Horizon Quantum Holdings Ltd. (HQ) Earnings Call Transcript & Summary

Good day, and thank you for standing by. Welcome to the Horizon Quantum Second Quarter 2026 Report Conference Call. [Operator Instructions] Please be advised, today's conference is being recorded. I would now like to turn the conference over to Katherine Bailon. Please go ahead.

Thank you, Lisa. Good morning, and thank you for joining us on Horizon Quantum's Second Quarter Fiscal 2026 Earnings Call. My name is Katherine Bailon. 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 second quarter ended June 30, 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 nonhistorical statements that I or the management team make during this call will constitute forward-looking statements within the meaning of U.S. federal securities laws, 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 20-F and other filings with the SEC 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 forward-looking statements made on this call, whether because of new information, future events or otherwise. Any forward-looking 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, August 4, 2026. This discussion includes non-GAAP financial measures. These measures are calculated by management and do not have any standardized meaning 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'll now turn it over to Horizon Quantum Founder and CEO, Dr. Joe Fitzsimons.

Thank you, Katherine, and thanks, everyone, for joining us on the call today. We've made quite a good bit of progress this quarter, and I look forward to running through it with you. Just to reiterate our mission, which we've been laser-focused on, our focus is really on building tools to unlock broad quantum advantage, building the infrastructure to allow developers to harness quantum computers to do real work and to create real value in the real world. As many of you know, we're taking a software-first approach on this. We're essentially building up programming languages at increasing levels of abstraction with the ultimate goal of allowing developers that are not experts in quantum computing to be able to harness quantum computers to tackle hard computational problems. I'm going to focus first on the R&D and operational highlights, in particular, on Beryllium. Beryllium, which is our object-oriented quantum programming language and our highest level of abstraction that's currently available is now available in early access to the -- for the group's early access users you have. This brings familiar constructs from the world of classical software development, such as classes, structs, scope functions and so on into the world of quantum computing. This will really be a catalyst, at least in my view, in enabling developers to really harness quantum computers to do much more serious work, to be able to construct much more complex programs than is otherwise practical. We've also made a lot of progress in developing libraries for Beryllium. One of the main advances -- sorry, one of the main advantages we have in building an object-oriented programming language is that we are -- it's very natural to have libraries to enable code reuse so that one developer can build classes, can build functions, can build components that another developer can then use. And that's really important, first of all, in building up a rich developer ecosystem, but also in enabling the abstractions that we seek, in enabling developers that are experts in quantum computing to build up components that can then be used by experts in the various fields that stand to be disrupted by quantum computing, but who are perhaps not experts in quantum computing themselves to harness quantum computers to tackle problems they care about. We've had really significant progress in building up the standard set of libraries there, including standard algorithmic building blocks, things like Quantum Monte Carlo algorithm and so on. Turning to our superconducting taskbed. Our Ember-1 system has now been opened up to first users. This is a superconducting quantum computer that we operate out of Singapore based on a chip from Rigetti. One of the advantages we have in operating this hardware ourselves and in integrating Triple Alpha, our software directly with its control systems, is that we are able to start enabling real-time execution of Hydrogen programs, which is our kind of portable assembly language, without having to rely on post-selection. This means we're able to do the full set of operations you would expect from a mature quantum computer. These are things like loops and recursive function calls as well as concurrent classical function evaluation without having to turn this into some kind of hybrid program where we're going back and forth between a classical computer and a quantum computer. So we can start to enable this in real time. This goes far beyond the limitations of quantum circuits, which is where many of the existing quantum programming frameworks focus. Turning to hardware collaborations. We have announced a new collaboration with Quantum Machines, that is aimed specifically at developing embedded calibration, both for use in our Ember-1 testbed system here in Singapore, and also potentially in other systems. The goal really here is to develop a set of calibration routines that can run regularly as the system is operating rather than having to do a giant batched calibration run at fixed intervals. So we're essentially aiming to maintain the system at its highest level of calibration continuously throughout its availability. We have also announced that our second testbed system, that's a 256-qubit IonQ system that will be coming in 2027, will be installed in Dublin. So we have operations both in Singapore and Dublin. And with this announcement, we will have systems operating in both locations. I also wanted to turn to recent progress in AI. And this is something that I wanted to flag. At the moment, the quantum computing community is coming to terms with the fact that recent frontier models, in particular, Fable and Sol have started to yield real results in quantum information. And in some sense, in the academic community, there's a bit of an existential crisis going on at the moment, trying to think through where the role of researchers sits relative to these AI systems. In particular, we have seen an announcement from OpenAI of a strong parallel repetition theorem, which is extremely important in the context of quantum information theory. As a company, we've been early to invest in AI agentic system and a dedicated team back in 2023, and we've been building up our own capabilities in agentic systems within the company to accelerate our R&D since then. With the availability of these new frontier models, it is starting to become increasingly clear that autonomous scientific discovery is becoming a possibility. And that means that it's becoming possible to harness these systems to make scientific progress without substantial human intervention. That's really interesting for us. It really potentially allows us to accelerate our road map. And in particular, in the last few weeks -- sorry, in the last few weeks, since these models have been released, we've seen our own agentic systems produce hundreds of research notes on open problems in quantum computing. Turning to applications. One of the defining questions, one of the most important questions for the quantum computing community is how quickly quantum computing will be adopted by industry. How quickly will we see quantum computing move from proof of concept to actual production deployment? And will particular industries be fast or slow to adopt quantum computing? I think that's essentially existentially important for the industry. What we have done is we've taken our first steps to really not leave this to chance, but rather turn it into an opportunity. And what we have done is taken the step of establishing a quantum applications team that is focused on 3 specific high-value problems in each of 3 industries. And the reason we're doing that is to ensure that progress is being driven on industry relevant problems and to ensure that we can drive adoption as it becomes possible to solve hard problems with quantum computers, but also to ensure that we stand to capture some of the value of that opportunity. We are not focused on doing proof of concept application with industry, but rather doing the entire end-to-end development within Horizon. You can perhaps think about this as creating a full chain from application through our software infrastructure with Triple Alpha and our compilers through our runtime environment down to execution on our testbed systems. This will allow the internal team to really make a sustained effort towards achieving real quantum advantage on problems that will -- that stand to deliver real value. The existing AI team that we have that have been working on agentic workflows is going to be rolled into that applications team to accelerate progress. Turning to the financial side. Our cash position has strengthened significantly. We have in Q2, $27.5 million in gross proceeds from the exercise of warrants, which is significantly strengthening our balance sheet. The total cash and cash equivalents have reached $113.3 million, significantly fortifying our balance sheet and financial runway and to -- allowing us to support increased investment in R&D. I'm going to hand over to Greg for more details on this. But before that, I want to address the net loss. So this quarter, we'll show a large net loss of $115.2 million. This is driven primarily by a $108.3 million noncash charge that's related to the remeasurement of warrant-related derivative liabilities. So this does not affect our runway in any way. In terms of operating expenses, there's increased operating expenses in R&D and G&A. And this is primarily related to the transition to a public company and scaling of the research and development team to meet the demand from collaborations. With that, I'd just like to summarize that we've made substantial technical progress with Beryllium and the testbed. We're tracking well against the milestones we set forth in the F-4, and we've increased our runway due to the proceeds from the exercise of warrants. With that, I'll hand it over to Greg.

Thanks, Joe, and good morning, everyone. I'd like to start off by reviewing our financial results in a bit more detail. But before I do, I'd like to mention that all financial results discussed here are in U.S. dollars. We are pleased to receive the cash infusion. As of August 3, Horizon has received notices from public warrant holders to exercise approximately 2.5 million common stock warrants, and that represents about 79% of the public warrants outstanding. The exercise of these warrants has generated gross proceeds to the company of approximately $28.7 million to date. This infusion of cash has further extended our operational runway. Now for the second quarter ended June 2026, a total of, as Joe mentioned, 2.4 million public warrants were exercised, and that generated $2.7 million (sic) [ $27.5 million ] in gross proceeds and combined with existing cash resulted in cash at the end of the June quarter of $113.3 million. This cash level is a net increase of $16.7 million from the $96.6 million as of the end of the March quarter 2026. Now total operating expenses for the second quarter of 2026 were approximately $7.2 million, and that compares to $2.8 million for the second quarter of 2025. This increase in operating expense was largely the result of hiring, primarily on our science and engineering teams. R&D expenses consist primarily of personnel-related costs for scientists, software engineers and technical staff engaged in the design, development and testing of our software and hardware systems. To a smaller extent, it also includes software and other cloud services and costs associated with the operation of our hardware testbed. For the second quarter of 2026, R&D expenses increased by approximately $1.4 million or 117% to approximately $2.6 million, and that compares to $1.2 million in the second quarter of 2025. Excluding share-based comp and nonrecurring comp adjustments, the period-over-period R&D increase was 100%, and that's attributable to increases in headcount and to a much lesser extent, hardware testbed setup costs. For the second quarter of 2026, sales and marketing expenses increased by approximately $0.13 million or 51% to approximately $0.38 million, and that compares to $0.25 million in the prior year period. Excluding share-based comp, sales and marketing expense rose 63% year-over-year, and that's due to increased trade show activity and industry engagement. G&A expenses for the second quarter of 2026 totaled approximately $3.8 million, increasing $2.7 million or 236% from $1.1 million in the second quarter of 2025. Excluding share-based compensation, nonrecurring comp adjustments and onetime business combination expenses of about $1.5 million and $0.3 million from the second quarter of '26 and the second quarter of 2025, respectively, G&A increased 191% period-over-period. This increase was primarily due to increases in headcount and other operating expenses, largely associated with our transition to a public company. For the second quarter of '26, our loss from operations totaled approximately $7.2 million, an increase of $4.5 million, or 267% from a $2.7 million loss in the second quarter of 2025. This increase was primarily due to hiring of staff across the company, including scientists, engineers as well as staff supporting public company operations. The company reported a net loss of approximately $115.2 million or $2.20 per ordinary share for the second quarter of 2026, and that compares with a loss of $2.9 million or $0.07 per ordinary share in the second quarter of 2025. As Joe mentioned, the second quarter result included a $108 million noncash loss from the remeasurement of warrant-related derivative liabilities. The increase in the company's share price during the quarter increased the fair value of these liabilities, resulting in the accounting charge. Importantly, this charge did not affect the company's cash balance. In fact, by contrast, the voluntary exercise of warrants during the quarter generated significant cash proceeds and increased share capital. I'd like to point out that all of the public and private warrants of Horizon Quantum that remain outstanding will continue to be remeasured at fair value each fiscal quarter. Accordingly, future movements in the trading price of Horizon Quantum's Class A ordinary shares could result in additional noncash gains or losses. Because these fair value adjustments can create substantial volatility in reported earnings without affecting underlying operating performance or cash flow, management also evaluates results on an adjusted EBITDA basis, a non-GAAP measure. Adjusted EBITDA excludes changes in fair value of derivative liabilities as well as share-based compensation, nonrecurring expenses associated with the business combination. On this basis, the company reported an adjusted EBITDA loss of approximately $5.5 million for Q2 2026, and that's compared to an adjusted EBITDA loss of $2.1 million in Q2 2025. In terms of employees, we ended the second quarter of 2026 at 57 full-time employees, and that's versus 34 at the end of the second quarter a year ago, reflecting a 68% increase period-over-period and a 10% increase sequentially from the quarter ended March 2026. Historically, scientists, engineers and other technical staff have represented 60% to 80% of the total headcount, and we remained within this range as of the June quarter 2026. Going forward, we expect this mix to remain consistent. In terms of cash flow, net cash in operating activities for the second quarter of 2026 was approximately -- used was $5.1 million, and that compares with the cash usage of $2.0 million in the second quarter of 2025 and a usage of $4.2 million in the first quarter of 2026. The bulk of the period-over-period increase in cash usage was due to growth in headcount. For the 6 months ended June 30, 2026, net cash used in operating activities totaled approximately $9.3 million versus $3.7 million for the first 6 months of 2025. 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 Alpha's technical potential is for our science and engineering teams to remain focused on the execution road map. 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 road map. 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'd like to share our thoughts on our cash balance, which stood at $113.3 million at the end of June. Since we are a software company, our model is a capital-light one. We believe our current cash balance provides us with significant runway as it is our intention to proceed during this pre-revenue period with fiscal discipline and rigor. And with that, I'll turn it back over to Joe.

I'd like to leave you all with 3 thoughts. With the availability of Beryllium to our early access users and the opening of Ember-1 to the first external users, we continue to execute as planned against our road map. With the establishment of our applications team, we're positioning the company to push towards Quantum Advantage for 3 high-value problems. And finally, the exercise of the warrants in the previous -- in Q2 has fortified our balance sheet and extended our runway, giving us the resources to aggressively pursue our plan. With that, we're happy to take your questions.

[Operator Instructions] Our first question today will be coming from the line of Quinn Bolton of Needham & Company.

Congratulations on the progress in the second quarter. I guess, Joe, I wanted to start off with the early access on Beryllium and hoping you can provide some thoughts on what kind of interest you're seeing from early access users, what kind of feedback you're receiving, but just -- kind of just the initial thoughts on how that Beryllium rollout is progressing with your lead customers.

Quinn, thanks very much for the question. So early access is still a relatively small group at the moment. It's mostly hardware partners. So we're just at the early stages of rolling that out. Similarly, with the testbed access, we've only recently granted access to that system. So we're still only seeing initial feedback. We're still debugging some of the system and making sure that we have it up to scratch. However, we've seen quite a lot of interest in Beryllium itself. It obviously extends the capabilities of what you can do on a quantum computer quite far beyond what is easily possible in other ways. So at least from that perspective, we've seen quite a lot of interest in using it. We've also had an active intern program here in the offices over the course of the summer, where the interns have been using this even before it was rolled out to early access users. And we've seen some really interesting things built with Beryllium, where the object-oriented nature of it has essentially allowed much more sophisticated programs than I have seen in any other programming framework to date.

And I guess I was going to ask a follow-on question, but you just somewhat touched on it, but you had mentioned in the script in response to this question, just how is the progression on the libraries going? Is most of that work today being done by Horizon? Are you starting to see initial Beryllium customers beginning to contribute to those sort of open source libraries to help build that ecosystem?

So I mean, what I would say is, at the moment, it is primarily internal to Horizon. We've only recently opened up the early access. So it was towards the end of the quarter. So we're only just seeing first use by external users. In terms of building up the library's ecosystem, at the moment, that is done by re-uploading code. It's obviously something we're looking at internally to make that process as smooth as possible to be able to really build up a community where developers can not only build libraries for their own use, but make them easily available to other Triple Alphas -- sorry, to other Triple Alpha users.

Got it. And then last question on the 3 applications you're developing. Can you give us some sense of what's the time line for development of those applications? My guess is it may be sort of a rolling process, but just when would you expect to complete some of the initial work on those applications to be able to start to engage customers on those 3 specific use cases?

So what I would say on this is that this is a long-term effort. It's not aimed at having an application to sell to users directly in the short term. It's intended to give us a pathway to capture value in the company from those applications and to drive the adoption of quantum computing with quantum advantage into those industries. So we're focused on 3 specific industries. We're not going to go into details of the specific problems we're focused on because obviously, we don't want to create a lot of competition for ourselves on those problems. But the purpose of pursuing these problems is not to build an application to sell. It is to build -- it's to achieve a real quantum advantage internally on high-value problems.

That you would then sort of partner with customers to sort of deploy. So it's more developing these applications that then use your software infrastructure, and long-term, that's how you would monetize it rather than sort of selling algorithms.

So what -- the way we have chosen these problems is essentially to choose problems that are essentially directly monetizable in and of themselves. So our purpose in this is not to build applications to sell. It is to build applications to drive adoption of quantum computing in industries.

And our next question is coming from the line of Tanu Chauhan of Rosenblatt Securities.

That's actually John. Tanu is my associate, but good to listen to you guys. Joe, Greg, Katherine, congrats on your second quarter as a public company and the progress. I have a few questions. Can you hear me okay?

Yes.

All right. Great. Calibration, I know is burdensome. You have an agreement with Quantum Machines. I think that makes a lot of sense. Could you elaborate a bit on how that will work? And is there some potentially revenue generation there from an IP standpoint? Or how should I think about that?

So essentially, we're working with Quantum Machines to develop embedded calibration, starting with our testbed system here, Ember-1. The purpose of this is to overcome the fact that you tend to need to take systems down for a significant period of time or rather make them unavailable to users for a significant period of time in order to calibrate them to calibrate the frequencies, to calibrate the shapes of the individual pulses that get sent into the processor to control quantum operation. And that drifts over time. So it's not a one-off process. This can drift as temperatures in your fridge shift or as there is changes in the lab environment if there's all sorts of changes can lead to a slightly different environment in the processor that changes the behavior of the qubits. And so you end up needing to recalibrate the system. What we're trying to do is avoid having to take the system offline or at least to make it unavailable to users to recalibrate at regular intervals. And instead, what we're looking at doing is embedding that calibration, so it's essentially running all the time in the background so that the system is maintaining its highest calibration. Are there opportunities to potentially derive revenue from this in the future with the network with Quantum Machines? Potentially, there is. But at the moment, we're focused on getting this working with our own system first.

Okay. I mean it sounds very useful. I know people complain about that process quite a bit you got to take the thing to a higher temperature and then you got to cool it back down, the whole thing. And then relative to the 256 delivery in Dublin, you've chosen the location, I think. Could you elaborate a little bit on that and when that might get delivered?

Sure. So the system itself is coming next year. We believe -- well, we expect installation to start next year. We're currently selecting the location to place it, but it will be in Dublin. We're building up a new premises there. As things currently stand, we're kind of exceeding the space availability in our current office there. So we're in the process of selecting a new office location in Dublin. And of course, with that, we'll have to build out the lab facilities.

Great. And if you could give me one more on error correction. How are -- what are you guys doing there? I think we've talked about how some software developers are optimizing some of the machine time access to improve the level of error correction as they write their algorithms. So I'm curious what you guys are doing at Horizon relative to error correction.

Yes. So error correction is one of the most important topics in quantum computing. Ultimately, these systems are very susceptible to error and we need to squeeze those errors out faster than they can accumulate. It may sound like a hardware problem, but actually a lot of it is a software problem. We've been tackling this in a couple of different ways. So one of the things that we have done with opening up access to the Ember-1 system, first of all, what we have done is to start enabling real-time execution of branching programs based on measurement results as well as the real-time evaluation of simple classical functions. And that gives us many of the ingredients that you need to start to do real-time error correction rather than in post-selection. So a lot of error correction experiments you see are post-selected, meaning that the error correction is not done in real time. Instead, the encoding is done, it's run many times on a system and then a subset of the results are picked out that have particular measurement outcomes. That pattern match the correction. In practice, we need to be able to do this in real time. So by building up that real-time execution engine to be able to run directly on our control systems, that's really starting to enable us to write these programs that can express full error correction in real time and execute them through the control systems. The other side of this in terms of what we're doing, one of the things we've put quite a lot of effort in that we haven't talked about so much to date has been in the development of what we call custom compiler layers. And the intention with our custom compiler layers is to enable developers and to enable hardware collaborators to build their own compilation workflow using our tools. And we are building towards enabling error correction through that, being able to include an error correction library for different codes and to be able to compile down into error correction codes into fault-tolerant encodings that can then be executed on hardware. Of course, where the whole industry sits with respect to error correction, it's still a very nascent technology. So it's not at the point where you have very large numbers of error corrected qubits that can be easily harnessed, but we're pushing forward on that front.