Horizon Quantum Holdings Ltd.

Good day and thank you for standing by. Welcome to Horizon Quantum first quarter 2026 report, conference call and webcast. 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, please 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. Please note that today's conference is being recorded. I would now like to turn the conference over to your first speaker, Katherine Bailon. Please go ahead. Thank you.

Good morning and thank you for joining us on Horizon Quantum's first quarter fiscal 2026 earnings call. My name is Katherine Bailon and I am vice president of investor relations at Horizon Quantum. Joining me on today's call are Dr. Joe Fitzsimons, our founder and CEO. and Greg Gould, our Chief Financial Officer. Before the market opened today, Horizon Quantum issued a press release with results for its first quarter ended March 31st, 2026. You can access a copy of the release in the investor relations section of our website at horizonquantum.com. Today's call is being broadcast live over the web and can also be accessed through the investor relations section of the Horizon Quantum website. Before we get started, I'd like to caution you that any non-historic statements that I or the management team make during this call will constitute forward-looking statements within the meaning of U.S. federal securities law, including the Private Securities Litigation Reform Act of 1995. You are cautioned that these statements involve material risks and uncertainties, including the risks Horizon Quantum has identified in its most recent annual report on Form 20F and other filings with the SEC and that we typically cite in our press releases that could cause actual results or events to materially differ from those anticipated. Except as required by law, Horizon Quantum expressly disclaims any intention or obligation to update or revise any financial or product pipeline projections or other forward-looking statements whether because of new information, future events, or otherwise. These statements made on this conference call contain time-sensitive information and are accurate to the best of Horizon Quantum's knowledge only as of the date they are made today, May 5, 2026. This discussion includes non-GAAP financial measures. These measures are calculated by management and do not have any standardized meanings under U.S. GAAP. These non-GAAP measures supplement GAAP measures, but should not be viewed as substitutes for GAAP measures. We have included a reconciliation of GAAP measures to non-GAAP measures as a table at the end of our earnings press release issued this morning. And with that, I will now turn it over to Horizon Quantum founder and CEO, Dr. Joe Fitzsimons.

Thank you, Catherine, and good morning, everyone. we appreciate you joining us on the earnings call today um on the call i'd like to highlight uh five key points um we've the first one is that we have grown our hair count rapidly from a year ago uh with a few more additions in the current quarter we believe this will uh we believe we'll be staffed to execute on our roadmap secondly in the march quarter we deepen two hardware partnerships and we expect to continue to add and deepen other hardware relationships in future Third, in our software stack, we've made progress moving upwards in abstraction with our language, Beryllium, expanding its feature set and improving stability. Combined with our other languages and tools, we expect Beryllium to make it easier for developers to create quantum applications to tackle real-world problems. Fourth, with our software stack, we are also tightening integration with quantum hardware, which we believe is critical to enabling quantum advantage. And fifth, we are grateful to our new partners and our public shareholders who have provided us with enough capital for the foreseeable future to blaze a path as a public quantum software company. But before we get into detail on the quarter, as this is our first earnings call as a public company, I want to take a few minutes discuss how we think about quantum computing at Horizon Quantum and why we believe we are well positioned to grow as the quantum computing industry evolves and pushes towards quantum advantage. I'll then walk you through our R&D and business highlights for the quarter before handing you over to our Chief Financial Officer, Greg Gould, to update you on the quarterly finances. So, in my view, the promise of quantum computers is to be better computers. They are, after all, computers with an extra trick. They have the ability to harness interference between different computational branches. And that makes them unique. The reason I first got into quantum computing, because I believe that there was a chance that quantum computing might mirror the trajectory of conventional computing. And 22 years on, I believe that more strongly than ever. And this historical perspective informs our approach to harnessing quantum computers to do real work. And make no mistake. beyond press releases, beyond proof of concept experiments, is of paramount importance that the quantum industry reach a point where software running on quantum computers is producing results that create real value. That's why our mission is to unlock broad quantum advantage by building the software infrastructure to solve the world's toughest computational problems. Today, the process of creating programs that harness quantum processing is laborious And it requires formulating algorithms to take advantage of quantum computing's peculiar ability to extract information from multiple possible branches of a computation through quantum interference. The industry has largely settled on three approaches to this problem. The first is what you might call an education approach, where the hope is that developers can be taught enough about how quantum computing works that they will be able to formulate their own quantum algorithms. While some beautiful educational material has been produced, which has been very helpful in attracting students into the field, the reality is that much of the progress in quantum computing algorithms is still made by people with many years' experience. The second approach is based on libraries, where pre-written quantum programs are called to accomplish particular tasks with these calls stitched together with classical compute. The approach, well, this approach works perfectly well when what you want to do is already available through one of the growing number of quantum libraries. But it falls short if no library addresses your particular problem or if customization is needed. It also misses out on the opportunity for compounding speedups that are possible when such primitives are combined as part of a larger quantum program rather than accessed as as classical programs, sorry, excuse me, rather than accessed in a classical program in a kind of black box manner. The final approach is based on professional services. This makes sense for a small number of high value use cases, but given the rarity of quantum algorithms expertise and the amount of work necessary to tackle each new problem, relying on this approach will likely substantially limit the rate of uptake of quantum computers. At Horizon Quantum, we're pioneering a new approach to this problem, using the creation of quantum algorithms as a compilation task rather than as an R&D project. Our long-term goal is to bridge the gap between classical and quantum computing by automating the acceleration of classical code on quantum computers. This means automating the construction of quantum algorithms to construct high-level quantum programs and then compiling and executing these on quantum computers. Doing this would essentially abstract away the quantum mechanics for developers not interested in those details, allowing domain experts in the fields that most sound to benefit from quantum computing to harness these systems in their work without having to become experts in quantum computing themselves. If we can achieve this, we expect it to significantly increase the breadth and depth of quantum computing use in enterprise. To achieve this goal, we need to overcome the fact that quantum computers are not yet living up to their full potential. Rather than being better computers, current generation quantum computers are much less capable than their classical counterparts. They are much more like the computers of the 1940s and 1950s rather than contemporary HPC systems. Most quantum computers still lack the ability to implement basic computing functions, including general control flow, allowing for the implementation of loops and subroutine calls, and are instead limited to executing static circuits. Fortunately, there are compelling reasons to believe that quantum computing has reached something of an inflection point. First, there is now strong evidence that quantum computers of multiple modalities are difficult to simulate. This does not necessarily mean that they are solving valuable problems, but it is an important first step. If we could simulate these systems with conventional computers, there'd be little point in building them. It also has the advantage of putting the ball firmly in the court of software. Secondly, quantum error correction is becoming a reality, with convincing demonstrations of multi-round error correction reducing the error rates below those of the constituent qubits. This is critically important for enabling the kind of long quantum calculations that are necessary to create a broad quantum advantage. Thirdly, the anticipated overhead of error correction has plummeted with the discovery of asymptotically good codes driven by progress in quantum LDPC codes. This means that the scale and complexity of the quantum computer needed to tackle a given problem has dramatically shrunk. Finally, there are now four distinct qubit modalities that have proven an ability to scale and offer viable and independent paths to full tolerant quantum computing. So software has never been more important to the trajectory of the industry. However, these systems are largely still being programmed with quantum assembly languages that are not too dissimilar from the microcode of the 1940s and 50s. If we want to reach our goal of using quantum computers to accelerate classical code, we need to close the gap between the software infrastructure modern code for conventional computers expects, the abstractions provided by programming languages and operating systems, and the capabilities of quantum computers. So we have essentially been attempting to speed run the history programming languages and system software learning what we can from the development of classical systems. So first we built Hydrogen to act as a single portable quantum assembly language that we could build upon that could describe programs both for today's systems but also for future quantum computers. Hydrogen is designed to be maximally expressive. so that code written in hydrogen can express anything a mature quantum computer would be able to do, independent of architecture or instruction set. And we built a compiler and execution stack to be able to take code written for these future systems, these future more capable quantum computers, and execute it on the current generation systems. And I would say that firmly places us in the 1950s. Next, we built Helium. Helium's an imperative basic-like quantum programming language designed to make it easier to express complex programs. Helium includes features such as dynamic memory allocation, IO management, and the ability to construct quantum circuits that implement functions written in C, bringing 1960s and 1970s technology into the quantum computing software stack. Most recently, we previewed Beryllium, an object-oriented quantum programming language. Barillium is intended to allow developers to start abstracting away quantum mechanics. We've designed Barillium such that a developer can define classes which exploit quantum processing to represent and manipulate various kinds of data. And when these classes are later used to represent that kind of data, by that developer or by another, they can spend their time thinking about how they want to manipulate that data rather than about how the underlying processing is achieved. Object-oriented programming languages are also special in that code reuse is a primary design goal. By allowing developers to build upon each other's work and each other's abstractions, they can create a strong network effect. where each module or library written unlocks new abstractions for the next developer, allowing them to accomplish more with less code. So for us, Brilliant is a key milestone, not just because it brings 1980s technology into the quantum computing stack, which, in my view, gives us a substantial technological lead, but because it enables developers to begin abstracting away quantum mechanics and to reuse code via libraries potentially unlocking the strong network effect. We believe that advancing our technology stack and working closely with the hardware manufacturers to unlock the full potential of their systems is the shortest path to achieving broad quantum advantage and capturing long-term value in quantum computing. Since software built through our tools is deployed through our infrastructure and executed through our stack, Our business model is centered on a combination of usage-based fees for cloud-based system and annual licensing for on-premises deployment. I'd like to leave you with three thoughts on this. One, we believe that software infrastructure is the key to unlocking the full potential of quantum computing, enabling broader use cases and more rapid uptake than would otherwise be possible. Two, we are focused on building that software infrastructure in a way that is hardware agnostic from the perspective of the developer, allowing them to focus on building software rather than trying to guess which hardware modality will ultimately win out. And three, ultimately, we are a software company, so we expect our margins and our CapEx to look completely different from hardware companies. Now, turning to the quarter, The first quarter of 2026 has been a rather eventful one for us. In January, we had the inauguration of our quantum computing testbed, including the first testbed system, Ember One, officiated by Singapore's Minister of Digital Development and Information, Josephine Cho. The Ember One system is based on a Rigetti Novera 9-qubit superconducting processor, an OPX1000 microwave controller tracked from quantum machines and a Maybel big fridge dilution refrigerator. It was integrated from parts in-house at Horizon Quantum, which I believe makes us the first quantum-focused software company to begin operating quantum computing hardware rather than relying solely on cloud-based systems. This allows us to more tightly integrate our software stack with the hardware without having to pass through third-party software layers on the way down. We believe that tight integration between software and hardware are critically important to achieving quantum advantage in the near term. Notably, this quarter we completed our business combination with DMY Square Technology Group, becoming a public company in the process and listing on NASDAQ. While this was a fantastic moment of celebration for the team, we are all keenly aware that we have only reached the starting line of a new race. The transition to being a public company, we have built out our board of directors. It's extremely gratifying to see the level of experience they bring. Danielle Lambert was the VP of human resources at Apple, where she helped to build out the teams behind iPod, iPhone, iPad, and Apple retail stores, among others. She later played Pivotal Road in the founding of Nest Labs before its acquisition by Alphabet. Peter Wee is the chief financial officer of Grab, a prominent Southeast Asia super app. He previously served as CFO of LegalZoom.com prior to its IPO and as CFO of MyLife.com. Earlier in his career, he held several financial leadership roles at Activision Blizzard. Jill Turner has more than 20 years of experience in global human resources leadership roles at Fortune 500 technology companies. She's currently a CHRO of Broadcom and previously held executive positions at Honeywell and Lumen Technologies, formerly CenturyLink. Harry Yu is currently a member of Broadcom's Board of Directors. Previously, he was a lead independent director of INQ. He's also an experienced public company officer, having held CFO roles at Accenture and Oracle, and an EVP role at EMC, among others. In anticipation of the closing of the business combination, we increased our headcount, particularly in R&D, where we saw a twofold increase in headcount compared to the same quarter in the prior year. We believe this headcount increase will both help accelerate our technological development and will allow us to work with a greater number of hardware companies. since historically our ability to take on new collaborations has been limited by the capacity of our science and engineering teams. Compared to the prior year quarter, we have also added key hires in GNA in preparation for becoming a public company, including Greg Gould as Chief Financial Officer, Catherine Bailon as VP of Investor Relations, and from May 11th, Catherine Fitzsimons as Chief Legal Officer. With these headcount additions, we are nearing a state where we believe we have the team necessary to execute on our plans. In the first quarter of 2026, our R&D efforts have included a focus on the continued development of Beryllium beyond its initial preview form, first shown last December, to a more feature-complete and stable form. We have also focused on integrating our first test system, Ember One, into Triple Alpha in anticipation of making it available to external users later in the year. A key part of our strategy is to try to ensure that our software stack is the best way to program quantum computers no matter the underlying technology. We track approximately 40 leading hardware manufacturers and seek to collaborate with companies where our software stack can increase the capabilities of their hardware systems or enable developers to build more complex software for these systems. Ultimately, our goal is to ensure that no matter which system architecture takes a lead in quantum computing, you'll be able to program it with triple alpha. At present, we're fortunate to see more interest from hardware manufacturers than we have capacity to service. Though we believe the increased hiring in our science and engineering teams will help alleviate this bottleneck. And so we prioritize collaborating with companies where we believe there's a strong potential to take a leading role in the path to large scale error corrected quantum computing. In the first quarter, We announced that Horizon Quantum had partnered with two additional quantum computing hardware companies. In particular, we're delighted to be working with Alice and Bob to streamline the development and deployment of full-tolerant quantum software. A first step in this process is to integrate their emulators into Triple Alpha to allow developers to interact with emulated versions of their planned quantum processors. We're excited about this collaboration in particular due to the potential of Alice and Bob's 2-bit approach to offer a shortcut to full-tolerant quantum computing and one that is potentially very hardware efficient. We're also delighted to be working with AQT to further develop quantum computing applications. This will involve integrating AQT's simulators and trapped ion quantum processors into Triple Alpha. We expect the collaboration to provide an opportunity to deepen support for and integration with trap-line systems within our software stack. Finally, I want to turn back to our testbed. Working directly with hardware has significantly accelerated progress, particularly in the development of a real-time execution stack, which would not be possible via standard cloud APIs. Tight integration between hardware and software is critical to achieving quantum advantage, and having direct access to a dedicated superconducting system has been a significant boon to our R&D efforts, enabling us to push further down the stack, enabling pulse-level programming and pulse-level optimization within our stack. Although there are many quantum computing modalities based on different underlying physics and different physical systems, at a very, please forgive this, but at a very coarse level, these can be divided broadly into two categories in two different ways. So the first categorization is that you can think of systems that need extreme cooling and microwave control on the one hand, and systems requiring optics and high vacuum on the other. While this doesn't tell you much about the characteristics of the underlying quantum system, it informs the requirements that are, excuse me, the requirements are important for informing facilities and operations requirements. So understanding how to operate these systems, developing the capabilities for operating each of these systems, more or less enables you to host and to operate other systems of the same category. The second characterization is between solid-state qubits versus atomic or photonic qubits. The first are essentially manufactured qubits, for which the parameters can be controlled during fabrication process, but which suffer from limits of precision and defects. The second category, so atomic and photonic qubits, are essentially perfect, uniform, indistinguishable quantum systems, but where the parameters are dictated by nature rather than by design. Since we took the decision to stand up a testbed several years ago, we had in mind a roadmap that would take us through three different systems. The first will be a small-scale superconducting system, since that was the most common modality in industry. And there is a maturing open-stack ecosystem, which allows for a cost-efficient entry point. This system would act as an exemplar of the first category of qubit modalities, according to both categorizations. The second system was intended to be a moderate-scale exemplar of the second category, being either a trapped ion or neutral atom system, with a modest number of qubits. The third system was anticipated to be best in class across all modalities and was expected to be the only one capable of reaching quantum advantage. Over the last two years, we've spoken to a variety of hardware manufacturers building both trapped ion and neutral ion systems. And in Q1, we entered into an agreement with IonQ for them to provide a 256 qubit trapped ion quantum computer. Importantly, the anticipated qubit count and fidelity of the system are expected to put it on the cusp of quantum advantage for at least some chemistry problems. As a result, this system will satisfy the criteria we had looked for in the two remaining planned testbed systems, allowing us to accelerate our roadmap with a single system rather than two. While the system is a significant expense, which we expect to recognize relatively over time, we believe that operating a multimodal testbed puts us in a strongly differentiated position to develop the most capable tools for programming quantum computers across all modalities. We're committed to operating in a fiscally responsible manner, such that we have the runway necessary to execute on our plans. Jurisdictions in which we operate have provisions for significant support for R&D activity, including by our refundable tax credits, which we take into account in cash flow planning, and which we expect could offset some of the costs associated with testbed operations. In our filings, we've disclosed plan for a second testbed system to be installed in calendar year 2027, and we believe with this new IonQ system, we're on track to achieve this. With these developments in mind, I believe we're well positioned to advance our software stack and position Triple Alpha as the most capable tool for harnessing the full potential of quantum hardware. And now I'll turn it over to Greg to discuss our financial performance in more detail.

Thank you, Joe, and good morning, everyone. i would like to start off reviewing our financial results in more detail but before i do i want to make three quick comments first all the financial results uh to be discussed are in u.s dollars second since we completed the business transaction combination on march 19th during the quarter all results presented prior to the closing of the combination reflect the financial results of our primary operating company horizon quantum computing private limited The results as of March 31st, 2026 reflect the results of the newly combined entity, Horizon Quantum Holdings Limited. Third, our quarterly historical financial results are presented in the last table in the press release for additional information. Okay, let's get started. Total operating expenses for the first quarter of 2026 were $6.5 million compared to $4.7 million for the first quarter of 2025. The widening loss was largely due to hiring primarily on our science and engineering teams. R&D expenses, which consists primarily of personnel-related costs for scientists, software engineers, technical staff engaged in the design, development, and testing of our software and hardware systems, and to a smaller extent, it also includes software and other cloud services costs associated with the operation of our hardware testing. For the first quarter of this year, R&D expenses decreased 1.19 million, or 36%, to 2.13 million compared to 3.32 million in the first quarter a year ago. This reduction was driven by a $2.28 million decrease in share-based compensation expense in the year-ago quarter, largely from a one-time share-based compensation catch-up expense of 2.5 million a year ago. Excluding this catch-up expense, R&D expenses increased 135% period over period, and that was driven by a 100% increase or doubling in the size of the science and engineering teams. Incremental costs from the setup of the hardware tests that also contributed to expense growth, albeit to a lesser extent. For the first quarter, sales and marketing expenses increased by 0.13 million, or 40%, to 0.46 million during the quarter ended March 31, 2026, versus 0.33 million in the prior year period. Excluding share-based compensation, sales and marketing rose 85% year over year, and this was due to higher trade show activity and industry engagement. G&A for the quarter totaled $3.6 million, increasing $2.7 million from $0.9 million in the first quarter a year ago. That's a 300% increase. Excluding share-based compensation and one-time business combination expenses of $1.9 and $0.3 million from the first quarter of 26 and 2025, respectively, G&A increased 191% period over period. For the first quarter, our loss from operations totaled $6.5 million, widening from $4.7 million loss in the first quarter a year ago, primarily due to hiring of staff across the company, including, as I mentioned before, scientists and engineers and staff necessary to become a public company. The net loss for the quarter on an as-reported basis of $3.6 million, or $0.09 per ordinary share, compared to $4.8 million in the first quarter of 2025, or $0.12 per ordinary share. Please note that the loss from operations for the just completed first quarter was $6.5 million, but the net loss was a narrower $3.6 million loss. That's because of a non-cash gain of $3 million in the quarter due to two factors. First, we recognized a $5.3 million gain from the fair value remeasurement of the warrant liabilities assumed by Horizon Quantum at the close of the business combination with BNY. This compared with zero a year ago since there were no warrants outstanding at that time. Second, the gain was partially offset by a $2.3 million loss due to the fair value reassessment and subsequent settlement of safe liabilities on the balance sheet. The safe instruments were a source of funding prior to the business combination and were converted into Horizon Quantum Class A ordinary shares at the completion of the business combination on March 19th this year. Drilling down a bit on the $5.3 million P&L gain from the change in fair value derivative warrant liabilities in the quarter, the gain was due to the decline in the price of the underlying ordinary shares from the first valuation date at the close of the business combination on March 19th to the end of the quarter on March 31st, a decline of 35%. This was more than offset by the valuation gains brought forth by the higher volatility assumptions since the close of the business combination and resulted in the net gain. To supplement our financial statements presented in accordance with GAAP, we used two non-GAAP measures, EBITDA and adjusted EBITDA. because we believe these metrics provide investors with additional meaningful methods to evaluate certain aspects of our results period over period. We define EBITDA as the net loss before net interest income or expense, depreciation and amortization expense, and income tax expense. We define adjusted EBITDA as net loss before interest income or expense, depreciation and amortization expense, income tax expense, but also share-based compensation change in fair value of derivative liabilities, and non-recurring business combination expenses. We use EBITDA and adjusted EBITDA to measure the operating performance of our business by excluding specifically identified items that we don't believe directly reflect our core operations and may not be indicative of our recurring operations. The reconciliation of GAAP net loss to EBITDA and adjusted EBITDA is presented in the last table in the press release. For the first quarter, our EBITDA loss was 3.3 million compared to 4.7 million in the first quarter of 2025. Our adjusted EBITDA for the first quarter was a loss of 4.1 million versus a $1.8 million loss a year ago. The widening of the loss was primarily due to hiring. In terms of employees, we ended the first quarter with 52 FTEs versus 29 at the end of the first quarter a year ago. That's a 79% period over period increase and is up 18 percent sequentially from the December quarter. Historically, scientists, engineers, and other technical staff have represented between 60 and 80 percent of the total headcount and remained within that range as of the end of the March quarter, 2026. Going forward, we do not expect this focus to change. Now, in terms of cash flow, net cash used in operating activities for the quarter was 4.2 million compared to a cash usage of 1.6 million in the first quarter of 2025. The bulk of the increase in cash usage from 25 to 26 was due to the growth, as we mentioned before, in headcount. Net cash provided from financing activities during the first quarter was $101 million, with the vast majority of the total generated coming from the pipe and the business combination with DMY. And to a lesser extent, the issuance of safes prior to the close of the business combination with DMY. Net cash provided by financing activities during the three months ended March 31st, was at 2025 was nil as we did not undertake any financing in that quarter. As of March 31st, 2026, we had $96.6 million in cash on the balance sheet. This amount reflects the majority of expenses tied to the business combination already having been paid out. We are pleased with our financial and operational performance. As has been the case throughout the company's history, we continue to believe that the best strategy to realizing Triple L's technical potential is for our science and engineering teams to remain focused on the execution roadmap. As a result, we do not pursue proof-of-concept services work, which, while potentially generating nominal revenue, we believe would serve as a distraction for our team and slow progress on our roadmap. We plan to engage in revenue generation when the larger quantum industry is reaching quantum advantage. That's the moment when quantum computing can finally solve the hard computational problems that classical computers cannot. That would be when we expect to transition today's early access users to paying users on a cost per use model, either on the cloud or on premise. Finally, I would like to share our thoughts on our cash balance, which stood at $96.6 million at the end of March. Since we are a software company, our model is a capital light one. We expect the cash balance raised during the recent business combination and related transactions to provide us with sufficient runway for the foreseeable future, as it is our intention to proceed with fiscal discipline and rigor. With that, I'll turn it back over to you, Joe. Thanks.

Thanks, Greg. And now we'll take your questions. Razia, I'll hand it over to you.

Thank you. As a reminder, to ask a question, please press star 1 1 on your telephone and wait for your name to be announced. To withdraw your question, please press star 1 1 again. Once again, please press star 1 1 and wait for your name to be announced. To withdraw your question, please press star 1 1 again. We are now going to proceed with our first question. The questions come from the line of Queen Bolton from Need Home & Co. Please ask your question.

Hi, guys. Congratulations on your first public call. Thank you.

Thanks, Quinn.

I guess maybe Joe, just to start off, is sort of a big picture question. Can you just elaborate on the go-to-market strategy? You've got the ability to sell triple alpha sort of through cloud access. but you also have the on-prem. Can you just elaborate on those two channels, and can you frame in any way what sort of revenue might you be able to generate from each of those two channels, either on an annual basis for the cloud access or maybe on a per on-prem system basis? Anything that can help shape sort of revenue expectations over the next few years?

Sure, so let me do my best to answer that for you. In terms of go-to-market and in terms of how we intend to generate revenue, ultimately our current focus is ensuring we are in place with the hardware manufacturers. So we're focused on working with the hardware companies to the best extent we can to ensure that our tools are the best tools to program those systems. and we're focused on technology development to ensure that we have a lead so that we get to a place where our tools are really able to achieve a broad quantum advantage that enables the creation of real value from software running on quantum computers. Now, in terms of how we intend to generate revenue from that, ultimately software built using our tools runs through our stack. That's necessary because actually a lot of the features we enable, things like the ability to support general control flow and so on, they require our runtime environment. So even if you have the compiler completely for free, you would still need to execute those programs to our stack in order to be able to take advantage of these features that go beyond the current state of the art in current quantum programming programs and in current quantum computing systems. Now, that puts us in a good position. It puts us in a position similar to where AWS sits with web applications, where applications built using them call APIs. In this case, applications built using our software would call APIs in our software stack. each time they're used. And that puts us in the position to charge based on usage for the deployed application rather than development. And it allows us to charge based on the value of the resource being accessed. Similarly, if you look at on-prem deployment, again, there's a pretty standard model there that allows you to charge effectively proportional to hardware costs by charging based on the capabilities of hardware. So if you look at like Windows Server or VMware, you see these pricing models where you have software license that is based on the capabilities of the hardware system it's running on. So that's our approach in terms of where those percentages are going to sit and so on. I'm not going to comment on that at the current time.

Understood. Thank you, Joe. Maybe for Greg, just looking at the adjusted EBITDA loss in Q1 of $4.1 million, as you look through the rest of 2026, would you expect that adjusted EBITDA loss to sort of range in the same level as Q1 through the rest of the year, or do your hiring plans sort of suggest that that adjusted EBITDA loss might trend higher over the quarters? And if it does trend higher, could you give any shape to that trend?

Yeah, sure. I'll let Joe comment as well. But the bulk of the hiring has already occurred in anticipation of the transaction. There'll be a slight increase and then a leveling off. But what I think is important, and Joe mentioned this also, is that we think we will have, with a few more hires, the team that we need to execute on the road map. Excellent.

But I think we're in a pretty good position at this point.

Yeah. Yeah, so it sounds like that adjusted EBITDA loss for Q1 is probably a pretty good proxy for the near-term run rate of the business.

We have a few hires still coming in, but, you know, you're not too far off base. Okay.

Perfect. And then... Lastly, I guess for either of you, you mentioned the IonQ system, the 256 qubit system that you would be installing, I think, in 2027. It sounds like there's probably some CapEx associated with that system. Is it possible to sort of give a sense as to what that CapEx requirement might be sort of as that system is installed? I know you mentioned it would be sort of the cost would be incurred ratably over a period of time. But I'm not sure if you can provide any thoughts on how that might affect CapEx in 2026 or 2027. Greg, do you want to take this?

Yeah. So I think this is the big capital expenditure. Aside from that, I think you'd see sort of normalized CapEx around hardware for employees and a few things. But this is kind of the big one. And the purchase of the decision of this was taken in the full context of cash usage and the opportunities from various jurisdictions to help fund it.

Let me ask. If you look at our 20F, you can see the amount where we're paying for the system there. You know, the payment schedule and so on is redacted, so I'm afraid we're not able to share the full details of that at the current time.

But as I say, it'll be recognized readily over a period of time.

Excellent. Thank you very much.

Sure. We are now going to proceed with our next question. And the questions come from the line of Tyler Anderson from Craig Hallam Capital Group. Please ask a question.

Hi, everybody. Thank you for taking my questions and congrats on the first call as well. Piggybacking off of the first question that was asked, would you be able to describe the customers that you're going to be marketing to within the cloud access portion?

So that's a good question. And what I can tell you actually is we get a fair amount of inbound requests for early access. At the moment, our focus is on the quantum computing industry and working with hardware companies. But we get a lot of inbound requests for early access to our system. And I would say they broadly cluster into six areas. So the first, let's say, is general big tech. But if you leave that aside, the five areas you would see are... One is in financial services, so you definitely see it. You see it in pharma and chemistry as well. You also see it in the energy sector, both with oil and gas and renewables. So that can be everything from like photovoltaics to like Kriging, which is a particular – it's a particular – mathematical model, well, machine learning model that's sometimes used to model oil and gas deposits and so on. And you see it in things like energy trading. The other two sectors that you see it from are automotive and then aerospace and defense. So hopefully that gives you a reasonable idea of at least where we're seeing interest.

Yeah, absolutely. And then for the hiring that you're doing, are these people who are only located in your two headquarters, or are you hiring for remote positions as well?

Well, I mean, I think Catherine is a perfect example of this in the sense that she's based in the US rather than either one of our two locations, but primarily we hire at the two locations we're based in, so Singapore and Dublin. Although we have occasionally people who are remote Generally, the way we approach this is that the strong preference is to have people work with us directly, you know, in the office. But we make exceptions when we know people by reputation or we've worked with them before and so on. So depending on the situation, we sometimes hire remotely, but primarily it's in the locations where we're operating in.

Thank you. And then just for everybody who may be listening who's not within the quantum industry, from a non-expert perspective, could you just describe the benefits that your platform gives to software engineers who want to adopt your platform? And these could be from the six groups that you had previously mentioned.

Sure. Ultimately, what we are building towards That's not to say where the current state of the system or so on, but what we are working towards is the ability to automatically construct quantum algorithms from code written for conventional computers. So what I mean by that is the ability to take code written for a conventional computer and automatically accelerate it on quantum computers. Now, the reason we are working towards that goal is because if we can do that, we can hand the domain experts in these different industries tools to be able to harness quantum computers without having to go out and get a PhD themselves. So, you know, if you think about it, how are people working in quantitative finance? How are people working in pharma? How are people working in, you know, in oil and gas? Whatever it happens to be, that are deep experts in their particular industry, how are they going to harness quantum computers? Are people from quantum computing companies going to go out to them and act as consultants and work with them? Maybe that works. But first of all, there's a real bottleneck in terms of the limited amount of expertise available. But the other side of it is you have this knowledge transfer problem. getting someone with a quantum computing background up to speed enough on these domains to be actually able to make a meaningful difference is, you know, almost as hard as teaching people in these, you know, who are experts in these domains enough quantum mechanics that they can start harnessing quantum computers themselves. So, you know, we think this is a real unlock if we can achieve this. Now, in order to get there, As I said earlier in the call, we've been trying to speed run the history of computing, or at least of software. So we've been trying to build programming languages that get to a higher and higher level of abstraction such that we can start to abstract away the quantum mechanics itself. And we're starting to get there with Beryllium. So Beryllium is an object-oriented programming language that allows developers to express computations that combine both classical and quantum information. So you can define objects to represent different kinds of data that include both classical and quantum data and classical and quantum algorithms for processing that data. And it turns out when you do this, you can start to embed quantum speedups such that a subsequent programmer using that library that you've developed in this way is then able to get those speedups without having to refer to quantum mechanics at all, because they're just accessing the different methods, the different ways of manipulating these data types that have been defined when the class was built. A way to think of this a little bit, I would say, is if you think about how a lot of machine learning calculations are done these days with Python, with PyTorch, and NumPy, and so on, you don't actually have to worry about how to implement matrix multiplication on a GPU. That's kind of taken care of for you by the developers of those packages. So you don't really need to care about the details of how a H100 works, for example. And that's where we're starting to get to, where developers can build up their own abstraction errors. So we can start to get away not just from the details of the quantum computer architecture, but we can get away from the quantum mechanics itself. the low-level details of how algorithms are implemented to just focus on how you're manipulating different kinds of data. And we think that's a real unlock, and that's going to be extremely powerful in enabling developers to build the kinds of sophisticated applications we need to take full advantage of the quantum computers that are starting to emerge now.

Well said, Joe. Thank you. I appreciate the time.

Thanks, Tyler.

We are now going to proceed with the next question. And the questions come from the line of Troy Jensen from Cantor Fitzgerald. Please ask your question.

Hey, gentlemen, and congrats on the results here in public company status. Maybe just, I guess, for either one of you guys, I know we talked about kind of the capex for the 256-qubit system. Can you talk to, has the capex been spent for the Rigetti system? And maybe disclose how much that's going to be also.

Yeah, maybe, Greg, you can speak to this. But, yeah, most of that expense has been incurred over the past couple of years. Yeah, it's behind us now.

Sorry, I broke up. What was that?

I was saying, do you want to dive into the details on that? But yeah, the only thing that was remaining, I mean, you can see it in our F4, there was a partial payment on a dilution refrigerator, still due, but generally all this in the rear view mirror.

Yeah. All right, perfect. Yeah, I thought that was an expense coming up, so that's helpful.

No, no, no. I'm obviously priced.

Yep, understood. And for you, Joe, I mean, you picked superconducting and you chose trapped ions. I mean, is that your belief in the methodologies that are going to get commercialized first? And what does this mean for you guys' positioning in photonics and neutrons?

That's a good question. I mean, the reality is there are four different models that are all charging forward together. When I talked earlier about these kind of course categorizations of systems into two categories, whether you're talking about stuff that needs to be cold and needs microwave control versus stuff that needs high vacuum and optical stability and so on, that kind of informs you, you get to learn how to operate systems of that class, of that category. So we went with a superconducting system first because it is an exemplar of the first category, which also includes things like silicones in there, various kinds of quantum dots, and so on. Even electrons on helium would also fall into that category of stuff where you need a DILFridge, you need microwave control, and so on. But superconducting of those is the most widely pursued in industry. It's like the most advanced. There's already this kind of emerging open stack so that it's not impossible to piece together a system yourself. I mean, it's challenging, of course. You need a bunch of physicists. But you're not developing the hardware from scratch. The choice to go with the trapped ion system really, if you look at where things stand, the 256 qubits, the transition to microwave control and so on that has happened at IonQ following the Oxford Ionics acquisition makes their roadmap very exciting to us. You know, I have to admit, I've known Chris Balance for quite a while. We were both, I think we overlapped at Oxford like a decade and a half ago, something like this. And they have achieved really spectacular fidelities for single and two cubic gates. You know, it's some of the most, I mean, I think they've had the record of literally the highest fidelities of any modality ever. for a significant amount of time. So being able to get to this point where there are a large number of qubits in the device, that's really, really attractive. Now, neutral atoms are, of course, also on this path where they're scaling up massively, significantly beyond the numbers of qubits that you see in trapped ion systems. But the level of control isn't there. And if you look at the commercial systems that are out there at the moment, you know, gate model quantum computing is still a little way off for most of the providers. And our focus at Horizon has always been universal quantum computing. So not going to like things like Ising machines or boson sampling or something like that, but focusing on universal systems. So it's not, you know, I'm not picking winners, but these are definitely interesting systems to us.

Okay. Well, I love the passion. Congrats and keep up the good work. Thanks, Troy.

We are now going to proceed with the next question. And the questions come from the line of John McPate from Rosenblatt. Please ask your question.

Okay. Can you hear me okay? Yeah. Great. Welcome to the public markets and the first of many conference calls, guys. I'm just going to ask this one directly. I don't expect an answer, but I'd love one. When do you expect delivery of the 256 qubit machine? Can you give us some sense?

That's a good question. I'm a little bit restricted in what I can say on that because of confidentiality agreements with our partners on that. The, what I can say is that, you know, back in our F4, we said that we expected to have a, you know, a second testbed system being installed in 2027. And, you know, we're currently on target to achieve that.

Okay, so just sort of that 12 month span, is that the way I should think about it, Joe?

I'm sorry, I can't be more precise than that again.

Fair enough, fair enough. And then on the software side, because it's what we're all about here, talk a little bit about turnkey algorithms, where those may or may not be in your future, where I can just click Shores and have a target and calculate... Talk a little bit about that, the ease of use to the enterprises and how that will proceed.

Sure. So I think what's important here is that, you know, I think a lot of the way the industry views how quantum computers will be used, and keep in mind that, you know, many of the people in the industry at the moment are coming from a hardware perspective. I think there's a lot of... there are a lot of assumptions that what will happen is that you have these kind of pre-baked algorithms, and they will be stitched together with classical compute. That they'll be maybe made available in libraries and stitched together with classical compute, used as kernels in some kind of hybrid computation. That is not my view of the future. That is not what I think is actually critical here. Let me tell you why. Let me give you a couple of examples. So you really benefit if instead of stitching together quantum algorithms with classical compute, you're stitching them together with quantum compute. So that's why we build quantum programming languages rather than Python frameworks for describing a quantum computation or something like that, because we want the information when we stitch together building blocks to happen in a phase coherent way so that quantum information can be passed between primitives. And an example in relation to this is that there's a lot of people think that Grover's algorithm is useless, that the overhead for full tolerance would swamp any gain you would get from it. But they're wrong. And the reason I say they're wrong is that you can imagine many different tasks where this is not true. So imagine, for example, you were working at NSA or something like that. You're faced with some number of communications intercepts, and one of them contains the relevant information for you, whatever plot is happening or something like this that you're trying to uncover. So if you're thinking in this kind of kernel way, like I'm going to call it quantum primitive, and so on, like, okay, I'm going to call Shor's algorithm to break each channel, and I'll do it one at a time or in parallel if I have access to a bunch of quantum computers, and then I'll just sift through that data classically at the end to decide, you know, what was the relevant message. But it turns out there's a much better quantum algorithm for this task, which is simply to construct an oracle for Grouter's algorithm based on Shor's algorithm. So you are using Shor's algorithm to decrypt the communications channel, and then you're searching over the possible communications channels coherently. And if you do that, the effort you need to put in is proportional to the square root of number of channels rather than the total number of channels times the difficulty of breaking each one. So you get a massive speed up and you don't really need to worry about the fault tolerance overhead because that has in some sense already been paid in implementing Shor's algorithm. Now, you know, that's kind of a particular example, but you can think of like much more, you know, much more industry specific ones. So let's say you're looking at finance, for example, and you want to look for mispriced derivatives. So you're trying to do some kind of computational arbitrage from derivatives that are mispriced. So you're going to search over instruments. You're going to search over hedging strategies. And then you're going to use Monte Carlo to price. And you're looking for a mispriced derivative. If you try to stitch together these primitives, like stitch together search, search for the outer loop, search for the inner loop, and then the Monte Carlo pricing, each one has a quadratic speedup on a quantum computer. But if you stitch them together with classical compute, you only get one of those speedups. You're really kind of having the sum of the runtimes of each one. But if you can stitch them together with quantum compute, then you are able to get to a point where you have compounding speedups. You have three polynomial speedups nested one inside the next. And the reason I bring this up is to answer your question about, are we looking at turnkey algorithms? No, we're looking at libraries. So we're looking at building the algorithm primitives to make them available in beryllium such that they can be stitched together and yield compounding speedups because we think that's a much more direct path to a broader range of quantum speedups than if we're using them as isolated programs.

Okay, that actually makes a lot of sense. And I learn something every time I talk to you, Joe.

Thank you. Thanks, John. Good to talk to you.

We have no further questions at this time, so I'll now hand back to Dr. Joe Fitzsimons for closing remarks. Thank you.

Thanks, everyone, for joining us on our first earnings call. I appreciate it, and I appreciate the quality of the questions asked. I think we're well positioned now to start executing on our roadmap and really start pushing towards harnessing the full potential of quantum computing as we go forward. Thanks very much, everyone.

This concludes today's conference call. Thank you all for participating. You may now disconnect your line. Thank you.