmarks and try to retrospectively prove that their technique has some predictive power. Um, something that the AI space is doing really well right now compared to, you know, other areas is they're constantly updating these benchmarks. It feels like every month, every other month there's some new benchmark that people are trying to produce that better captures the idiosyncrasies of something that's super intelligent.

That just doesn't happen in the chemistry space. Um, the, the rate at which we release new benchmarks to try ourselves against, uh, and the relevancy of those benchmarks to the problems that people actually care about, uh, in the world such as drug discovery or material sciences. There, there is correlation there, but it's not as good as you would want.

And I think a lot more effort goes into trying to beat these benchmarks than goes into trying to build good benchmarks that are worth beating. Uh, and I think that's probably something we could afford to pay more attention to, uh, at the researcher level. Folks, you heard it here.

We think quantum computing is cool, but maybe quantum chemistry is even cooler still because, well, we're talking about the building blocks of life at the end of the day. Hey, lo, thanks for being on the show. Thanks so much, Mike.

All right. And thank you all for watching the latest episode of the digital CXO Leadership Inside series. You can find this episode and others on our website.

We invite you to check all those out. Until then, we'll see you next time.