Home / Transcripts / Xanadu Quantum Technologies Limited (XNDU) · August 5, 2026

Xanadu Quantum Technologies Limited (XNDU) Earnings Call Transcript

August 5, 2026

Toronto Stock Exchange CA Information Technology Software earnings 28 min

Earnings Call Speaker Segments

Operator operator
#1

Hello, and welcome to the second quarter 2026 earnings conference call for Xanadu Quantum Technologies. I'm Jerico, and I'll be moderating today's session. <<operator_script>> Just a reminder that this call is being recorded so it can be replayed. At this point I'd like to hand things over to Brett Harris, Vice President of Investor Relations. Please go ahead.

Brett Harris executive
#2

Thanks, and good afternoon to all of you. It's a pleasure to welcome you to the conference call covering Xanadu's results for the second quarter. With me on the line today are Dr. Christian Weedbrook, our Founder and Chief Executive Officer; Michael Trzupek, our Chief Financial Officer; and Rafal Janik, who serves as our Chief Operating Officer. You can listen to this call as it is streamed live, and afterward a recording will be posted on Xanadu's Investor Relations website. The information we discuss is time-sensitive and holds true only as of the date of this live webcast, which is August 5, 2026. During this call we may present forward-looking statements and information grounded in expectations we hold at present. Such statements and information carry substantial risks and uncertainties, so actual results could differ materially from the conclusions, forecasts and projections contained in those forward-looking statements and information. In reaching the conclusions and preparing the forecasts and projections reflected in the forward-looking statements and information, certain material factors and assumptions were applied. Anyone weighing an investment in our securities should review the discussion of those risks, factors and assumptions found in our periodic filings with the SEC and applicable Canadian securities regulators, including our most recent quarterly and annual reports. Except where the law requires it, management undertakes no obligation to update these forward-looking statements and information. At this point I'll hand the call over to Dr.

Brett Harris executive
#3

Christian Weedbrook, Xanadu’s founder and chief executive officer, is with us. Christian, over to you.

Christian Weedbrook executive
#4

Thanks, Brett, and good afternoon, all. At Xanadu, a single mission guides every call we make: to build quantum computers that are useful and available to people everywhere. That means a fault-tolerant quantum computer capable of solving real-world problems. The past few months have been eventful, with steady progress on the technology, partnership and fundraising fronts. I'll take you through where things stand on our hardware, our software and our partnerships, and then Michael will walk you through the financials. On the hardware side, we achieved further progress on optical loss. This is among the most important engineering challenges in photonic quantum computing and a critical driver of system performance on the path of fault tolerance. Specifically, this quarter delivered a new milestone: ultra-low-loss photonic chip packaging with an average edge coupling loss of just 0.085 dB per facet. That's a meaningful step forward in our ability to deliver high efficient, scalable quantum hardware. And that results from several things coming together at once. Our internal advanced photonic chip packaging facility, which we brought online last year, a close collaboration with Corning on customized fiber and fiber array solutions, and with DISCO on wafer singulation. Behind that packaging work, we accelerated the pace of fabrication itself.

Christian Weedbrook executive
#5

Within our foundry partners, we boosted fabrication runs across our two core material platforms. Thin-film lithium niobate rose roughly 75%, and silicon nitride climbed roughly 50%. With more tape-out volume, we can iterate more quickly on chip designs that feed straight into development of our qubit factory and speed up improvements to fabrication processes. Meanwhile, we keep putting investment into the infrastructure that will speed our development. In July, we announced a substantial expansion of our U.S. operations, anchored by our growing presence in Albany, New York. Albany has turned into a genuine hub for quantum computing and semiconductor research, providing us a strategic base near key partners and closer teamwork across groups. This forms part of a wider U.S. build-out. Our workforce here has expanded more than 5-fold since 2023. We expect that head count to rise significantly once more by year-end. Albany additionally provides us direct access to established U.S. semiconductor and photonics supply chains. That's a genuine benefit of our photonic approach, and we think it will help speed both our R&D and manufacturing timelines. On the software front, we keep finding ways to boost efficiency within the quantum computing stack. Our research team published an algorithmic breakthrough in Quantum Read-Only Memory, or QROM.

Christian Weedbrook executive
#6

There's a common subroutine that appears across algorithms for loading classical data into a quantum computer, and it's a bottleneck that hasn't seen any meaningful improvement in roughly 7 years. Our approach replaces iterative qubit swapping with a more efficient copy method and eliminates an unnecessary data unlocking set, which reduces the Toffoli gate operations required by about half — making quantum algorithms more efficient right now, or lowering the resource demands on the fault-tolerant hardware of tomorrow. This breakthrough is covered by a patent filing as proprietary Xanadu IP, and it's available today within PennyLane. We're also pushing the boundaries of quantum machine learning. Our team published new work that uses Fourier-based methods to train a quantum model with over 1 million parameters. It's large enough to accurately learn the distribution of ribosomal RNA, and it stands as an example of quantum techniques scaling to problems of genuine scientific interest. We shipped PennyLane version 0.45 and Catalyst version 0.15, and together these make it easier for our users to assemble complete quantum algorithms end-to-end. Also, the foundational PennyLane white paper has passed 2,000 citations, cementing a new industry standard. Each new researcher and developer building on PennyLane deepens both our commercial moat and our community moat. As we like to say, the people learning and building on PennyLane today are our partners and customers of 2029 to 2030.

Christian Weedbrook executive
#7

With Lockheed Martin, we're taking our current research partnership further through a joint quantum machine-learning initiative, tackling a major industry-wide bottleneck: talent. We unveiled a fresh program aimed at expanding quantum training via their Quantum Talent Pipeline, which provides engineers throughout the organization with access to PennyLane, our educational materials, and specialized workshops delivered by our team. Combined, these efforts are developing hands-on quantum capabilities for aerospace, defense, and advanced engineering. It stands as an example of how a top-tier organization can cultivate quantum readiness internally. We also expanded our collaboration with Los Alamos National Laboratory, where we're involved in their 2026 Summer School by co-supervising student projects. Additionally, we became a member of the Unitary Foundation, cementing a collaboration that has existed for a long time in support of open source quantum software and PennyLane. Regarding high-performance computing, we teamed up with Oak Ridge National Laboratory to make PennyLane available on the Frontier exascale supercomputer, among the most sophisticated classical computing systems globally. They introduced distributed computing capabilities into our Lightning simulator, allowing multi-node quantum simulation on Frontier's AMD-based hardware. We also conducted a practical workshop for the Oak Ridge Leadership Computing Facility community. We keep reinforcing ties throughout our partner ecosystem, renewing multiple enduring collaborations while broadening our involvement with the financial services sector.

Christian Weedbrook executive
#8

We renewed a multiyear partnership with Rolls-Royce centered on quantum algorithms for computational fluid dynamics and aerodynamics — 2 challenges that are fundamental to their engineering efforts. This stands as a strong illustration of how chronic computing can drive transformation, since these are precisely the types of problems at the heart of modern aerospace design. We're likewise seeing expanding opportunity within the financial sector. Our continuing research collaboration with the Fidelity Center for Applied Technology, FCAT, has involved work published jointly that adapts the hidden subgroup problem for practical data analysis. Outside of FCAT, we're engaged at an advanced stage with several major banks and quantum algorithms, and we keep actively scouting throughout the sector for high-value applications. We're modeling systematic risks, capturing multi-input correlations using quantum methods. With that, I'll pass it to Michael to take everyone through the financials. Thanks, Michael.

Michael Trzupek executive
#9

Thanks, Christian. I'll walk through where we stand financially, the capital moves we made during this quarter, and how operations performed. Revenue came in at $1.5 million for the quarter, versus $2.8 million in Q1 2026 and $1.1 million in the same quarter a year ago. The primary driver of revenue this quarter was DARPA Stage B. R&D expense totaled roughly $19.7 million, which is $2.4 million higher than Q1 of 2026. As we mentioned last quarter, we still anticipate that R&D spending will rise over time as we bring on more engineering talent and speed up wafer and chip production, and that expectation is very much unchanged. The modest quarter-over-quarter rise reflects growth in head count and stock-based compensation tied to bringing aboard additional manufacturing and engineering teammates as we scale our R&D capabilities. We expect R&D growth to pick up pace entering Q3 as we keep expanding the team and increase tape-outs. G&A expense came to approximately $11.1 million, compared with $9.8 million in Q1 of 2026. That sequential increase mainly reflected higher head count, stock-based compensation tied to annual grants, the synthetic ATM, Capital Market adviser fees, and other expenses related to being a public company. During this quarter, we recognized $2.3 million in nonrecurring fees connected with our SPAC combination and the launch of our synthetic ATM program.

Michael Trzupek executive
#10

The adjusted EBITDA loss came in at $21.3 million, versus a loss of $13.9 million in Q1 2026. That bigger loss was primarily the result of higher R&D and G&A spending, along with reduced grant revenue. CapEx investment rose to roughly $6.4 million during the quarter, largely reflecting spending on specialized equipment used to test, refine and scale our chip manufacturing capabilities. We anticipate CapEx will rise further in the second half of the year as we keep investing in R&D and manufacturing infrastructure. We closed the quarter with $312.8 million in cash. In May, we signed a standby equity purchase agreement with Yorkville Advisors, setting up a synthetic ATM equity facility of up to $300 million, which gave us the ability, though not the obligation, to issue Class B subordinate voting shares to Yorkville over a 3-year term, opportunistically depending on market conditions and valuation. During the quarter, we raised $67.2 million under the synthetic ATM facility, selling 5.5 million shares at an average net price of $12.28. As we said when this facility was announced, we intend to remain disciplined and only draw on it when we believe conditions favor shareholders and net proceeds flow directly to Xanadu's balance sheet as primary issuance.

Michael Trzupek executive
#11

Having said that, I'll now hand things over to the operator so we can move into the question-and-answer portion of the call.

Operator operator
#12

<<operator_script>> The opening question is from Tanu Chauhan of Rosenblatt.

Tanu Chauhan analyst
#13

This is Tanu, and I'm calling in for John. You've released figures concerning edge coupling losses, and you also discussed those earlier today. Could you say a bit about the 16 remaining components you had been working on to bring losses down?

Christian Weedbrook executive
#14

Absolutely, and thanks for asking. So, as you said, 16, 17 major components. Loss reduction is among our biggest challenges. We therefore examine the major components with an eye to cutting them down and bringing loss even lower. One route is architectural: take beam splitters — originally we required, say, x amount of them, as just one example. Now, for instance, we can eliminate half. That's one approach. Another approach concerns the platform, or the loss propagation of the substrate itself. We use 3 different types of substrates. As for the components you listed, they truly cover that end-to-end side of the computer. Anywhere the laser passes through will produce loss. So there are the qubit generation elements, the gate side of things, and the detector portion. Beam splitters are a classic example. Phase shifters are another. Ring resonators, which form a nonlinear light medium and let us build the squeezers, fiber to the chip coupling, the edge coupling you brought up, and then various detectors, along with their detector efficiencies.

Christian Weedbrook executive
#15

Consequently, that gives you 16 or 17 of those. Those represent merely a handful of illustrations of the components.

Operator operator
#16

The next person in the queue is Kingsley Crane with Canaccord Genuity.

William Kingsley Crane analyst
#17

Good to talk with everyone. I wanted to ask about Project Optimism. Is there any news on the chances of locking in that funding, or on when it might happen? Just any additional detail on the milestones would help.

Christian Weedbrook executive
#18

Yes. The most recent update we shared publicly is that it’s now in final discussions. There isn’t much more we can say, but perhaps the best way to put it is that these are final, final discussions at this point. So keep watching over the next month or 2, and we hope to have more information for you.

William Kingsley Crane analyst
#19

All right, that's quite useful. Then perhaps regarding the joint QML plus workforce training initiative you have going with Lockheed Martin — I'm curious about how it originated, how enthusiastic you are about that program, and what potential exists to spread the word about PennyLane, and about quantum computing more broadly as well?

Christian Weedbrook executive
#20

Sure, definitely. PennyLane has had Lockheed Martin as a user for quite a while. They were pretty enthusiastic about teaming up with us, initially on the QML front and more recently around workforce training. Each year, they aim to turn a certain count of their people — AI developers or other research engineers — into individuals capable of building quantum algorithms, yet the supply of in-house quantum specialists constrains them. As a result, they are drawing on our platform, our teaching materials, and PennyLane itself to speed that effort up at a very brisk rate. From our perspective, that is genuinely thrilling. We can place PennyLane in the hands of a great many additional developers, drive forward innovative algorithm work with one of our major partners, and keep expanding our educational offerings.

Operator operator
#21

The following question is from Nehal Chokshi at Northland Capital Markets.

Nehal Chokshi analyst
#22

Continuing further on the topic of reducing optical loss—0.085 dB per facet sounds like a genuinely strong figure, doesn't it? But could you provide some background on what that number was previously? Across the system, how many facets are there? And really, what I'm hoping to understand is how significant this particular factor is compared with the other success factors you're also aiming to lower. Is it proportional, or is it just proportionally much bigger than all the others here?

Christian Weedbrook executive
#23

That is a good question. I would say this is among the big ones, though not the sole one. As we said, there are 16 or 17 of them. A few were already — so, as you're aware, we have to reach levels below fault-tolerant that let you scale up, in principle, without limit. Several of these were already at the correct quantity of 9s when it comes to fidelity, which is genuinely impressive work by the team. So it's difficult to really gauge how impressive this particular one was relative to the whole picture. When you examine the actual fiber-to-chip coupling in this specific example, it is world class. So on its own, we believe it represents the lowest loss anywhere in the world. And so it is simply one more important building block for the overall system. One more thing worth mentioning, which we brought up previously when we went public, is that we'll be having a detailed loss road map and hardware milestone road map coming at the end of summer. So we remain on track for that. And the reason I raise that, if you recall the earnings, our first earnings call and also our investor deck, when we were public, we had loss reduction broken into 2 major paths.

Christian Weedbrook executive
#24

So really, if you think about that as a summary, all 17 elements sort of come together into 2 different paths or directions, and we had that up until the point when we went public. So we're hoping to have it by the end of the summer, extrapolating out to 2030, plus or minus, basically where we need to get to. So stay tuned for that, and that's a much better way to kind of examine the context of how everything fits together.

Nehal Chokshi analyst
#25

All right, a quick follow-up from me. Reaching the 2030 goal depends heavily on R&D engineers and the talent behind R&D. Since the company went public, has that pipeline of R&D engineering talent shifted in any material way?

Christian Weedbrook executive
#26

That is a fair question. I'd say it's affected us positively, though. We've brought on quite a few people, and I believe awareness of Xanadu has grown since we went public, and we've seen much more inbound interest. Equally important, we're better positioned to draw in even stronger talent to add to the excellent team we already have. A big part of that is that compensation can now shift — you're able to offer RSUs and similar things, plus eventual liquidity. So overall, I'd say it has genuinely benefited us, beyond the capital: the visibility has led to things such as better recruiting, and even though we already had truly outstanding people, we're seeing a bigger pool of talent applying for our roles. So all in all, I'd say yes, it has changed for the better.

Nehal Chokshi analyst
#27

That's great. So would it be fair to say we should expect the pace of R&D engineers getting onboarded to Xanadu to speed up even more?

Christian Weedbrook executive
#28

Absolutely, and I don't want to make it sound like this is simple, either. We always aim — we try to take pride in telling it straight, and it remains a very different kind of challenge. But the more things we can improve, such as recruiting and bringing in more excellent people, the more that risk goes down. So yes, definitely. It's still a significant challenge, yet having additional staff and especially additional capital lets us reach our ultimate goal of a large-scale quantum computer.

Nehal Chokshi analyst
#29

Alright, so for my final question: your press release references a seminal PennyLane paper with 2,000 citations. I'd like to know when that seminal PennyLane paper was actually published, since reaching 2,000 citations sounds truly remarkable if it happened only recently.

Christian Weedbrook executive
#30

Yes. It initially appeared—the first version was placed on the archive. As you’re aware, you’re able to post items such as a preprint there, and that happened roughly 8 years back, with formal publication likely coming a year later. But broadly speaking, citations can begin accumulating the moment it’s posted to the archive. So about 8 years.

Nehal Chokshi analyst
#31

Understood. Alright. So turning to the traceable metric you shared previously — the count of 30-day active users — has that shifted materially since the start of the year? You had a metric on that.

Christian Weedbrook executive
#32

Yes, certainly. That figure keeps climbing. We are observing an increase both in active-user counts and in how many universities around the globe we are working with, plus how many corporate organizations we are able to collaborate with. So in terms of adoption, PennyLane keeps expanding. We aren't disclosing the precise figures at this moment, but that is something we will examine as we head toward our Analyst Day.

Operator operator
#33

Todd Coupland of CIBC has our next question.

Thomas Ingham analyst
#34

Could you share your thoughts on how optical loss has performed up to now, and maybe rank that against what DARPA C anticipated, and also include an update on where you believe you could be by year-end, when you expect an announcement on that?

Christian Weedbrook executive
#35

Todd, appreciate that. Perhaps I'll take the last item first. We're in good shape for Phase C, though we can't promise anything, but I'd say we're meeting every requirement, and things look good. Still, nobody knows until you know. So we'll certainly keep you posted on that. Regarding the loss, I think going back to what I mentioned, the plot of L1 and L2, just think of them as two separate quality paths that are independent, where the layer travels, and you pick up loss each time it propagates through the path. It's really a function of a loss gap more than anything. And you think of it as a loss gap because you can actually reduce the loss through the physical implementation. So doing more and more chip runs. I think from the press release, you would have seen that we've increased by 50% and 75% two different foundry runs. So think of that as ultimately loss reduction and speeding that up, which is great. So it's more of a gap where you improve the hardware, but also improve the bounds through theoretical work. So given that you can think of it like a gap, the gap has actually decreased in the last 4 years up to 200x, I believe, down to where we are now.

Christian Weedbrook executive
#36

There's still roughly a 5 to 10x reduction, a 5 to 10 improvement, remaining before we reach what you might call the promised land. So we remain on track for that. And as mentioned, we will certainly share more specifics on where we stand during the Analyst Day call when we hold it.

Thomas Ingham analyst
#37

Thanks. My next question is about how OpEx and CapEx are expected to grow during the year's second half. What should that look like compared with the figures you just reported?

Michael Trzupek executive
#38

Yes. So, as we discussed last quarter, we're still not giving explicit guidance on that, but we are giving high-level guidance, which is that we're certainly increasing — as we mentioned on some of the other questions — our engineering talent, and we're also increasing how many tape-outs we do. You're beginning to see a little bit of that and the growth on our balance sheet this quarter. And I think you're going to see that grow substantially in the second half of the year. And by the time we get to our Analyst Day, we hope to give you more color then.

Operator operator
#39

Moving on, we have a question now from Paul Treiber of RBC Capital Markets.

Paul Treiber analyst
#40

On the fab runs topic, you brought up the 50% rise and the 75% rise. What run rate are we at right now on an annual basis for fab runs? And then how big an increase should we expect relative to where we are today?

Christian Weedbrook executive
#41

Yeah, good question. So we use two foundries, and what we have is something called a WIP corridor. In simple terms, there's a minimum wafer count you can access at any given moment. For example, at NY CREATES — they handle our silicon nitride, which is mainly what we use to make our qubits. At the beginning of last quarter we had roughly 100 wafers there. That has since risen 50%, to 150 wafers, and it will probably climb further, based on what Michael mentioned as well, since we're investing more to reduce that loss. Then there's another one, Wavetek. That's a subsidiary of UMC in Taiwan. We have around 100 there as well, a bit more actually, and that figure has now increased to 175 wafers per month, and it's also likely to rise again next quarter.

Operator operator
#42

That brings us to the end of the question-and-answer session. At this point, I'll hand things back to Christian Weedbrook, who has some closing remarks for us.

Christian Weedbrook executive
#43

Thanks so much, everybody. It's really wonderful to receive those questions. And we're always grateful that all of you join us for this earnings call. One quick reminder: our mission is to create quantum computers that are genuinely useful and accessible to people all around the world — something very much shaped by our photonic-based approach together with our PennyLane software. And maybe one last sort of reminder that in our view, thinking over the long term is cumulative. We don't measure it quarter by quarter, even though we're glad to sort of produce the earnings and quarterly reports and keep you posted on the terrific progress we're making. But it's something we're targeting for the end of this decade, as we said, 2029, 2030. And we look forward to staying connected, and thank you all for your time.

Operator operator
#44

That wraps up today's call. We appreciate you being with us. You're free to disconnect at this time.

Read the full transcript via the API

You're viewing the first half of this call. Get the complete Xanadu Quantum Technologies Limited transcript - plus 132,000+ transcripts from 7,000+ companies, speaker segments and full-text search - through the EarningsAPI REST API or hosted MCP server.

Get an API key View API docs →

For developers and AI pipelines

Programmatic access to Xanadu Quantum Technologies Limited earnings transcripts and 132,000+ others is available through the EarningsAPI REST API and the hosted MCP server. Quarterly plans from $162 - full transcripts, speaker segments, full-text search, and the /api/v1/transcripts/recent polling endpoint for ETL pipelines.