Good day and thank you for standing by. Welcome to the Quantinuum Third Quarter Earnings Conference Call. At this time, all participants are in a listen-only mode. After the speaker's presentation, there will be a question and answer session. To ask a question during the session, you will need to press star 1 1 on your telephone. You will then hear an automated message advising your hand is raised. To withdraw your question, please press star 1 1 again. In the interest of time, we respectfully ask that each analyst limit themselves to one question. Please be advised that today's conference is being recorded. I would now like to hand the conference over to your first speaker today, Shubh Mukherjee, Head of Investor Relations. Please go ahead.

Good afternoon, and welcome to Continuum's third quarter of 2026 earnings call. With me today are Raj Hazra, President and Chief Executive Officer, and Nitesh Sharan, Chief Financial Officer. Before we begin, I'll cover the Safe Harbor Statement. Today's remarks contain forward-looking statements within the meaning of the Private Securities Litigation Reform Act. These include statements about our technology roadmap, commercial expectations, and 2026 and 2027 outlook. Actual results may differ materially. Please refer to the risk factors in our earnings release issued earlier today and our most recent Form 10Q filed with the SEC. Statements speak only as of today and we undertake no obligation to update them. In addition, during today's call, we will be discussing certain non-GAAP financial measures. Please refer to our earnings release for additional information regarding these non-GAAP financial measures and reconciliations to the most directly comparable GAAP measures. With that, Raj, over to you.

Thank you, Shiv, and good afternoon, everyone. Quantum computing, the largest market opportunity across all quantum technologies, is moving more rapidly from promise Thank you very much. and most interestingly, enterprises are investing now in the expertise, partnerships and applications needed to put quantum to work. As I said last quarter, the central question is, who will lead the creation of this new era of computing? Our view of what it takes to lead and a leader is very clear. Leadership starts with building quantum hardware that can deliver accurate results at the speed customers need. Customers need to solve business problems, which means having answers that they can trust in a timely manner. Efficient, fault-tolerant quantum computing at scale is the foundation of that trust. But having the best fault-tolerant quantum hardware alone is not enough. Leadership also means integrating quantum with classical computing and AI and giving developers the tools to build applications that solve real business problems. And that is what Quantidian is building, a full stack platform designed to deliver useful quantum computing at scale. Our third quarter marked significant progress across all three parts of that strategy. First, in product technologies, we achieved world record logical fidelity on Helios, our commercial system, and significantly de-risked the hardware and error correction capability that will be required to deliver universal, fully fault-tolerant quantum computing with our Apollo system in 2029. In the area of software, we announced a multi-generational Nexus roadmap designed to accelerate application development on today's systems and enable those applications to scale well into the fault-tolerant era. And perhaps most importantly, in commercialization. We added new customers, deepened customer engagements, secured the largest cloud agreement in our history, and expanded access through major cloud and enterprise ecosystems. These are not three separate initiatives. Together, these efforts position Quantinium to accelerate commercial value creation before the end of this decade, Delivering universal, fully fault-tolerant quantum computing by 2029 while building the software, applications, and the ecosystem customers need to use it today and in the future. Let me take a few minutes to walk through the details of our progress this quarter. I'll first focus on product technology and supply chains. The path to false tolerance requires two things advancing together. Accurate hardware, including an architecture that can scale qubits sufficiently, and scalable error correction. We significantly advanced both in the third quarter. While not just advance, we demonstrated results that illustrate our significant lead in false tolerance hardware with the rest of the industry. More of this in a minute, but first let me do a quick update on our systems roadmap. SOL remains well on track for delivery in the second half of 2027. And given our step-by-step de-risking approach for these systems, we have greater confidence in its timeline today than we had for Helios at the same stage of development. For Sol, following the first trap's return from fabrication that I spoke about in the second quarter, in the third quarter we validated Sol's entire control system for moving qubits in the two-dimensional trap at full product scale. That demonstrated the control and movement of ions required to validate our QCCD architecture in Sol and machines beyond like Apollo. We are now in full system integration and assembly on Sol and on track to complete that in the first quarter of 2027. Next will come some validation and verification, including running real jobs before commercial availability in the second half of 2027, as planned, as promised. Sol is expected to be the world's first commercial quantum system with 100 logical qubits. targeting a logical error rate of 10 to the power minus 5, that is, approximately one error in 100,000 operations. Apollo, our universal fully fault-tolerant quantum computer, is progressing to plan through its design phases and remains solidly on track for 2029. Our teams are progressing through key subsystem designs, including validating our cryogenic cooling system, which is critical for a machine at Apollo scale. We are also building the supply chain we will need to scale production for Apollo and future machines with deep longstanding partnerships with Global Finderies, Infineon, Honeywell Aerospace, Monarch Quantum, and others. These partners bring deep expertise early into the design of critical components, helping us reduce complexity, improve reliability, and make production more repeatable. In September, we finalized our CHIPS R&D award following the previously announced letter of intent, which will help accelerate our supply chain efforts. Quentinia was the only trafting company selected for that award. Beyond the roadmap, as I had promised, I'll get back to our industry-leading result in false tolerance, which is very, very exciting, not just for us, but for the industry at large. In Q3, we introduced a scalable end-to-end logical architecture that brings together tailored AI decoders, all-to-all qubit connectivity, high-fidelity physical operations, and hardware-level error corrections. These capabilities, working together to improve computational accuracy, providing the foundation for universal, fully fault-tolerant quantum computing at scale, is essential for leadership. The end-to-end logical architecture of flexibility also allows us to test, refine, and combine different error-correcting codes, balancing our accuracy, efficiency, and speed with for an increasingly larger set of diverse demanding business applications. We have already implemented this logical architecture in Helios. In September and October, two different error correction codes running on that architecture on Helios delivered back-to-back word records for logical fidelity. October's result reached 99.9995%, equivalent to five errors in a million operations. That error rate, just to put it in perspective, was 10 times lower than our own September record and 100 times lower than Google's, which had published the previous best results amongst the various hardware modalities. Critically, these results point to significantly lower error rates with lower error correction overhead. Each two-step increase in the error correction code distance reduced logical memory errors by a factor of approximately 12, compared with roughly two in Google's published superconducting results. This stronger error suppression means that we will need 30 times fewer physical qubits to reach target logical error rates, reinforcing another significant scalability advantage of our trapped-iron modality and our QCCD architecture. Today, we are very clearly leading in scaling accurate high-fidelity Next, let me move to software. Our Nexus platform had a very significant quarter of progress as well. Nexus is a platform as a service that brings quantum application development and deployment into one environment. and converts these hardware advances into capabilities customers can use to build meaningful applications and deploy them on hybrid quantum classical infrastructure. Today, 215 organizations are building on Nexus, up nearly 20% from the 180 I had quoted last quarter. More than just the number, total job submissions increased tenfold in the 12 months through August 2026 compared with the preceding 12-month period. Today, we announced a significant advance for Nexus, a multi-generational roadmap to develop this platform into what we believe will be the industry's first platform-as-a-service designed for the fault-tolerant era. It aligns software capabilities with each generation of our hardware through Apollo and beyond, giving customers predictability for application development and deployment. This is a fundamental aspect of accelerating the adoption of useful quantum computing. We are already delivering the foundations of that roadmap. Our September release of Nexus extended our Guppy programming language with tools for programming logical qubits and applying supported error correction codes, including our own optimized codes and third-party codes, thereby reducing the need to rewrite application logic for each type of error correction code. Developers can now test programs in emulation, execute supported programs on Helios, and customize their approach while benefiting from advances in error correction. The upcoming December release will expand capability to deal with hybrid workflows along applications that have both classical and quantum code to respond to quantum results as soon as they become available, thereby making the performance of the systems more efficient and more effective. It will also add supported error correction codes and execution analytics to help developers optimize further the performance they get on our leading hardware. As our hardware and software advance together, the applications, the workflows, the expertise customers build on Nexus will become invaluable. That gives them a compelling reason then to build with Quantinium today and carry that investment forward with Quantinium into the fault-tolerant era. And last but not the least, let me talk about our progress with commercialization. We are adding new customers, developing applications around valuable customer problems, securing commitment across hardware generations, and expanding access through established cloud ecosystems. Let me give you just a few examples of new developments this quarter. Our work with Johnson & Johnson applies quantum computing and AI to the creation of new cancer therapy materials with the potential to accelerate drug development and enable more effective treatment sooner. With GE Vernova, we are advancing research into superconducting materials in order to build smaller, lighter, and more efficient electrical infrastructure. A new collaboration brings together Abu Dhabi Maritime Academy, part of the 80 Ports Group, AI-powered logistics company Quincas, and Quantinium to explore quantum AI and high-performance computing for complex maritime and logistics applications. We also signed our largest ever cloud agreement, a multi-year, multi-generational commitment from a leading global financial institution, for access to our systems through Apollo using Nexus, supporting fault-tolerant application development and deployment across multiple core business functions. In addition to this, we are also expanding access to our infrastructure through cloud and enterprise environments that already serve millions of developers. Building on our announced partnership with Oracle last quarter, We are advancing commercialization of the hybrid quantum plus classical capabilities in the Oracle Cloud Infrastructure. We jointly launched a beta program for select customers to explore hybrid quantum and AI applications in the areas of defense, financial services, automotive, manufacturing, and general research. We demonstrated our joint offering at Oracle AI World, and Helios deployment into the Oracle Cloud Infrastructure AI Data Center remains well on track for full deployment later next year. In addition, we also signed a new agreement with a leading hyperscaler serving millions of developers and enterprises worldwide. bringing our hardware and software into that ecosystem significantly broadens our reach and is expected to accelerate the integration of quantum into existing AI high-performance computing and enterprise workflows. Alongside this enterprise and cloud momentum, we are seeing continued government investment and strategic focus in the United States and internationally benefiting Quentinium. In the United States, we continue to progress through the DARPA Quantum Benchmarking Initiative, or the QBI initiative. We are actively discussing a QBI stage C scope of work with DARPA as we continue to execute on our stage B contract. Separately, with the USG, we are participating in four winning university proposals in phase one of the Department of Energy's Genesis mission. These draw on our systems and expertise across algorithms and nuclear reaction design, quantum chemistry and new material creation, and AI-driven design of next-generation quantum processors. Beyond the United States, internationally, our second Helio system is on site in Singapore and is being assembled with commissioning on track for next spring. Next week, we are commissioning an upgraded computer at Rekind in Japan as part of the world's most advanced hybrid quantum supercomputing system of its kind. We are also discussing potential deployments with several other governments at this stage. Together, these developments show how our commercial strategy is successfully accelerating the industry's adoption of quantum computing. In particular, developing applications around customer needs, securing long-term commitments to our products and services, and building the distribution and deployment pathways that are needed for broader adoption and scaling. Lastly, on the ecosystem front, supporting our technological and commercial progress, we are growing the broader ecosystem around our full-stack platform. First, we worked with Sandia National Laboratories and NVIDIA to introduce QOPS, A new performance benchmark that gives customers a capability that they were lacking. That capability is a credible, repeatable way to assess the size of a useful computation a system can execute and how quickly it can execute it. We are building alignment amongst partners in the ecosystem and government stakeholders towards an industry standard around co-ops that will help customers compare systems and make informed purchasing decisions expediently. We are also growing the community that develops applications and brings new capabilities to our platform. We've now grown QNET, our community providing developers and researchers with resources to do more with our platform, to over 800 participants and added eight startups New startups to our Quentinian Startup Partner Program. Finally, we are working closely with our government and commercial customers to grow the quantum ecosystem. A case in point is a quantum challenge we designed and ran in Singapore, supported by Singapore's National Quantum Office. Over 200 teams submitted proposals, and the winner will be announced later this month. These initiatives help build practical expertise and applications around our technology stack. These efforts reinforce one another. Clear benchmarks help customers evaluate systems, developer programs build expertise and applications, and public-private programs create pathways to adoption. Our third quarter execution increased our technology lead. With deeper customer commitments and expanding routes to market, we believe we are positioned to turn that leadership into momentum, into accelerated commercial growth, and further extend our industry leadership. Our momentum across technology and commercialization has been accelerating through the years. As I look at 2027, I'm more confident of our opportunity and our ability to capitalize on it than I was when we spoke last quarter. With that, over to Nitesh to take you through our financial results and outlook.

Thank you, Raj, and good afternoon, everyone. Let me start by building off of Raj's thoughts. Our goals were clear this year. Through the IPO, we shared an aggressive and credible plan, grounded in reality. Nitesh Sharan, Rory O'Byrne, Robin Schulman through the engineering required to be the first default tolerance to scale manufacturing with a diverse and proven supply chain and to execute the go-to-market strategies that we expect will unlock the greatest value for customers. We are fueling this from a position of capital strength that sets the foundation for disruptive growth and outsized returns, and that continues to translate into near-term financials that outpace even our own expectations. Let me illustrate that in the Q3 results. Revenue for the third quarter was $9.1 million, up 233% from the prior year. Growth was driven primarily by strength in our cloud platform, along with research and services. Revenue remained well diversified across customer types and geographies, with government representing the majority and enterprise also a significant contributor. Customer concentration also continued to improve in the quarter. Let me emphasize a point that is important to understand when contrasting our results with others in the quantum sector. Every dollar of revenue report is directly related to quantum computing. As Raj noted, that is the grand prize we are focused on as the largest part of the quantum ecosystem, and it is what will catalyze the world-changing breakthroughs that we see on the horizon. Bookings are a forward indicator of future revenue. In the quarter, we recorded bookings of $98 million, including the Oracle deal we announced during the last earnings call. These deals were across government and enterprise and served several of our core industry verticals, notably financial services, chemicals, and pharma. Our 2026 year-to-date bookings now stand at approximately $104 million, and we expect to achieve more than $120 million in bookings over the full year, setting us up well entering 2027. More on that when I get to the outlook. Moving to the cost structure. Cost of sales was $3.2 million in the third quarter, including roughly $0.7 million of stock-based compensation. Non-GAAP gross margin, which excludes stock-based compensation and other non-cash charges, was 74.6%. The year-over-year improvement in growth margin this quarter was largely driven by revenue growth. This line will move with revenue from quarter to quarter and be heavily influenced by product mix. The gross margins this quarter reflect a heavy mix of cloud business. Over time, we will increasingly emphasize design for cost and manufacturing for scale, and we have a great set of supply chain partners working closely with us on these objectives. Research and development expenses were $97.5 million, including $27 million of stock-based compensation, up from $42.4 million a year ago. We are investing at every level of the technology stack, our leading quantum hardware, the Nexus developer platform, and in end-user applications. These capabilities amplify one another and are increasingly fueled by direct customer engagement, providing direct clarity toward unlocking end-user value. Sales and marketing expenses were $14.9 million, compared with $4 million in the prior year period, and included roughly $4 million of stock-based compensation. We are building our government, commercial, and consulting direct sales engines. Our indirect channel partners will enable us to further leverage this line item to capture more of the market. and we are enabling all of that through targeted high ROI marketing spend across brand development, demand generation, and lead execution. General and administrative expenses were $22.1 million compared with $5.3 million in the prior year period and included $5.5 million of stock-based compensation. The increase reflects the investments required to operate as a public company including finance, legal, compliance, and operations. We continue to invest in our controls infrastructure from the foundational IT to modern applications that improve overall process and administration. In total, stock-based compensation was $37.2 million in the quarter. Adjusted EBITDA, which excludes stock-based compensation, depreciation and amortization, and other one-time and non-cash items, was a loss of $80.5 million compared with the loss of $45.3 million in the prior year period. As a reminder, At the GAAP net income level, we separate the economics attributable to legacy shareholders who have not yet exchanged their shares into the post-IPO public entity as non-controlling interest. Our total GAAP net loss for the third quarter was $117.9 million. Of that, the net loss attributable to Continuum Inc. was $17 million, or a 46 cent net loss per share. On an adjusted basis, Non-GAAP net loss was $75.2 million, and non-GAAP EPS was a loss of 29 cents per share. These measures are calculated across our full shareholder base, and they exclude stock-based compensation, depreciation and amortization, and other one-time and non-cash items. Turning to the balance sheet, we ended the quarter with over $1.9 billion in cash and equivalents with no debt. We have a strong balance sheet. with what we believe is more than sufficient cash to fund our technology roadmap and invest in the critical capabilities that we expect to fuel the path to profitability. Maintaining a strong balance sheet is a critical priority for us and a source of strength. Before turning to guidance, let me describe the accounting related to one of our big wins in the quarter, the TIPS Award. As we noted, we were awarded $100 million from the Department of Commerce to help develop the US quantum ecosystem. As part of the structure of the deal, Nitesh Sharan, Robin Schulman Just note that this is an incremental investment to fund our pathway to fault tolerance, providing incremental capital protection and diversity of funding sources that may not be fully reflected on the cash line. So turning now to our outlook. Our execution velocity and commercial momentum have picked up through the year, and this gives us higher conviction in our near-term revenue. We are raising our full-year 2026 revenue outlook to $31 to $34 million, from $28 to $32 million previously. It also makes us more bullish on 2027. Based on our current bookings and backlog and our assumptions around contract timing and revenue recognition, we continue to expect revenue growth of more than 100% in 2027. That growth now builds off of a higher 2026 pace. So put simply, our expectations for both 2026 and 2027 are higher today than they were just last quarter. In closing, our momentum is clear. Given our early stage nature, things will not move linearly. Each quarter will have its own unique element, but we expect the longer-term trajectory is exponentially upward. The technology leadership Raj described is converting into commercial traction, and that traction is showing up in our financial results with accelerating momentum. We believe we are building a diversified business with an attractive long-term financial profile that we expect to deliver sustained growth, profitability, and high returns. With that, we'll now open it up for Q&A.

Thank you. At this time, we will conduct the question and answer session. As a reminder, to ask a question, you will need to press star 1-1 on your telephone and wait for your name to be announced. to withdraw your question, please press star 11 again. We respectfully ask that each analyst limit themselves to one question. Please stand by. Our first question comes from Krish Sankar from TD Cowen. Please go ahead.

Krish Sankar from TD Cowen. Hi. Thank you for taking my question and congratulations on yourself. Rajeeb and Nitesh, you know, the calendar for outlook improvement from three months ago, Just wondering, what drove it? Is it like cloud, enterprise, government, R&D? Is it in the U.S., overseas, any specific industries? And just like extrapolating it onto the 2027 revenue growth of 100% of a higher base, what kind of specific verticals are driving the growth in 2027 too? Thank you very much.

Sure. Thanks for your question, Krish. On the 2026 raise, it really is broad-based. You know, we are seeing great progress on the cloud business. We're seeing it across governments, not just in the U.S., but across the world and in enterprises across some of the verticals we highlighted in the prepared remarks. And as you go forward into 2027, certainly we were excited last time to raise that outlook as well, talked about Nitesh Sharan, and many more. Our sales team and the pipeline that they're building, we are seeing it across all of the pieces. And so that's why we had enough conviction, which still, in our view, reasonable amount of conservatism and prudence in our planning that we would be able to raise 2026 and also give you an early view of an increase in 2027.

Thanks, Nitesh.

Thank you.

Thank you. Our next question comes from Vivek Arya from B of A Securities. Your line is open.

Hi, thank you. I think we're taking our questions. I wanted to start just on the error correction side. You said the October result needed about 30 times less fiscal qubits to hit your target logical error rates. Could you discuss how that changes how you design Apollo and Lumos? and particularly how we should think about LUMOS in terms of size and that logical qubit count longer term. Thank you.

Thanks for the question, Rebecca. So as you stated, there were two really big things we announced in the error correction. One was reaching that absolute five errors in a million operations. But the second big thing was the error correction efficiency where we achieved an exponential reduction in error correction for a reasonably small amount of increase in the number of physical qubits that are used to create a logical qubit. And that comes from the architecture that we have, our modality, and the very strong all-to-all connectivity that we support in our particular implementation of two-dimensional QCCD. As we think of that going forward, there's two things. One is, obviously, these results, even achieved on Helios with a universal gate set, give us very high confidence that as we move to Sol and Apollo, and the physical qubits themselves get better, that we are in very good de-risk shape to achieve Apollo's targets that we put forth and achieve them at full fault tolerance across all operations of the universal gate sets. We have not made public yet kind of the logical specs on LUMOs, but what we expect is that LUMOs to continue the physical qubit improvements to some extent, but most importantly, aggressively adopt these error correction techniques that we have now proven are scalable from the scale of helios to a million plus qubit machine that we will effectively continue to run that exponential and therefore deliver essentially the best error rates in the industry for the smallest amount of system size. And that is very important from a reliability standpoint, cost standpoint, and simply manageability standpoint of these systems.

Thank you. And then on your question, Largest EverCloud deal, the multi-generation commitment through Apollo. What's driving customers to sign multi-year commitments before fault tolerance arrives? How is their spend trending? What kind of conversations you're having with them? And are they moving from research and exploration projects into production use cases?

Great question, Rebecca. There's really two drivers. One is being engaged early actually having a use case roadmap that tells them what can be achieved at what portions, with what portions of unlock from machines like H2, Helios, Sol, and Apollo. So the first thing is conviction that their business relevant use cases can grow. The second is the predictability of our roadmap now and the confidence that if they invest in technologies or application building input in, say, the sole timeframe, that that carries forward in a timely manner to a machine called Apollo, which isn't maybe at some point, but it's definitely in 2029. The nature of our conversations very clearly, as you see these kinds of deals that we are seeing both extend in time, expand in size of dollars, and extend over multiple generations, that came from the early engagement which was kind of more proof of concept to now it is almost pre-building of the application that in some cases will unlock at sole at a full production scale or Apollo. But that path of getting ready so there is not a delay between when the technology is available and when the applications are deriving business value, that gap is minimized. So we are very much from the Thank you.

Our next question comes from John McPeak from Rosenbat Securities. Your line is open.

Hey, guys. Congratulations on the progress here. Question is the same question I had last call, actually. I'll ask it again since one quarter has passed. The de-risk since the last call relative to the transition from Sol to what's likely the first fault-tolerant machine, Apollo. Maybe you could just give me the key points there, what you've accomplished. Because you did the crossing with Helios, and now you have the chip, I guess, running, and you're crossing ions in the trap. So just anything there would be great, and then I just have one follow-up.

Sure. Maybe I can answer that. And even though you asked the same question, I will give you a different answer and a better answer this time. Because we have advanced. We have advanced. And what we have done is, last time I told you that Sol is the building block for Apollo. Why? Because it's the same two-dimensional QCCD. Sol chip came back. And the big thing in Helios, we had one crossing. So you had to manage only one pair of qubits crossing each other. of course in Sol with 198 qubits we have near 100 pairs of qubits to actually move around the chip and pipeline them and that's a complex control system but it is the heart of what makes a 2D trap for Sol and Apollo work and that we have proved on real silicon at the 192 qubit scale and then simulated it beyond to give us confidence that the basic heart of the Sol's Thank you very much. The scale that Apollo is, you have to ensure that that system is balanced to the load and the larger chip that Apollo is. The second thing is, of course, advances in what we call the laser optic system because Apollo carries more qubits. Even though you can move the qubits around, as we've proven with Sol and Apollo, the ability to actually show that you can have beam formation, beam delivery, and working with our partners like Monarch Quantum prototyping those, to us then being able to say, yes, we can see Sol working in 2027, and by 2027, 2028, we would have de-risked our path significantly through proven subsystems into Apollo, which then we integrate. That has been our strategy. Prove out the most critical subsystems first in emulation, then in silicon, test silicon and prototypes, and then build them and integrate it and build the whole system together with far less risk than waiting for everything to come together at the end.

Excellent. And then with the recent novel error correction that you're using, triangular error correction, and the results that you're getting there, Is there potentially upward pressure on the logical error rates? Because on the logical error rates in Sol, Apollo, and Lumos?

Actually, with the error correction results that we got, that we have said, shown on Helios, which is five errors in a million operations, which is five times 10 to the power minus six, we have de-risk. Thank you very much. physical qubits themselves are better. That gives us a significant boost in being able to do more error correction per round than we've even been able to do on Helios, which is about six times better than the published result on any other modality. Does that make sense, John?

It makes sense. When I say upward pressure, I guess it's upward pressure on fidelity is really what I meant, not on error rates. And then The final one is, you know, what I think some of other modalities are struggling with is the non-Clifford gates and the distillation magic states and all that and the overhead that's required. So you guys are going to give us at some point soon, I think, some information on non-Cliffords and universal gates, I guess, right, with the new error correction codes.

Absolutely. In fact, I want to be very clear. We have already, and we were the first to demonstrate a universal gate set. So our results are universal gate sets and a distribution of non-Clifford gates along with Clifford gates. We define universal full fault tolerance as what we need for the era of full fault tolerance. So it's not about just doing the Clifford gates and not the non-Clifford gates. So that is already in our results. If you look then, if you're trying to compare, for instance, where our magic state or some of the components of end-to-end error correction are, we have results. In fact, we published it in 2025 where we have slightly better results than what Google published, which was their 4DX reduction in magic state errors. with a significantly higher acceptance rate. And what I mean by that is the way you do magic state distillation or cultivation is you have to try over and over and over again. And sometimes they work and sometimes they don't work. And so the acceptance rate is how accurately you're doing the magic states versus doing them and throwing them away and then only keeping the good magic states. and what that practically means is if your acceptance rate is high, then you're using the physical qubit resources very efficiently and you're also cutting down the time required to get these magic states built. So all in all, we enjoy this end to end logical error rate advantage due to a number of factors, one of which is best in class magic state fidelity and magic state utilization if you think of it as space time utilization which is really how well we can scale as we get to larger qubit counts.

Great. I'm really looking forward to so on. Apollo, thanks guys.

Thank you.

Thank you. Our next question comes from David Vogt from UBS. Your line is open.

Great. Thanks, guys, for taking my question. Maybe, Raj, I just wanted to follow up on John's question regarding the Apollo roadmap. I appreciate all the color on Sol, where we're going to be from a validation and a commercial availability perspective next year. But can you help us understand from a simulation, from an emulation to silicon emulation roadmap, What other sort of subsystem designs or what other roadmaps should we be paying attention to as we move forward with Apollo outside of sort of the build-out of the supply chain in terms of tracking the progress? Obviously, I know Sol is critical next year, but just want to keep an eye on what else is being focused on from a subsystem perspective for Apollo as you go to a fault-tolerant solution and then have a follow-up.

Absolutely. The first major subsystem, as you see, and Apollo being a larger version of Sol, is the transport subsystem that we've completely now validated at scale. It is important to get that right even at scale, and why does scale matter? Because all of the pipelining and therefore the functioning, the time-based performance of the chip, the speed of the chip depends on very efficiently being able to move these qubits, not just being able to move them. So that is the first subsystem, is the transport subsystem. The second subsystem, as I mentioned, was essentially the laser, the optic subsystem that then interacts with the qubits on the chip that are controlled electrically in their movement. And that, we made significant progress along with our supply chain partner, Monarch Quantum, that we've been engaged with for now over two years in building out the exact, what we call the beam delivery and the beam concentration and the beam localization capabilities so that you can actually have those moving qubits being brought to the computational zones, which are many more in Apollo than in Sol, and being able to accurately use them to give us the kind of fidelity that we enjoy today in Helios and see moving forward in Sol and Apollo. Third is the cryogenic system. Again, we don't need cryogenics to actually manufacture qubits like superconducting folks do. We need the cryogenics, which is, I would say, gentler cryogenics in order to create a good vacuum. But at Sol's scale and then at Apollo's scale, it's important that we have a very efficient cryogenic system, which is not just a technical matter, but it's also a closed loop so that it does self-recovery of the cryogenic material, which is helium, and that impacts the practical TCO of the machine. So a percentage of two of Thank you very much. will be able to use Apollo's hardware and offer developers who bring in just their application a choice of error correction codes that they can link in like a library. And in some cases, if they don't know which ones, the machine will actually, using some level of machine intelligence and understanding the application, suggest what may be the best to use for that application and the resources it needs. So a very sophisticated environment in software that aids developer productivity and effectiveness enabled by very high performance, fully fault-tolerant quantum hardware.

Great. Now that's incredibly helpful. Maybe just a quick follow-up for Nitesh. I know, Nitesh, you're not in the habit of giving sort of longer-term outlooks, but just trying to understand when I look at your Outlook for next year that you expressed confidence in and the bookings number that you alluded to earlier for this year, I think 120 million bookings for the full year. How do we think about the way bookings start to filter in when Sol becomes commercially available later in the year? Like what's the lead time? When do you expect sort of commercial engagements to really start to impact those metrics and then ultimately convert it to revenue? Recognizing that obviously still have some validation and integration work to do, but just want to get an understanding on how the model works in that sense.

Yeah, maybe I'll make two points. One is that we're seeing some of that commercialization happening now. In fact, some of the contracts that we have actually have a graduation of how people will have their journey going forward. And I think to an earlier question Raj talked about, a lot of these customers that we talk about continuing with us, upsizing their contracts, renewing, oftentimes there's a long journey of algorithmic development, how these use cases will build and build upon one another. It goes further out beyond Seoul into Apollo as well. So I think you're certainly going to see greater emphasis on that as we go through next year in terms of the opportunity. But I think if underpinning one of your questions, how does this model play out over several years? We're very early in this opportunity, and we do see the market growing substantially, especially when you get to fault tolerance. And the opportunity in terms of the value of these systems grows step function, and so that'll be catalyzed in our revenue and our bookings. And ultimately, we're talking getting towards this billions of dollars of opportunity in the not-too-long horizon, capturing a meaningful part of the market because, again, our focus is quantum computing. We are leading. We think we have a multi-year lead against competition. The engagements we're having with enterprises, governments, and so forth is showing that back to us that we're doing the right thing. We're trying to make our systems more readily accessible to folks and that's happening not only through the hardware side but also through the Nexus side as Raj just talked about. So we do think all of this leads to ultimately we're in hyper growth mode and the numbers that we're laying out for the near term, the next 12 months and so forth, it will move to exponential growth as you get into the latter part of this decade and we start to move from Helios to Sol to Apollo and to Lumos. Great, I'm sure.

Thanks, Nitesh. Thanks, Raj.

Thank you. As a reminder, we ask that each analyst limit themselves to one question in the interest of time. Thank you. Our next question comes from Joe Moore from Morgan Stanley. Please go ahead.

Great, thank you. I wanted to ask about the DARPA stage C. You talked about the progress that you're making there. I guess to the extent that some others did pass through stage C, just wanted to give you a chance to address if that means anything. And certainly, we're confident that you're in a strong position. And I guess to the extent stage C is about whether the system can be built and operated as designed, what key pieces of LUMOS do you need to demonstrate to get there? Thank you.

Hi, Joe. Thanks for the question. Just the facts, what DARPA has said is, Stage C is a phase set of awards, depending on which companies were in their Stage C commitments and delivery. We continue to work with finishing up on our Stage B and our active discussions with DARPA on our Stage C scope of work. We see the DARPA as an important validation confidence builder, but we are also engaged very deeply, as Nitesh talked about and I talked about, with enterprise customers. Our view of what these systems need to be, what error rates they need to reach, what scale they need to reach, and how we build them is built partly on the conviction of our own technology view and partly on the conviction that our customers say these are the right kinds of systems, these are the right kinds of system capabilities. So we're moving down with Apollo and Lumos on our roadmap very confidently as we de-risk critical components of that fully fault-tolerant computing stack and in parallel continue to work to finalize our state C scope of work based on this particular de-risked, highly credible roadmap that we put forward for the industry.

Thank you.

Thank you.

Our next question comes from Mark Mortler from Bernstein. Please go ahead.

Thank you very much for taking my question. Congratulations on all the progress you're making. It's really quite impressive. I'd like to switch gears a little bit and talk about the hyperscale of partnership you have and likely maybe you'll get more of these going forward. I'd like to understand some more on the economics of this. Who's selling the usage? How is it being monetized? Are they reselling? Just the compute or the software platform, is it consumption agreements or are they paying something up front? Any code you can give so that we can get a better understanding as the scales of what the opportunity and monetization is and what the use cases they're using it for. Thank you.

Thanks, Mark.

I'll start and then Raj can add on. I mean, so first, yeah, we're excited to have continued growth, both direct customer activity, but also through these channel partners. Excited last time to disclose Oracle and then excited now to talk about Hyperscaler. And to your point, yeah, there's a lot of opportunity for us. What's most important is that the arrangement is a strategic one, that we're providing customers with differentiated service. So ultimately for us, it's a go-to-market access play. We are providing the full stack capability. We talked last time about how Oracle were actually the first to put a quantum system in a data center, and it's going to be natively built up. and so forth here. It's not exactly that construct, but it is another one where we're able to have joint go to market. The revenue model is openly we're still working through some pieces, but there is ultimately that if you have an end customer of the hyperscaler that wants access to the best breed of quantum system out there, they can access our systems and then we will obviously accrete the revenue associated with that.

Yeah, the only thing I'd add, Mark, is The hyperscaler models are different depending on how the integration happens. And we provide the flexibility through our software stack and our business model to be able to do that. But in the Oracle case, as we said, Oracle bought the system and our full stack along with it, monetizing it. We may have future opportunities to build on top of that as we bring additional customers in. In other cases, we are a full stack organization. Think of it as a full stack capability within that, and we get paid for the value we unlock for customers that are already on that hyperscaler or come to that. The key point is twofold. One is we are seeing increasing number of end customers already on the public cloud wanting to integrate AI, HPC, and quantum workloads, which is what's driving these hyperscalers. to look aggressively at incorporating the QPU along with the GPU and the CPU in the data center. And number two is we are monetizing our full stack. This is the hardware, our Nexus layer, our value-added libraries and applications and chemistry, and not just commoditizing the hardware in these kinds of partnerships with our hyperscalers.

Extremely helpful. Thank you.

Thanks, Mark. Thanks.

Thank you. Our next question comes from Vijay Rakesh from Mizuho. Please go ahead.

Hi, Raj and Nitesh. This is Rahal covering for Vijay. Congratulations on the strong results. And yeah, I was wondering how many of those organizations that you have on Nexus could become on-prem clients over time? And is there something specific that will drive their conversion from cloud to on-prem? Is there some like usage level or maybe a customer size?

It's a great question. I mean, the first thing is, you know, when customers get on Nexus, it's a highly sticky environment because you're choosing the programming language, you're binding your applications to specific libraries, and it's a very sticky environment that then you want transportable on whether you want it in a private cloud or you want it on a public cloud. And the private cloud is the on-premise cloud, right? So we have architected our full stack to be able to do that. The organizations, we have about 215 organizations today on Nexus. And a large majority of them are using the hardware in the Quentinium Cloud as we move forward. But some of them are already on a private cloud. For instance, we have Singapore, for instance, as a customer, Japan as a customer, that are actually using a private cloud in Nexus on their private cloud where they have the hardware on their premises. As we move forward, what I see is how the work played out for enterprise workloads. You will see some of the workloads and some portion of the workflows happen in the public cloud, but some stay on the private cloud for data governance, data security, and simple economics of moving large amounts of data back and forth. And we have customers across all of those segments which is why we have a good set of inputs as to how to create a flexible architecture that essentially says we meet you where you need the service.

I see. It's very helpful. Thank you. And maybe one follow-up. When you think about 27 growing over 100%, how much of it is already covered by Backhook and Pipeline? And maybe... What portion of it is an Oracle deal or maybe other big deals that you have secured?

A good amount is covered. I won't give you too precise an answer, but I'll just say that we are feeling like we're in really good possession. Obviously, there's still details that play out of when actually revenue plays through, but the line of sight we have, including the existing backlog, gives us a fair amount of confidence that we can deliver at least 100% next year. and it is, Oracle's important but there are definitely other deals. We talked about a few of them in the earnings call this time. So it's broad-based. I think I answered an earlier question about our revenue is really a nice composite of governments across the world, national labs, universities and enterprises across industries and our bookings to revenue profile for next year reflects sort of that diversity as well.

Thank you.

Thank you. Thank you. Our next question comes from Quinn Bolton from Needham and Company. Please go ahead.

Hey, guys. This is Shadi Milalian for Quinn. Thanks for taking the question. I guess on the agreement with J&J, I know in the past you guys have talked about outcome-based IP being a potential revenue stream. So I was just curious to hear if that's something that can come out of this agreement. Thank you.

Yeah, we see it certainly monetizing our full stack. So where we have... We are working with some chemistry customers. For instance, I'll give you one example where we have IP around a workflow to deal with metal organic frameworks, which can be used from anything from carbon sequestration to delivery, creation of drug delivery elements. That's an IP we own and that we are actually actively monetizing in a couple of the engagements we have with pharma customers. So that is very much still a way these services get delivered for end users into solutions and we see that as a value add for us going forward.

Great, thank you and congrats on the results.

Thank you.

Thank you. Our next question comes from Troy Jensen from Kanner Fitzgerald. Your line is open.

Hey gentlemen, congrats on the results. I'm sorry I jumped on a little late here but I guess a question Farage. Just, you know, with the balance sheet you have right now, just the thoughts of acquisitions, you know, software, and I guess ultimately, I'd love to know if multimodality is something that Continuum would ever pursue, or is it always going to be just trapped ions? Thank you.

Great question. It would be kind of getting into speculation at this point, and I don't want to do that. What I will simply state is Number one, we have great confidence in the modality path we've chosen and our delivery just demonstrates further confidence in the way we are acting. On the software spaces, we look at a gamut of possibilities from deep partnerships to enabling folks on Nexus, collaborations on standards and benchmarks. And we have, with the balance sheet we have and our position, we are always actively looking at What's the best way to propel our business and deepen our moats? And we will continue to be good stewards of that capital and use it for those purposes to drive our business forward and accelerate where necessary. Yeah, okay.

And just maybe a follow-up. I guess, are your thoughts on trapped ions? Do you think that's the time to solution and time to cost winner for all applications? Or does it cover most of the applications? Or I'm just kind of curious. I always thought that there could be different modalities that are good for better applications, but I'm just looking at your thoughts on that.

I'm glad you used the word time-to-solution because time-to-solution incorporates not only the speed of the computer but the accuracy of the computer as well because there's no point calculating the wrong answer quickly. As we look at our lead in fault tolerance, for instance, and being able to generate with fewer qubits and fewer rounds of error correction the entire chain of correction, we actually see time-to-solution being in favor of those and many more modalities, including Trap 10 and us, that actually have the highest fidelity and fidelity being more important in order to deliver an effective, real, usable answer, of course, within the time-to-solution bounds that that answer needs. And so we actually look at our approach on building fidelity and building the best fidelity and then innovating to cut that down as fast as we can deliver that fidelity in terms of execution speed as the right way to meet our customers' workloads and use cases that they want from us.

Understood. All right. Thank you, and congrats going forward.

Good luck going forward. Thanks again.

Thanks, Troy. And operator, I think it's sensitive to time. We'll take one final question.

Okay. Thank you. Please stand by. Our last question comes from Tyler Anderson from Craig Hallam Capital Group. Please go ahead.

Hi, this is Tyler Anderson on for Craig Hallam. Thank you for taking my question. I was hoping I was going to make the cut. So I was reading your work on the error correction codes. And to me, this seems like something that can be used to separate out, whether it be a Magic State factory or just something to create T gates. such that that area requires much less resources than what would be otherwise and then have a separate area for, let's just say, the logical or the logic compute of the system. And I'm wondering, to what degree does that free open logical qubits for you and enable you to calculate with greater depth and a faster time to solution?

So if I understood your question correctly, Tyler, thank you for asking it. What I'll tell you is if you want to build a universal false-tolerant quantum computer, you're going to have to have magic state distillation or at least magic state cultivation to a certain extent. What is critical is that the magic state output is high fidelity. But you also, because magic states are built or magic state factories depend on the number of magic states you may have to inject and how effectively you can build them. It actually, we stand with a very distinct advantage because we can build those magic states using far less physical qubit resources than anybody else can because of the improved physical qubits we have, the all-to-all connectivity and our ability to be more efficient, therefore, in a higher acceptance rate. So when we look at a full logical computation which has a universal gate set, we are already lower in the number of magic state resources that we need. And I just told you why. And so therefore, what we see is not complicating the developer's life by giving them two completely separate paths because we are already generating the world's best leading logical error rates and have a path forward on continuing to do that without running into a wall.

Okay, that makes sense.

And I agree with the trying to abstract a way for end users. I think that's very important. That's all for me. Thank you. Thanks again, Tyler. Thank you.

Thank you. This concludes the question and answer session. Thank you for your participation in today's conference. This does conclude the program and you may now disconnect.