Okay. We'll go ahead and get started. Good afternoon, everybody. I'm Quinn Bolton with Needham. It's my pleasure to introduce our chat with IonQ. IonQ was the first public, sorry, pure play company, quantum computing company to come public in 2021. IonQ's quantum computers are based on ion trap technology, which has a number of advantages compared to other types of qubit technology, including longer coherence times and higher gate fidelity. Based on these advantages, IonQ's systems outperform competitor systems on benchmarks based on real-world applications. Joining me from the company today are CFO Thomas Kramer and VP of FP&A and Investor Relations, Jordan Shapiro. Thomas, Jordan, thank you for joining us.
Thanks, Quinn.
I'll start with a real quick overview of quantum, sorry, of IonQ. Maybe just give us a quick history of the company and the ion trap modality. Tell us about the sort of pros and cons of ion trap versus the other qubit modalities in the quantum space.
Absolutely. Thank you. And thank you, everybody, for coming. Quantum computing is complex. It offers the premise of solving problems that classical computers cannot solve in, either cannot solve, period, or cannot solve in a reasonable amount of time because the computation is done differently with massive parallelism inside the compute itself. People have known about quantum computing since the '70s, where Feynman, who was a scholar, academic, and broke lots of ground in this area, said that, well, nature isn't classical. It's not digital. It's quantum. And you can't model it as well when you translate everything to zeros and ones. So that was great. And we knew about quantum mechanics, which is a field within physics. So people tried to find out, okay, how do you do quantum computing? Turns out that that wasn't easy.
And it wasn't until the 2010s when one of our co-founders did research on atomic clocks that he realized that this technology was so stable that you could compute on it. And he demonstrated the first quantum gate. The atomic clock project he worked on got a Nobel Prize. And we're busily waiting for ours. But out of that, the reason why we are using what's known as ion traps is because of this research. And we found this to be a very stable compute platform. We literally take atoms from nature, and we encode them with values, do computations, and read out the results. The main other technology that I'm sure you've heard of is called superconducting. And this is using a silicon substrate, and you jam in some synthetic qubits, and you compute on them.
Qubits, the quantum bit, the main building block of a quantum computer, they are very, very sensitive to the environment, and they have errors in them, but if you manufacture them, you have more errors, and as a result, in order to, and since they're susceptible to the environment, putting them into superconducting with heat and vibration deteriorates the computation more. One of the ways that have been sought to solve this is to cool down the entire computer to 0.1 Kelvin or near absolute zero. This is costly, time-consuming, and hard, and it hasn't led to better fidelity. Fidelity is a measure of the quality of that qubit, and so, meanwhile, we've been producing quantum computers that operate at room temperature. They draw 12 kilowatts, so the equivalent of two and a half wall sockets in your office, and we are charging ahead.
The important thing to note is that the number of physical qubits isn't meaningful unless you can compute on them, and so there is a term called algorithmic qubits or AQ. We think of them just as good enough qubits to compute on. Nobody has a platform today that can compute on 500 qubits, so if their chip has 500 qubits, that's a lot of qubits that are not being used, and that is our goal, is to drive up the computational capacity by introducing more and more qubits that can be computed on, and we can do this because of the much higher fidelity and the more stable platform that ion traps offers. The decoherence time that was mentioned, that's actually a measure of how long your computer can compute anything before it needs to be reset or literally falls apart and can no longer compute.
On superconducting platforms, this is measured in microseconds, so millions of a second. On ion trap, the current rating is 3.5 seconds, which is actually a really long time. But if you can only do that in microseconds, then you just—you don't have a lot of time to compute anything. But that's a brief overview of the platforms. Happy to do follow-up questions afterwards.
Yeah. No, well, I've got a lot of questions. So I guess I wanted to move next to quantum computing. It's certainly caught the eye of many investors over the last three to four months. There have been a number of sort of just broader announcements in the industry. And so I was wondering if you could comment. What do you think some of the bigger announcements in the last three to four months have been at the industry level? I know the Willow announcement from Google was certainly one. You've had a couple of very nice contracts yourselves. And so maybe highlight a couple of the industry events or announcements.
Absolutely. One of the biggest things for us as a young company was we announced an acquisition of a company called Qubitekk. And Jordan here did this deal. So maybe you can talk a little bit about it.
Yeah. Qubitekk is a company that's focused on quantum networking. And that's the idea of linking quantum computers together, which enables very secure communications, virtually unhackable communications. And so this is one of the bigger announcements coming out of IonQ recently. About a year ago, we announced we would do quantum networking in addition to quantum computing. We came out and recently announced a large deal around quantum networking. That's in September. We announced a $54.5 million deal with the Air Force Research Lab, or AFRL, all focused on quantum networking. We came out with another one on Monday, a $21 million deal with the Air Force as well on quantum networking. And the Qubitekk acquisition reinforces that.
It's a company that has been focused on this technology, bringing 118 patents to IonQ, U.S. and international patents, bringing a team that is expert, focused on quantum networking for a decade, and then bringing a long Rolodex of customer contracts and contacts. So this is just one of the areas that we've expanded into. But there are also, Thomas, I think, other recent announcements that we've made that are key.
Absolutely. One of the largest announcements we made last year was that we announced that we had sold a contract to the Air Force Research Lab for $54.5 million, where we're going to work with them to do, among other things, quantum networking as well as quantum computing, and yesterday, I was in a press conference with the Governor of Maryland, Wes Moore, and the President of the University of Maryland, sorry, president of, yes, University of Maryland, Darryll Pines, where they announced a $1 billion initiative to create the Capital of Quantum, where they want to create a campus for quantum AI and intelligence, and we have been chosen to anchor this initiative with them, where we will move into an entirely new building. That will be our headquarters in Maryland and work with the university and other players to develop new quantum solutions, which is certainly a very strong effort.
It mirrors the announcements from the University of Chicago, except with a compute platform that is commercially available now, as opposed to in six-to-seven years.
So that's just a few of the announcements in the last few months. You've got the acquisition. You have these large Air Force deals on networking. You have the partnership with UMD and Maryland State. And that's just coming out of IonQ. And I think to your point, Quinn, there's been a ton of deep interest in quantum computing that has really surged recently. And it's because the momentum in the space is becoming so clear at IonQ and across other companies in the industry as well.
Wanted to get your response to comments last week from Jensen Huang at NVIDIA, who said that it may take 15 years-30 years before we see a commercially available quantum computer. How do you respond to that? I assume you may have a different view.
In one word, yes. But we invite everybody who makes prognostications about the industry of quantum and the timeline to insert at the end of their statement on our platform. And since NVIDIA doesn't have a hardware platform, it would probably take them that long to create one. And you also see between different modalities, like superconducting, have always said that they are 10 years-15 years out on their platform because there are scaling challenges with the superconducting platform because you need to have an error correction rate of 1,000 physical qubits to one logical one to compute on. That just puts it further out in time. Our correction algorithm requires 13 physical qubits to one, which means that we need far fewer physical qubits.
We also have partial error correction that works on only certain gates, but the most widely used gates called Clifford gates, where we only need three physical qubits to one. So I think the—I think that comment, frankly, has been taken a little bit out of context. It was a Q&A to a different speech. But the people invest in opportunities that will give you both long-term and near-term payoffs. What we have seen so far is that at the end of 2023, there was $42 billion invested in quantum. And that is now a year ago. So there's more now, I would say more than $50 billion. The likes of Microsoft, Google, NVIDIA, Intel, obviously us, are all investing heavily into quantum. It's hard to fathom that this much activity goes into something that will have a payout that's that far in the future.
It's frankly hard to fathom that you would build an entirely new version of CUDA for quantum if you thought nobody would be using it for that long. And what we're seeing from where we're sitting is that we're doubling recognized revenue year over year since we went public. There is incredible demand out there. And given the press that we get from Jensen, there's even more interest. So I would just say, well, insert the on my platform after the statements. And we are incredibly proud of the fact that we're going to release a 64-qubit system at the end of this year. 64 qubits is enough for a quantum computer to evaluate 18 quintillion possible solutions to an optimization problem simultaneously. What is a quintillion? It is not a workaround on your name. It is actually a number. It's a scale of numbers with 18 zeros.
It's so large, it's hard to put in your head. But the world's largest supercomputer is called the Frontier. It sits at Oak Ridge National Labs in Tennessee. It can compute 1.3 quintillion floating point calculations in a second. So that gives you a reference point. The optimization problem will be done in much less than a second. So I would say that we're looking at the industry now, and we are seeing more activity, more interest, more contract discussions than we have ever in the past.
There was a nice sort of segue talking about your AQ 64 system. You guys introduced the AQ 35 in 2024, about a full year ahead of schedule, which was a great accomplishment. Going from 35 algorithmic qubits to 64 seems like a bigger jump. So what are the biggest challenges, or what do you see as the key hurdles to reaching that 64 algorithmic qubit machine?
So at this point, there are no hurdles. There were hurdles. These were R&D efforts that we have worked on and have come up with separate answers to separate parts of the puzzle. For instance, we introduced a new chip, a chip that holds the qubits. So this is what's going to be in Tempo. It's made with actual gold, but in a very thin layer. And the qubits themselves sit in a vacuum chamber above it, floating in midair, and then we compute on it. Coming up with this chip takes time. And reiterating on it and making sure that it works also takes time. Coming up with a new system for how to target the lasers onto the qubits also takes time. But these have been done. We are also cramming in more qubits on the chip than we have done in the past.
So instead of having all the qubits in one long line, a chain, we're now going to introduce several chains. So we will have multiple cores on the chip. That has also been figured out. What we're doing now is the engineering of putting it all together. And that is going on track. I was reading an update from our head of engineering today. And I was smiling so widely that people were asking me what I was laughing at. And I'm like, yeah, it's just good to see all the developments that come out of all the work that we've put in.
It's notable. Nothing about this chip is quantum. The only thing that's quantum in our quantum computers are the ions themselves that are behaving as qubits. That's a classical chip, and I think that's indicative of what's happening throughout the computer, which is that these are known technologies to us, and there's a lot of engineering work that goes into iteration, but there's not novel science. We're lucky to say that we've solved science problems, and we're focused on the engineering problems from here.
You mentioned that the partial quantum error correction technology that you introduced back in, I think, August. Expand on that. Why is it so important? Your overhead is much lower, I think, in some of the quantum error correction work that you've done. But talk about this approach for partial error correction on the Clifford gates. And how important is that?
So it's very important, and it allows us to get to scale quicker. It introduces a partial solution to a challenge that quantum computing has in that all computing is error-prone. We, as users, don't see it because the errors are caught ahead of time and corrected. So when we're on a Zoom call, that is fixed by sending more packets than the recipient needs. You just bombard the recipient with pieces of your video call, maybe 2x, 3x what you need for that call. And then the recipient's computer pieces together with what it needs, and then it drops the rest. This is a fundamental technique in all computing. Now, error correction on a quantum computer comes with a tax. There's two types that have been known as error mitigation and error correction.
Error mitigation has to do with correcting the input factors, the data, because you can spot where bad data is, and you can lock it out of your computation, thereby improving the result. That is time-consuming. So it will increase the length that your computer has to run to solve a problem. Error correction is the art of taking several physical qubits and combining them to create one logical qubit that takes strengths from all of those qubits, and you compute on this result instead. That's very taxing on the number of qubits because now you just reduce the capacity of your computer. So finding a partial error correction solution that only needs three means that it preserves capacity of the computer. It's completely manageable in software. So you can turn it off and on when you need it. So you don't lock out two-thirds of your capacity immediately.
It's just for those computations you need, and it's done on a particularly noisy gate that, when noise just means that more of the answers you get out are incorrect than otherwise, and so if we can fix the noisiest gate, and the gate is the lowest level of computing, then you've got a lot of your work done for you, and you've got it done very cheaply in terms of the capacity of the computer, so this is a massive advancement for us. We continue to believe, and we've proven, that you don't have to wait for a fully fault-tolerant, error-corrected computer to make good use cases and prove economic value of a quantum computer, and this is important because we are not operating a research lab. We are operating a commercial quantum computing manufacturing operation. We are also operating a commercial software operation.
If there wasn't a commercial interest in it, then it would take too long and be too expensive.
People ask us why we're so confident that quantum is now. And a lot of it comes back to these three trends of the hardware getting better, meaning we can put more qubits into a quantum computer today. The software is getting better, so the algorithms require fewer qubits to get interesting problems done. And then the error correction is getting better as well. So when we can show that we have a much smaller overhead to run a problem that is meaningful on a quantum computer, the fact that those trend lines are coming together means that quantum computing is a lot nearer term than some people might think.
Yes. You guys had mentioned the sort of push into quantum networking earlier and wanted to expand on that a little bit. I believe in the past you've sort of mentioned you think the company has an inherent advantage in quantum networking, and what if you could sort of explain why you think that is and what attracts you to the quantum networking aspect of quantum?
Yeah. This goes back to our architecture, which from our founding was always this modular architecture where you could connect multiple quantum computing chips together via what we call photonic interconnects. It's a fancy way of saying you connect them with light. And so networking has been something that was on our roadmap forever. It's always been. But it was something that we didn't anticipate doing as early as we've ended up doing it. And the reason that we moved on it earlier than expected is we would have conversations with customers who would say, we'd love to purchase a quantum computer or work on your quantum computers, but we also want to be able to connect them. We're interested in the networking side. We're interested in the communication side.
And while for us on the compute side of the house, we thought about that as creating clusters that were more powerful than the individual system. On the networking side, it means there's a whole different set of applications that you can tackle. So we started selling what we first thought were one-off type networking deals. And what we realized was actually a large burgeoning and growing industry in quantum networking. So there's been a lot of excitement about that. But that kind of connects to the acquisition we made and the recent deals that we've been selling. And we think that quantum computing and quantum networking are both large industries and have a ton of potential. And IonQ is uniquely suited because of our architecture to go after those opportunities.
Think about your iPhones. Almost everybody has an iPhone. Or if you have something else, it looks like an iPhone if it cracks like an iPhone. But we can all sit and play Solitaire. And it's great. Graphics are good. We can take pictures. But it's a phone. If you can't call anybody, it has limited usefulness. In fact, most of our computing today is done when you're connected to other machines. That doesn't exist for quantum computers yet. We're creating it. This will be the internet of quantum computing. And figuring that out is vital to the industry, which is why we've had it on our roadmap. And we just found people realized that sooner than we had thought.
That has allowed us to put more money into developing these solutions because our customers are actually paying for them by buying early quantum networking equipment that allows us to push these boundaries further.
You'd highlighted the $54.5 million deal with AFRL, the more recent, I think just this week, $21 million deal, both of those quantum networking related, and then also the deal with ARLIS. So you guys are getting close to, I think, the $100 million mark in terms of value across those three awards for quantum networking.
That's right. That is right. Why did Jesse James rob banks? Because that's where the money is, and when we're seeing that there's enough economic interest in this area, we are willing to pursue it now rather than waiting, so AFRL, or the Air Force Research Lab, is obviously a defense institution, and quantum computing can be incredibly important and vital for defense operations to calculate troop movements, how much supplies do you need, which routes should they take, and also what are the potential routes that a ballistic missile can take and how can you counter it. Now, that is great. You need to calculate that, but you also need to be able to communicate that calculation to installations on the ground and other operators. Hence, you see the government being very interested in networking, but it goes further than that.
I think I get one email a week with a notification that my password and username has been leaked. So while we have a modestly secure infrastructure for IT and our digital lives, it's not very safe. And it's not much more safer when it comes to bank information than it is when it comes to checking up house prices on Zillow. So what's needed is just better networking, secure networking, something that quantum can provide a much stronger layer of than classical. So that's just on security of networking information, classical information, too. Additionally, you can network quantum computers. And it turns out that the overhead of doing distributed computing on quantum computers is much lower than classical. So you can indeed envision a hive of quantum computers working on the same problem set that being physically distributed.
Jordan, you talked to us a little bit about the Qubitekk acquisition, which furthers your position in quantum networking. Maybe just give us a little bit more detail. What did Qubitekk do? What technologies or components does the company bring to IonQ? And then I think it's got an affiliation with the EPB Quantum Network. And so they've actually tested this in real-world optical networks. And so maybe tell us about the importance of that network.
That's correct. Yeah, so Qubitekk was focused on quantum networking through entanglement generation and basically communicating from one node to the next and having qubits at either end so that you could send signals across. And one of the unique things about the company is that they have created one of the first quantum networks in the world. And that's the one that you mentioned, Quinn, at EPB. That's the Electric Power Board of Chattanooga, Tennessee. It happens to be a very forward-thinking utility. And EPB is the utility that brought the gig economy to, or Gig City as a name, to Chattanooga. And they've also invested in building up this quantum network where they're actively running experiments with customers today. There was just an announcement a few days ago between Oak Ridge National Lab and University of Tennessee at Chattanooga running quantum networking experiments across the EPB Quantum Network.
And so that was one of the things that attracted us to Qubitekk is that they were building real products that were in production with customers and were running live experiments on quantum networks that were in situ. So in addition to that, the company had a number of relationships with universities, with local research institutions. And so we saw a ton of potential both on the technical side and on the commercial side with the company, not to mention the team coming to us as well.
Do they also expand your team into quantum cryptography and quantum key distribution? And is that a further rationale for that transaction?
That's right. So quantum key distribution is a slightly different flavor of quantum networking. It's not based on entanglement in the same way, but it exchanges quantum keys, effectively quantum code, to two nodes and allows them to basically pass information securely through those codes rather than through entanglement. It certainly expands our TAM. There's interest for QKD. We've seen especially large interest for QKD outside the U.S. So if you look at the APAC region and the Europe region, there are a lot of opportunities there as well. And Qubitekk brings all that technology to IonQ. It's not technology. Specifically, QKD is not technology that IonQ is already working on. So it's an expansion net new to IonQ.
Great. I wanted to move over to the quantum application side of things. On the third quarter call, you announced two collaborations with leading companies, AstraZeneca and Ansys, to develop or co-develop quantum applications for drug discovery and computer-aided design. Can you tell us about those partnerships? And ultimately, will the quantum portions of those applications run on the IonQ cloud?
So ultimately, the entire world's quantum compute demands will run on IonQ computers, of course. But jokes aside, we had originally thought, and this we were wrong at, that the pharmaceutical industry would be one of the last bastions to fall to quantum computing. But when they did, it will be one of the most profitable. This is because of the challenges with drug development. It is very time-consuming, expensive, and fraught with just human error. Because today, when we develop drugs, it's just trial and errors, two, three people sitting in a lab and saying, hey, what if I try and put these two things together? And you get some tissue, and you put it on there, and then you move on to mice and then to people. We don't have today a way of modeling chemical formulations, which is what drugs are.
And because when you model these combinations, you don't actually need to model just the combination of the atoms. You need to model the combination of all the spins of all the electrons. So it becomes a factorial where you need to see what are all the possible combinations of all the electrons of all the atoms in a combination. Very, very quickly, classical computers will collapse because they can't handle that compute level. Even for quantum computers, to do all of that for large problem sets, such as proteins and protein folding, will take more qubits than is currently available. However, what we're seeing is that there are smaller combinations that you can actually model and that will give you great benefits. We did a version of this for the Naval Research Laboratory last year where we were able to model rust, corrosion.
Corrosion is a $20 billion problem a year for the Navy. And the way this is solved currently, you develop protective coatings, i.e., paint, that's supposed to guard the ship from rusting. But the way you do it is you have your paint formulation now, and then you sit around, oh, what happens if I put coffee in there? Stir it around. You paint it on some slabs of metal. You put one in a marsh, one in fresh water, one in salt water. Come back in one and two and five years and see which one fared better. This is a hugely inefficient way of developing these combinations. Now, that is what Pharma wants to do.
Because if you bring a blockbuster drug and the value of a blockbuster drug hasn't been changed for 25 years, it's still just one that brings in $1 billion per year, far less than Ozempic, as we all know. But if you bring in approval of a drug by one week, you can pay for an entire quantum computer. So eventually, pharmaceutical industry will have lots of these. However, before we get up in large enough compute capacity to model proteins, we can use it in terms of clinical trials and choosing a patient population, choosing site selection, analyzing the data that came out of the trial. Because right now, you prove or not prove the indication you're looking for, and you scrap the data. Famously, Ozempic is a diabetes drug that nobody thought of because people aren't mining that data.
And so we see a lot of interest in this. One of the things about quantum computers is that they can do machine learning with far less data. So in statistical sampling, a lot of the time what comes out with insufficient data, there actually is sufficient data when using quantum algorithms as opposed to classical algorithms. And there's a lot of interest in this, obviously, in the pharmaceutical sector. So that's AstraZeneca. And the learnings here are not just applicable to AstraZeneca, of course. Ansys makes simulation software. So if you're manufacturing big things in engineering, also small things, we actually use Ansys software internally to model our chips and our architecture. But if you make a new airplane, if you're designing fuselage, and before you hammer it out and put it in a wind tunnel, you're going to simulate it on a computer.
That computer can run for six months. Ansys has 65% of the simulation software market for industrial engineering. What they want to do is to have quantum computing or IonQ as one of their compute engines. They see what it is they try to model. This is easy. We send it to a regular CPU. This requires more. We send it to a GPU. This is so complex that we need to send it to a QPU. We're going to white label our software into or our compute power into their software. That is the vision for that partnership.
Perfect. We've got about 90 seconds left. I wanted to ask, quantum stocks have performed very well through the fourth quarter. Many of your pure play peers have taken the opportunity to raise equity, take advantage of the stock prices. IonQ is the one who has not yet raised funds, so you are always sort of the best funded pure play quantum computing company, but just give us thoughts about the balance sheet. Is your lack of activity in the capital markets just an expression of your confidence that you're funded through profitability and you're comfortable with the balance sheet?
Yeah. So the reason why we haven't raised more money is that we have been and still are the most well-funded and capitalized company within quantum. We haven't seen that need. There's also we don't want to be known as the knee-jerk company that because the stock is trading at $50 now, let's go out and raise money today. I don't think that's the bargain that we're making with investors. Investors are not just investors. They're owners of the company. They're part of the company. We will want to make measured moves, which we did when we raised $650 million when we went public because we knew that this is a capital-intensive industry, but it's not as capital-intensive as all that. Unlike some, we have chosen not to create a fab because fabs are expensive. There are people who do it much better than us.
If there's things we can outsource, we do. And as Jordan said, most of the components in a quantum computer are entirely classical in nature. We buy them from catalogs. And apart from certain special things that we design ourselves, that's the way it should be. But that being said, I'm incredibly happy about some of our colleagues being able to raise money and extend their runway because I think it's good for the industry to have several operators that can together come up with solutions to the world's largest problems.
Perfect. Well, we should probably wrap there. Thomas, Jordan, thank you very much for joining us at the Needham Growth Conference. Really appreciate it.
Thank you, Quinn.
Thanks, Quinn.
Thanks.