Welcome everybody to day two of the Annual Age of AI Conference that Rosenblatt has been hosting for six years. The first one was actually before ChatGPT, which I think is interesting. My name's John McPeake, I'm the senior analyst here, and I cover quantum computing and AI software. I'm really happy to have Inder Singh, CFO and COO of IonQ, and Hanley D'Onofrio from IR of IonQ. Inder's been there since September of 2025. He was on the board of directors since December of 2021. He was previously EVP and CFO of Arm. He holds M.S. and B.S. degrees in engineering from Columbia, and M.B.A. from NYU. Me too. I've known Inder since we worked together at Prudential. I think that was his first finance job in the early 2000s. While we have a history, I have a buy rating on IonQ and a $100 target. We have lots to talk about here. It's great to have you with us guys. Thank you, John. Thanks for inviting us. Great. Let's start kind of high level, sort of the company vision. You guys have transformed really from a single product quantum computing company to a full platform. You're spanning computing, soon foundry to support it, networking, security, sensing, even space, all in roughly 18 months under Niccolo's leadership. How do you think about what holds all that together strategically and where do you think the center of gravity sits as you look out, going forward, Inder? Yeah, that's a fantastic question. Yes, it was great to work with you when I was on Wall Street, and learn from you and then maintain this resilient relationship over the years. I think that as I look at IonQ, where it was, of course, when I joined the board and was at Arm, it had basically no revenue, no real product deployed in the field, had one in its own lab, and was trying to build a quantum computer. Things that probably weren't envisioned five years before that, just as ChatGPT suddenly happened. The quantum advantage is still ahead of us, but the journey of building a compute device for this company probably started 30 years ago when it was originally founded by the founder, but really in earnest in 2021 when it went public. Since then, it's gone from essentially nothing in revenue to being the market leader in virtually every metric that you can imagine. But also I think importantly that, as you mentioned, it's gone from being a one-product company, meaning a quantum compute creator, to a platform company with multiple quantum and adjacent products. There's a slide that we have, which we use sometimes, and if Hanley could bring that up, it might be instructive. It's a slide I created, actually, when I joined the company in September to say what's changed. When I was on the board, and we were doing a CEO transition at the beginning of last year, as lead director, my role was to help the company find the right successor for Peter, who was retiring. At that time, frankly, we were in that bottom left hand tile that you see on this tile. We were building, at that point, probably our fourth generation computer, four years, one generation each. We had a definitely distinguished product in terms of the number of qubits and the compute power, et cetera, but still early days. We were envisioning and then are now deploying into the market our fifth generation computer. We'll come back to this one, but that's what got us here. That's the multi-generational leadership of creating quantum computers that scale and then work in the field, once you deploy them, and then add compute power each year. This year, 100 qubits. Last year was 36 qubits. Next year, we'll be selling the 256 semiconductor-based version of our compute device. We have a roadmap that extends out to 2030, going up to 2 million qubits and unmatched in my view from anyone that I've seen. Virtually every other quantum computing company you would name, almost all of them would be in the bottom left-hand tile. What Niccolo then did was quickly get us into the other tiles that you see. These are additional TAMs is how I look at it, John. There's a TAM of quantum compute in which we have a lead, and we intend to maintain that lead. What about connecting machines to each other? Much as a Cisco would do. They are a quantum machine to ours to someone else's. A unique capability that we kind of had in-house already, I wouldn't say this one required a major transaction, but we had acquired a company previously that we now started to say, "Okay, we can scale this and start connecting machines to each other." Realizing that as we're building more powerful compute devices, John, it's not just a qubits discussion, it's always like a when do you get to what I call fault-tolerant computing discussion. At that point in time, at that moment, probably within that year or the next year, we would be able to have a machine that hypothetically could run algorithms that break all kinds of encryption. To protect against, frankly, ourselves in a way. We needed to have the antidote to that. Quantum security, cybersecurity was an acquisition, a company called ID Quantique. They had number one, number two market share, depends on who you listen to. Incredible customer penetration in geographies like Asia, the APAC rim, where countries are really concerned about being hacked, et cetera, preparing to bolt down stuff, telecom networks first, with quantum-proof cybersecurity. That capability, again, distinguished from virtually every other quantum company. Yes, as we think about national security applications and if we think about potential national security customers, we acquired a few other types of assets that brought us the other two tiles that you see. The quantum sensing, which is basically not just timing, like atomic clock and the most accurate, but also jam-proof sensing that uses the gravitational field of the Earth to determine exactly where you are to a much higher precision than GPS ever will, but also is very hard, if not impossible, to tamper with and degrade the ability to operate with knowing exactly where you are, especially in a denied environment, in a confrontation situation. Also listening to customers, it would be nice to be able to put that sensing up on a satellite as well. Either ours or somebody else's. You now have in one shop, to your point, a multi-product company instead of a one-trick pony. You have a company that's also got, I think, the largest, if not the single largest, but almost largest quantum application development team in-house. As we build the capability of the machines, the iPhone, if you will, we're building the App Store to go with it. Again, I think unique in that differentiated way. You mentioned the foundry, which of course, we announced our intent to acquire SkyWater, which is a secure foundry in the United States. Listening to our customers and potential customers, quantum is so important, quantum is so transformative, we have to trust the entire supply chain, not only from a design standpoint, but from a manufacturing standpoint. It should be on shore, preferably. SkyWater seemed to be the one that was differentiated in that way, had those top-secret clearances needed to sell to government contractor, government customers, and had some quantum knowledge as well. Very unique in terms of a quantum foundry. We are working with them today, even through a commercial agreement, and really doing well so far. That's the narrative around how we went from being a one-trick pony, one product that's nice, big TAM, and we have a lead, to being a one-stop shop. That's how I look at the business, and you have multiple ways of getting customers. Some who start with computing or some who start with, "I need your security first," and then you do a land and expand. That's helpful. Thank you. You guys went into the public market a while ago, via a de-SPAC process, like all the other companies in Quantum. We did just have a regular way IPO, completed with Quantinuum, which just coincidentally does share your qubit modality. We'll talk about that in a little bit, but I'm curious whether you're seeing from your chair an increased level of knowledge or interest in quantum from us, from the investment community, over the past few weeks and months. Because you've seen the CEO of Intel going to the board of PsiQuantum, and the U.S. Department of Commerce made some investments in companies. It just seems like it's becoming more real, if you will, just over the last few months or so. Yeah, absolutely. I think that to your point, it's risen really in terms of the priority list of countries. I can't think of one that it's not really affecting already to being right up there, with almost anything else you can imagine, right? The potential of quantum, like when I joined the board, it was an early stage. Company had just gone public, to your point, so had two others. No one knew that any of these companies, frankly, will build a successful computer and which timeframe, et cetera. To me, the interest in joining was having sold Arm to NVIDIA and being in that holding period of getting approvals, maybe help look at, get a lens into the future. Like what will computing look like when you have, at scale, a quantum computer that can out-compute any other classical computer of any type on Earth, simply because it uses a different approach to computing, an exponentially more powerful approach. This company has proven now that not only can you build it for a government lab, but take it from the lab into the field and use it for useful things, even with fewer qubits than frankly are needed for really interesting solutions to life sciences and other problems. Proof points along the way, as I said earlier, with each new generation, new capabilities. As I look at where we're positioned, the modality that was chosen by the founder of this company 30 years ago, to your point, was ion trap. There are other modalities. There's superconducting, there's neutral atom, there's photonic, I could name some more. As an engineer, I don't say one is better than another. Some just have a natural advantage over others from the get-go. For, in a 100-yard dash and you spot me 40 yards, I have less to run than the others do. In a similar way, ion trap begins with fewer errors, begins with better fidelity, begins with the ability to have higher coherence time. These are all jargon that not everyone needs to know. You need all of these things to work in unison to be able to do computing at scale. Others will say they have one advantage here or one advantage there, but then they have to solve for three others. I think that ion traps in general, because the ions and atoms exist naturally in nature, you don't have to take temperature down to zero degrees Kelvin, for example, then create matter atoms, manufacture them and then use them, because then you're generating multiple sources of error. Again, one is not better than the other, it's just different. One has natural advantages, like ours is room temperature, almost, slightly below. Others, not so much. In an environment where you're creating a device that could be mobile one day, hard for me to imagine a superconducting thing being mobile in the same timeframe. When you have less error correction, you don't have to talk so much about it. You can say, "Logical qubits, how do I get there fast? What's my roadmap?" Just build a plan to put the numbers up that frankly, it would be very difficult for anyone else to match. I look at other players, like the ones you named, including Quantinuum, by the way, as fellow companies in the journey. We're in the early stages of quantum. We need the ecosystem to develop. These other companies, over time, will keep us honest in our innovation. I think that some of them will learn some of the lessons that we learned over the last couple of years, especially, that if you really want to scale, you can't use, for example, laser control of ion traps even. They will learn that getting to four- nines fidelity, which we achieved last year, ain't so easy, right? Those are the things that I think the learning curve of this company has benefited from. When Niccolo started putting together the entire platform, it differentiates it further. Yes, there are others that are using ion traps, so are we. We realize that the more qubits you have, the more lasers you need, the more expensive the machine, the bigger the machine, the more energy consumption. It's the opposite direction to what you need. All right. Picking the semiconductor path helps us solve that. That's a good jumping off point to talking about tech a little bit. You do have this 256 chip, electronic qubit control chip, I think 16 by 16, because that's how you get to 256. You have exhibited four- nines of two-qubit gate fidelity in a prior-gen chip. Could you walk us through where we are right now? You've talked about various tape outs, I think, with this device. What are the kind of engineering milestones between now and customer delivery? That'd be great. Yeah, of course. Thank you. I think to your point, we have announced that we have created the first working prototype of the 256-qubit chip. Remember, 100 qubits this year are being deployed with customers, and then a queue forming for the next generation machine with 256. Achieving that milestone of creating the chip is really critical. Working with our foundry partner, engineer-to-engineer team, they've been able to do that in a couple of months versus nine or 10 months that it might've taken otherwise. We're happy to see that not only can we create the prototype, but we can do it in much less time than even we thought. It allows us to turn our attention to the next generation chip, right? It doesn't mean we give up and we're done entirely with the 256. We have to put it into a system, test multiple systems, because we sell a working system. We deploy it at the customer site. That's the work still ahead of us. The really unique part of it was using ion trap and scaling to your point, using semiconductor. 256, to us, proves that. 10K is an extension of the 256 work, so the same team can begin to turn some of its attention to the 10,000 qubit chip, which probably we would've been focusing on more a year from now. We find ourselves in a position that the path that we've chosen, ion traps, helps with the science part getting solved. The path that we're choosing with the, really the expansion of 256 into 2 million, becomes more, in my mind, well-defined. You see here the things that are left and the things that are achieved already in terms of the check marks. Hanley, if you could, at some point, try to maybe bring up the tech roadmap one. I think that we're in an industry, again, where different modalities exist. We're an industry where some are still building a machine. We're an industry where there's still a learning curve. We've benefited from discovering that lasers just don't scale. Lessons like that and using now semiconductors going forward and having the first proof point of actually creating the semiconductor puts us onto this journey. A roadmap that I think from everything I've seen in public roadmaps that any company has issued is unmatched in terms of number of qubits. An important point for you and investors to understand, I know you do already, is it's nice to have the chip, it's nice to have the system, but the system needs to be industrial grade, and that means it needs to be fault-tolerant. The machine that we create must be fault-tolerant with enough qubits to matter. We've announced our walking cat architecture. It goes back to Schrödinger's cat. I won't go down that rabbit hole if you don't want me to. Essentially uses what's called cat states in quantum mechanics to scale from 10,000 to 20,000 to 200,000 to 2 million. In our case, without using more lasers, without using more energy, in fact, using less, and in our case, with the bill of materials, frankly, able to go down versus up, and then creating a roadmap that gets to quantum advantage potentially the fastest. I think fault-tolerant computing, which is probably somewhere in the middle of this period that you're seeing, is essential and candidly the only thing that really matters. You had talked about, I think it was like A, B, C, D, or something relative to the- Yeah. ...tape outs. Yeah. We're done with the tape outs on the chip? Is that the way I should think about it, Inder? Yeah. I think as with anything in semis, it's all about multiple tape outs usually before you have a semiconductor logical design burnt onto silicon and works. Multiple tape outs to get to that, you go for yield. We've gone through, to your point, A, B, C, which are early prototypes of any chip design company. D, where now you have the features come in and the metrics that matter, the prototype that you want to create. Doesn't mean it's perfect, doesn't mean it's got everything working. We'll iterate maybe once or twice more if needed. We feel comfortable enough where we are moving from what you see there as the third box, the chip prototype testing, into component testing, system level testing. System is the entire machine that surrounds the chip. We don't sell the chip. We actually build a quantum system deployed at a customer site. We'll do multiple iterations of the system, John- ...to make sure the system is stable and works and has the features that it needs to deliver, SLAs that we sign with customers. Again, this will go into our sixth-generation machine. Five generations of machine learning, like learning ourselves, helps us perfect the other parts of the machine, then having the semiconductor now take the place of a bunch of lasers and use less real estate, less energy, less cost gives us an advantage. When the 10K comes, John, the upgrades from machine to machine become modular, so you don't have to forklift out the old machine to put in the new machine anymore. The chassis stays the same, most of the machine stays the same. You have a few modules that you have to upgrade. It's like a new NVIDIA GPU in my gaming PC or something. You just pull it out of the chassis and put a new chip in. Is that the way I should think about it almost? Exactly. For a customer, you become very sticky. "They've made the investment of the machine." It may have 10,000 qubits in it day one. By year two when we have the 20,000, they don't replace the machine. They still keep our machine and they get our new modules and so on. Yes. In other companies I've worked in, Cisco, et cetera, that's how you establish footprint and incumbency. Right now we're all about establishing the fastest customer footprint around the world and then leverage those relationships over time to not only sell them more powerful computing devices, but the other things that we bring like networking security as well. In terms of the 10K chip, what kind of timing are we looking at there from an R&D standpoint? Is that occurring now? Is the 256 kind of in a chassis now? How should I think about where we are in the brown, I guess, part of the whatever color that is in process? Yeah. Zooming out a little bit. Yeah. I think that if we could get that five-year roadmap back up again for a moment, Hanley. This year we're shipping our 100 qubit laser-based system. Next year we'll be shipping the 256 qubit system. The year after that we're intending to ship the 10K system. By getting to the hardest part, which is scaling using semiconductors, ion traps using semiconductors, ion traps controlled by electronics, not by lasers, the first proof point happened much earlier. I'm not prepared to say that we can now shift the entire roadmap all the way to the left, but we have indicated that the acquisition of SkyWater enables us to shift some of the outer years to the left. Engineers are engineers. When you press them and say, "Can you do it faster?" They will say no. We're building machines that have never been built. Ever by anyone. In that context, we want to stay true to form. When we say that we're going to achieve 10,000 qubits in a chip on this roadmap in 2027 and then ship it to customers the year after, if we can do it faster, we will, of course. At the same time, we want to make sure that when you get to quantum advantage, and like I said, it's the middle of this timeframe, that that's a fundamental moment. It's not like ChatGPT, where it's going to be thrown out at consumers, and they can suddenly have it. It'll be a machine that can do amazingly good things, in life sciences and financial services and battery and chemistry and all kinds of things where there are too many variables for classical computing, also do things like cracking encryption that you may never read a press release about. But that may be the ultimate app. Right. Are you going to manufacture the machines in the United States? I know Oxford Ionics is obviously over there. They got their R&D lab, et cetera. When I joined the company last year, we had basically demand exceeding supply, we increased the manufacturing capacity, both in our College Park, but principally in our Seattle, Washington area factory. We've now added two more locations for the 256 and beyond. One is in Oxford, we will be assembling the systems. Also in Colorado, in a new facility, we'll be also building the system. Chips come out of SkyWater, systems come out of our four factories. Four factories for the 256. For all the generations of semiconductors. For all the generations. Yeah. If we have to build two more, we will. I think that's where having the financial firepower we do gives us optionality. There are other ways to go about the manufacturing over time, and we'll look at those when we have a repeatable model and manufacturability, because that's what we're designing for. I should think about the customer deliveries of the machines. I think we talked about this on the call a little bit, sort of 2027. We'll see the machine this year running. That's the timeline, yeah? Yeah, that's my expectation. Internally, as I said, we'll be building multiple physical prototypes of the machine. Got you. Like system, another system, another system, and perfecting each one. That's still internal to us. I'm assuming no revenue, frankly, from the 256 in this year. That happens in next year, we start taking orders for machines that we'll be delivering the following year. You might have some bookings this year? Yeah. We announced the beginnings of that already on the last earnings call. I think people want to see the fact that we're using semiconductors, the fact that the bill of materials we have is lower, the fact that we have milestones, kind of like scientific milestones behind us. They're signing on for not just the first, but hopefully more than the first generation of semiconductor-based chips. Yeah, we have a customer who's with us now for three generations of machines, including the 256- ...in terms of access to a machine like that. Yep. People saying we want to be first in line, we're trying to get their roadmap mapped out and delivered. Again, we want machines that turn on, they would do what they're supposed to do. Really important, as I mentioned, create the fault-tolerant machine. Create the fault-tolerant machine that discovers its own error, fixes its own error, and moves on. Understood. I got a lot of material here. Why don't we jump a little bit to growth rates and organic versus inorganic, which is a question sometimes I get on you guys. I did calculate for 1Q, it was a fair amount of four-function math, but I got to more than 100% organic growth. Could you talk a little bit about how investors should think about that, relative to your commentary and the segments of the business now because you are getting to be a more diverse company? Yeah. As I look at our organic business, meaning our quantum computing business, that first tile on the first slide I showed, that's what got us here, right? That's why we acquired Oxford Ionics for a little over a billion dollars. That's why we announced the acquisition intent of SkyWater for another almost $2 billion, or $1.8. All in, that's about $3 billion of investment dollars we, as a single company, have put behind our semiconductor roadmap. That's all in quantum computing, right? That is all about staying number one and establishing footprint, becoming de facto leader in compute. That's the organic growth you're talking about. Yes, of course, we're putting multiple billions to drive that innovation engine. The roadmap that I mentioned, the manufacturability I mentioned, and then scaling to meet demand at the same time. In our guidance for this year, I did indicate that I expect that business, computing, to grow 100% year-on-year for the year. There may be a quarter, maybe Q1 was slightly above. Depends on when you ship the machine, who you ship it to, when you turn it on, et cetera. That's how rev rec works. Over the course of the year, 100% growth continues. Even as the company gets bigger, that part of the company, which everyone else compares to- ...for us is still growing 100% year-on-year. Absolutely, we're. That's what brought us here, and that's what differentiates us, the lead we have in that. At the same time, being able to be the first company to be able to network computers together and actually do it in a way, in a quantum sense, is absolutely unique. We can add security to those that want to just begin with security, for example, to prepare for Q-Day. They might look for other things over time. The other products, and there are many of them, each of those tiles is many products, gives us more ways to monetize the quantum advantage or the quantum lead, I should say, we have as a company. You guys did demonstrate photonic interconnect and entanglement across fiber optic, I think, in April. When you get to what number of qubits will you require that to be part of your architecture? How many qubits are going to require- So in- ...photonic interconnect? Yeah. Technically, inside our computer, we believe we do not need photonic interconnects anymore, because, again, we're using CMOS to scale. Right. Our view and architecture and our plan is basically not to use interconnects all the way to 2 million. Okay. Yes, if we were using lasers and we were trying to scale the ion traps, or any other modality for that matter, then you need interconnects, to your point. We have interconnect capability already in-house, and we now use it for our networking. The networking goes over distances, and essentially either that form factor or something very similar to it that we also have, can be used to regenerate the signal to be able to have the entanglement happen over longer distances. Absolutely, the interconnect thing we do not need for our semiconductor roadmap up to 2030. Beyond that, who knows? Interconnects also can be used to connect quantum computers that sit next to each other to each other. Now I'm getting to territory where you become very unique. Think about the GPU example you gave, NVIDIA does what it does because it connects many GPUs together, not because it has one that's super powerful. Inside a data center, you have so many of them. If you can do similar things with just a couple of quantum computers, each quantum computer, at advantage, will outrun any other classical device of any form. Two of them working as one, it sort of starts to boggle the imagination a little bit. I don't want to get too far ahead of our headlights here. We want to make sure that it can exceed the expectations we set out there. With respect to powerful quantum computers, I did a panel earlier today on quantum's collision with cryptocurrencies in particular, RSA 2048 requires some number of qubits according to the most efficient algorithms that have come out recently. I'm curious what the company line is these days in terms of when RSA 2048 will likely be broken. This is a matter of some debate, some people think that while the hardware goes right, the algorithms will get more efficient at the same time. I was talking to Marco about this topic, I'm just curious what you guys are thinking in terms of number of qubits and fidelities required to break the internet, if you will. Well, look, I think that there are two plus two, really, but two mostly widely adopted encryption protocols. One is RSA and the, quote, "break-proof" thing was supposed to be RSA 2048. Under most of your cryptocurrencies, it's ECC 256, elliptic curve cryptography, but only 256 digits. Compared to 2048, it's thought to be as hard to crack as RSA because it's elliptical, not linear. Both of those require a certain number of qubits if you're running Shor's algorithm, which is the full-on algorithm that's been published 30 years ago and been waiting for a machine powerful enough to do it. We can get to that in some timeframe. Some of the algorithms that are being shaped, by others, in fact also, Google has talked about it, others have too, have that moment coming nearer and nearer. Exactly as you described it, the algorithms become more efficient and need less qubits, and the machines develop more useful qubits, and you have a fault-tolerant machine. Fault tolerance, I think, opens the gate to deploying one of those types of algorithms, maybe not on 10K, maybe on the one after, but in that area. Shor's might need another generation, for example. Again, I'm going to just use a independent party that has said when they think it'll be broken. The Department of War has said that it will be broken within three years. An adversary of the country has said they'll break it in four years. It's in that zone, and who gets there first kind of matters. I think it's a race not between us and any other quantum company, which is why I don't consider any of them as competitors, candidly. It's a race of nations, and I think that matters, as I said, to a lot of people. I have a couple questions from the audience. One, I'll just ask it verbatim. When will we see performance results of the 256 qubit system along with use cases utilizing the 256 qubit system? You kind of answered that, I guess, but I'll just throw that out there. Yeah. It's a great question. Every successive generation machine we're putting out there, remember, leverages the power of quantum mechanics. Yep. In quantum, as you know, it's not a one zero. It's not like a binary system that classical uses. You double the number of bits and it becomes twice as powerful. This is Moore's law. In quantum, it's like two to the power of N, not two times N. If you think about 256 versus 100 even, that we have this year, I said we're going to ship machines with 256 next year. I look at revenue as a really good indicator of the adoption of what you would put into the market. That should begin driving revenue next year for us, right? If we can do it sooner, we'll do it sooner. I'm not counting on that, and we'll be happy to be surprised to the upside if that happens earlier. That's when I would expect 256 realistically, then the year after that, hopefully the 10,000. We just have to, again, as I said, execute on our roadmap that uses mature nodes in existing foundries that know how to scale CMOS. Most of the science stuff is behind. Now it's just leveraging mature CMOS technology. That's fair. Relative to M&A, you guys have been fairly active over the last year or so. You do have a fair amount of liquidity. Should I think about the pace of acquisitions continuing at the same rate, or should we think about it as tapering a bit now that you have a more broad portfolio of assets across? Yeah. I hope it's not at the same rate. As COO. Well, yeah, I ran M&A for Cisco, you can do one a month, but you need a machine and you need to have an ingestion engine and all that stuff. Here, I think you do M&A when it's something that can be done faster than you could build it yourself, obviously, right? The time value and the return on that investment is different if you acquire it. What we've been acquiring is companies that have revenue mostly. That means that they have customers. They have a product and they have customers, and we can help them scale- by taking them through our channel or vice versa. If we do something, it won't be of the size or magnitude that we've had to do, because simply when we need to build a roadmap that scales rapidly and faster than any other player, frankly, for the needs of a national security of a country, then we do it. We do have financial firepower that we can bring to bear, as we've demonstrated with a $3 billion investment on our semiconductor roadmap going forward. If a customer says, for example, "Sure, it would be great if you had this other small piece in your portfolio," whatever. If we can build it, we'll build it. If we are pointed to a company that has a product, we will look at it. I think we can now choose to be much more selective, John, having built the main pillars of the platform. One of the markets, although you did highlight the importance of non-governmental markets, but in the last earnings call. The government defense market could consume your offerings pretty well as a bundle, I would think, in a lot of cases or as a portfolio of technologies, and I guess SHIELD/Golden Dome is hanging out there. I'm curious as to when you might see some unlock from or announcements coming on that project, just because I think President Trump has said he hopes to complete part one of the project in his term, which would be the end of 2028. Well, we've announced a few things that are, I would say, the early beginnings of that already. Yeah. I think we're well on the journey and to go back to kind of like in the five tiles that I described, one of them is around sensing and atomic clocks and space-based assets. Yeah. We've put together the building blocks of what we think that looks like. It's a question of, I think, also what are the elements of Golden Dome and the architecture of Golden Dome, and remember that that's not a simple thing to build. No. It requires many things. We don't make weapons, but we make sensing, we make some of the compute power that I mentioned. We deliver a lot of things potentially to that table. That said, the reason I point to sort of like the non-government business is because the adoption is already happening. In places like data centers that are starting to say, "We want to put a QPU next to the AI factory of GPUs." One set of those doesn't make a trend, but now I've seen three of those, and you start to see a trend emerge. The other aspect of that is also to point out that it's no longer being bought only by labs. Yes, labs will still buy it. Yes, universities, we want them to buy it, actually. I'll take government business any day, right? Even today, we are deploying machines, multiple machines, with a single customer even, and that's like the beginning of a flywheel effect, John. Where security, as I mentioned earlier, our business that is based in Geneva has been prolific in providing security solutions around the world, in the APAC rim especially, for the geopolitical reasons that you describe and can imagine, and also in Europe. In the U.S., humble opinion, we've had our head in the sand a little bit in terms of people telling us, "Oh, breaking encryption is 20 years away." When a few companies say it, Google says it, others say it, then no, it's like a fraction of that, security becomes suddenly relevant. We want to have the best portfolio, and be agnostic. If someone wants one flavor of security or a different flavor of security, we intend to be the one-stop shop for all of that. Okay. We're a little overtime, we started a little bit late, and I appreciate going over. If you had to pick one thing that you think the investment community, I guess my part of that, I don't know, what do they get wrong? What are the misperceptions about IonQ and/or quantum at large, Inder, that you see out there right now? Yeah, look, I think respectfully, I think no one gets anything wrong. I think because it's quantum, it's complicated. It's hard to understand. When will it be real, et cetera. Even Einstein called it spooky science, right? There's skepticism around will quantum ever happen? My philosophy is put up the numbers. Show the growth, show the revenue. A good metric is, of course, sources of revenue, but also revenue growth. To your point, organic growth. I think that there are other players, as you know, using different modalities, that as they scale, they become more expensive. As they scale, they become more energy consumption centric. As they scale, the total cost of ownership is higher. Not knocking any technology, trust me. When you begin with a cost advantage, and then you start deploying it and then saying to customers, "Well, if you buy the 10K machine, you won't have to replace it. You'll swap out some modules," like I said earlier. Right. That sticky relationship, and then getting them to buy into the roadmap, and then you execute to that roadmap, creates what I learned at Arm or Cisco, at other companies, is that's how you become de facto leader. Right now, it's all about grabbing market share as fast as possible because no one is saying you can't. There's no incumbent to dislodge. It's not Cisco trying to beat Nortel and fighting over price. This is like grow fast and then feed into it. That's why I'm covering the space. All right, well, that's a great finishing note. Thank you, Inder. Thank you, Hanley, for taking the time today. Thank you, John. Really enjoyed the discussion with you, as always. Yeah. Bye-bye.