This conference call transcript was computer generated and almost certianly contains errors. This transcript is provided for information purposes only.EarningsCall, LLC makes no representation about the accuracy of the aforementioned transcript, and you are cautioned not to place undue reliance on the information provided by the transcript.

Rigetti Computing, Inc.
3/5/2025
Thank you for standing by and welcome to Regetti's fourth quarter and full year 2024 earnings conference call. At this time, all participants are in a listen-only mode. After the speaker presentation, there will be a question and answer session. To ask a question during the session, you will need to press star 1-1 on your telephone. To remove yourself from the queue, you may press star 1-1 again. I would now like to hand the call over to Subodh. Kakani, President and CEO, please go ahead.
Good morning, and thank you for participating in Rigetti's earnings conference call covering the fourth quarter and year ended December 31st, 2024. Joining me today is Jeff Bertelsen, our CFO, who will review our results in some detail following my overview. We will be pleased to answer your questions at the conclusion of our remarks. We would like to point out that this call and Rigetti's fourth quarter and year ended December 31st, 2024 press release contain forward looking statements regarding current expectations, objectives, and underlying assumptions regarding our outlook and future operating results. This forward looking statements are subject to a number of risks and uncertainties That could cause actual results to differ materially from those described and are discussed in more detail in our Form 10-K for the year ended December 31st, 2023, our Form 10-Q for the three and nine months ended September 30th, 2024, and other documents filed by the company from time to time with the Securities and Exchange Commission. These findings identify and address important risks and uncertainties that could cause actual events and results to differ materially from those contained in the forward-looking statements. We urge you to review these discussions of risk factors. Today, I'm pleased to provide an update and report on our progress at Rigetti Computing. Rigetti recently entered into a strategic collaboration agreement with Quanta Computer Inc. a Taiwan-based global Fortune 500 company, and a global leader of computer server manufacturing with the goal of accelerating the development and commercialization of superconducting quantum computing. Digeti and Quanta have committed to investing more than $100 million each over the next five years pursuant to the collaboration agreement, with both sides focusing on their complementary strengths to develop superconducting quantum computing technologies. In addition, Quanta will also invest $35 million to purchase shares of Rigetti common stock subject to regulatory clearance. Quanta's collaboration with Rigetti is designed to strengthen our position in this flourishing market. Our company's complementary strengths, Rigetti as a pioneer in superconducting quantum technology with open modular architecture, enabling integration of innovative solutions across the stack, and Quanta as the world's leading notebook server manufacturer with $43 billion sales, will support us in our goal to be at the forefront of the quantum computing industry. On the sales front, I'm pleased to report that we sold a Novara QPU to Montana State University in December 2024, which was our first QPU sale to an academic institution. The NOERA will be located at MSU's Q-CORE to educate and train scientists and engineers on quantum computing technologies, in addition to being used to create a testbed for quantum computing R&D. MSU's Q-CORE is a new center of excellence for quantum enabling technologies established to accelerate workforce development and the regional quantum innovation ecosystem. I should also note that in addition to the MSU sale, there was an additional Novera sale in the fourth quarter to the UK government. On the technology front, we launched our 84-qubit ANCA III system in December 2024. ANCA III features an extensive hardware redesign that enables superior performance. We achieved major two-qubit gate fidelity milestones with ANCA III. successfully having the error rates in 2024 to achieve a 99.0% median ISWAP gate fidelity and demonstrating 99.5% median fidelity with F-SIM gates. Our newest flagship quantum computer continues to feature Regate's scalable, industry-leading chip architecture with 3D signal delivery while incorporating major enhancements to key technologies. ANCA III is available to Regate's partners via Rigetti quantum cloud services platform and to the general public via Microsoft Azure and Amazon Bracket. In other developments, AI power tools from quantum elements and crews remotely automated the calibration of a Rigetti QPU integrated with quantum machines control system. This work was part of the AI for Quantum Calibration Challenge hosted at the Israeli Quantum Computing Center. The two companies participating in the challenge, Quantum Elements and Cruise, automated the calibration of a 9-qubit Regattino Vera QPU, integrated with Quantum Machines' advanced OPX1000 control system, and NVIDIA DGX Quantum, a unified system for quantum classical computing that NVIDIA built with Quantum Machines. This achievement showcases the potential of AI in quantum computer calibration and also highlights the growing collaboration within the quantum computing ecosystem. Quantum Elements crews and quantum machines are members of Rigetti's Nowhere RQPU partner program, an ecosystem of quantum computing hardware, software, and service providers who build and offer integral components of a functional quantum computing system. We believe that another advantage we leverage is our modular approach to developing our technology. By enabling our partners to integrate their technology with ours, we can explore and advance creative and flexible ways to improve quantum computing capabilities. In summary, we believe that superconducting qubits are the winning modality for quantum computers, given their fast speeds and scalability. We have developed critical IP to scale our systems and remain confident in our plans to scale to 100-plus qubits by the end of the year with a targeted 2x reduction in error rates from the error rates we achieved at the end of 2024. We believe our leadership in superconducting quantum computing continues to be reinforced as we push the boundaries of our system performance, as evidenced by the success of AMCA III. Thank you. Jeff will now make a few remarks regarding our recent financial performance.
Thanks, Subodh. Revenues in the fourth quarter of 2024 were $2.3 million, compared to $3.4 million in the fourth quarter of 2023. Revenue is an important part of our strategy to fund our ongoing research initiatives. Renewal of the U.S. National Quantum Initiative, sales to U.S. and foreign governments, and NOVERA are all important to future sales. Gross margins in the fourth quarter of 2024 came in at 44% compared to 75% in the fourth quarter of 2023. The lower gross margins on a year-over-year basis were due to ongoing revenues from our contract with the UK's NQCC to deliver a 24-qubit quantum system, which has a lower gross margin profile than most of our other revenues. On the expense side, total OPEX in the fourth quarter of 2024 was 19.5 million compared to 19.7 million in the same period of the prior year. Stock compensation expense for the fourth quarter of 2024 was 3.4 million compared to 3.7 million for the fourth quarter of 2023. Net loss for the fourth quarter of 2024 was $153 million, or 68 cents per share, compared to a net loss of $12.6 million, or 9 cents per share, for the fourth quarter of 2023. The non-cash change in the fair value of derivative warrant and earn-out liabilities negatively impacted our net loss for the fourth quarter of 2024 by 135.1 million. compared to a favorable impact of $4.6 million in the comparable prior year period. The derivative warrant and earn out liabilities are non-cash in nature and Regetti will never be required to pay cash to settle these obligations. Cash, cash equivalents and available for sale investments totaled $217.2 million as of December 31, 2024. During the fourth quarter of 2024, we received net proceeds of 153.3 million from the sale of 88.1 million common shares through a registered direct offering and completion of our at-the-market equity offering. We also prepaid in full all of the remaining amounts owed under our loan agreement with Trinity Capital Inc. We believe that our existing balances of cash, cash equivalents, and marketable security should be sufficient to meet our anticipating operating cash needs for at least the next three years based on our current business plan and expectations and assumptions considering current macroeconomic conditions. Thank you. We would now be happy to answer your questions.
Thank you. As a reminder, to ask a question, you will need to press star 11 on your telephone. To remove yourself from the question queue, you may press star 11 again. Please stand by while we compile the Q&A roster. Our first question comes from the line of Craig Ellis of B Reilly. Please go ahead, Craig.
Yeah, thanks for taking the question and congratulations on the partner progress and the Novera QPU sales. Sue, but I wanted to start off by following up on the Quanta announcement. And the question is this, can you help us understand the deal's genesis? Were you reaching out to them, they to you? And over what time period has this deal been gestating? And is there anything exclusive about any of the technology that you or they would develop underneath the agreements?
Thanks, Craig. So yeah, it's an exciting partnership announcement we did with Quanta Computer. As our announcement said, Quanta is a large company based in Taiwan with almost $43 billion in annual sales. They are well known for both their laptop as well as server manufacturing. I believe they have the number one market share in GPU servers right now. So they are a close partner of companies such as Nvidia, Apple, and many other companies. They on their own have been looking around for how is the best way for quantum computer to get into quantum computing because they clearly view quantum computing as the next big thing after GPUs. And they've always done that. They have always been on the leading edge of the new technology curves. So they were searching for the right partner. They did their own homework, looked at all the different modalities, decided that superconducting gate-based quantum computing is the most likely modality to win. Within that, we are clearly a leader competing right at the top along with IBM and Google. So we don't know all the companies that they talk to, but they certainly started talking to us close to a year ago or so. So we have been discussing with them for a while now. On our side, we know that we cannot be building the commodity type items of the hardware stack in Berkeley or Fremont, California. Doesn't make any sense given the cost structure in those places. So we were looking for appropriate contract manufacturers in the long term as volumes pick up who's the right contract manufacturer for us. So it was a mutual decision where we thought they are the right potential partner given the critical role they play in CPU, GPU servers today. They believe that we are the right partners from a technology standpoint. Clearly one of our needs was to have money and that's what they offered is as you can see They are buying $35 million of our shares at $11.59 pending regulatory clearance. But more importantly, they have committed to more than $250 million over the next five years to be invested in the non-QPU portion of the hardware stack. So essentially going forward, we will continue to focus, as we always have, on the quantum chip fabrication part, and we'll be responsible for the whole stack. but we will start relying on them as our contract manufacturer for things such as the control system, the dilution refrigerator, cables, and all other accessories that are extremely important, but just not that high value add. So we believe it's the right partnership in the long term. We certainly are counting on them to help us out on that part. And certainly between the cash we already have plus the $35 million they will give us for shares plus the $250 million commitment, We effectively have close to $500 million right now to be invested in the next five years. So we feel really good about that position and how we can deploy that investment to accelerate the pace of our quantum computing development. Hopefully that answers your question.
Yeah, that's very helpful, and I think it is significant that as a leader in first x86 servers and now GPU-based servers, they've chosen to partner up with you and certainly with significant financial commitment. The second question I wanted to ask was related to your take on where we stand with government funding. Recently, there was a bipartisan bill introduced by Senators Daines and Durbin regarding $2.5 billion in potential funding for the U.S. government. Can you just give us an update on where things stand federally in the U.S. and and your views on what that might be able to do?
Sure. So as you correctly said, there is a bipartisan bill that has been introduced for about $2.5 billion for over five years. Most of that is slated to go to the DOE labs, such as Fermilab, Oak Ridge National Lab, and other DOE labs that we depend on for our funding. That bill seems to have bipartisan support. There's no indication it won't go through, but as of today, it hasn't been signed yet. We are optimistic that it will get signed here soon, and then the money gets appropriated to the right DOE labs. We are hoping that somewhere in the second quarter of this year, the money will start getting appropriated so we can start getting contracts from those DOE labs. In addition to DOE, DOD has several initiatives going on right now to fund quantum computing The biggest one being the DoD DARPA, Quantum Benchmark Initiative, QBI. We expect them to make some very important announcements in the next month or so, as to which is the group of companies that they have chosen going forward to build what they call utility scale quantum computing. Basically, that's like the world's best, biggest quantum computer that has to be built by 2033 time period. And they have several hundred million dollars, more than $300 million. in budget for that. In addition to that, there are several other line items in the DoD bill that will be able to fund quantum computing. So overall between DoE and DoD, we are expecting sizable increase in US government investment in quantum computing. We are just waiting for the bills to get signed and the money to get appropriated, but we are pretty optimistic as 2025 rolls along, sizable amount of investments will be available from DoE and DoD.
Thank you. And then finally for me before I hop back in the queue, regarding cash, congratulations on getting it to such a significant level. The first question related to that is, does it change at all how you look at near-term intensity for either R&D or sales and marketing? And then the longer-term question, since Jeff did indicate potential sufficiency for the next three years or so, How do you feel about its ability to get you to a level where the company is self-funding? Thank you.
So as I mentioned earlier, we have $217, as Jeff mentioned, million dollars at the end of last year. Plus we have this commitment from Quanta for $35 million plus the $250 million. So Rounding off the numbers, we have roughly $500 million available to us for the next five years, which is a sizable amount of money given our burn rate. And as Jeff said, at least for three years, we don't need to worry about cash and probably longer. Certainly, we hope that these government initiatives get funded by the US government, the UK government, where we have very good relationships, as well as some other friendly governments around the world that we have been talking to. Assuming those initiatives materialize, we certainly hope we don't need to raise cash. We'll certainly look at opportunities, but it's not that we need to raise cash if those initiatives materialize and we manage to get the contracts. But clearly we are in an R&D stage right now. We don't believe commercial sales to increase anytime immediately. I know there's a lot of discussion going around as to where exactly quantum computing is in terms of sales expectations. Our general view is we are still in R&D. We are still roughly about four to five years away before commercial sales matter, which is why we keep highlighting that it's R&D milestones that are far more important right now than this one-off government-type contracts. We certainly welcome those and we want those. And as government increases the budget substantially, we will certainly depend on those to help us get to positive cash flow in the next three, four years. But our focus right now continues to be squarely on R&D and making sure we are in the lead with superconducting quantum computing. Hopefully that answers your question.
Yeah, that's very helpful, Subodh. Thank you.
Thank you.
Thank you. Our next question comes from Quinn Bolton of Niedermann Company. Please go ahead, Quinn.
Hey, guys. This is Shadi on for Quinn. My first question is on the QPU sale to Montana State University. Can you guys discuss if the sale to MSU was a competitive process? And if so, what was some of the feedback from the university that led to them choosing Rigetti's technology?
Thanks, Sean. As you know, we have been building 9-qubit QPU and making it available to commercial customers, particularly academic researchers and government national lab type customers, not quite the classic data center type customers. The real objective of enabling that is to build a quantum ecosystem, and it's all for research applications. Clearly, the Montana State University is a research case where they are trying to understand the fundamentals of quantum computing. So they are doing basic experiments with those like pulse shape, pulse sizes, really understanding how to design algorithms and those kinds of things. So it gets to the fundamental understanding of quantum computing. You're really not going to use a nine qubit QPU to try to compete with a CPU or GPU. So it's really not meant for any practical application or demonstrating quantum advantage or anything. It's all for research purposes. But it's a nice convenient product to have in your lab where you can get hands-on experience It fits into many of the research customers in this area already have purchased a dilution refrigerator for various reasons. And a 9-qubit QPU is a relatively simple product that fits into your existing dilution refrigerator. So it's a relatively simple thing to ship, get it integrated into your system, and you can start working with it fairly quickly. So it's a fairly user-friendly way to get into fundamentals of quantum computing ecosystem. Hopefully that answers your question.
Yeah, it does. Thanks for that. And then I want to follow up on Craig's question, and sorry if I missed this, but is that DOE Quantum Leadership Act the same bill as the National Quantum Initiative Act, but reintroduced as a different name? Or are these two different bills going through Congress?
No, there's only one bill right now. It is the same one that Craig mentioned, introduced by Senator Daines and Durham. It's a $2.5 billion initiative over five years.
Got it. And I have one quick modeling question, but what can we expect the share count to be in Q1?
Yeah, I think in terms of share count, we ended the year at 200 and call it 84 million shares. So I would think, let's say 290 would be my estimate.
Awesome. Thanks for taking my questions, and congrats on the progress.
Thank you.
Thank you. Our next question comes from Krish Sankar of TD Cohen. Please go ahead, Krish.
Hi. Thanks for taking my questions. This is Stephen calling on behalf of Krish. Subodh, I guess the first one for you regarding technical milestones for the year, specifically on the scaling front for the 9-qubit modular tile architecture. So I know you guys have a target of reaching 36-qubit chip by middle of this year and then over 100 qubits by the end of the year. I was just kind of wondering, this level of scaling, should we assume that that rate of scaling can continue in future years as well, or are there certain physical limitations to the packaging or processing process in terms of like reticles or signal integrity so on and so forth and also related to it from a from an error rate standpoint you mentioned having or yeah 2x reduction in error rates by the end of this year just just wondering if all of the i guess the aggregate of different software algorithms and other improvements that you guys are working on both internally and with partners can that error rate be outperformed or better than you currently are estimating?
Sure. All very good questions, Stephen. So I'll try to answer them. So regarding, let's take an assessment of where we are. First, as a modality, I mean, you can clearly see what's going on in the quantum computing world with superconducting gate-based quantum computing. I mean, between our announcement, there's Google below chip announcement, plus recent announcements from companies like Amazon, Microsoft, even the Chinese Academy of Sciences, which is the government of China's sponsored organization. To us, it's becoming amply evident that superconducting gate-based quantum computing is the most likely winning modality here. I mean, the amount of investment going on from all the large companies and organizations. But look at the data. I mean, collectively, we are in the roughly 100 qubit range right now. Collectively, we are in the 99 to 99.5% median two qubit gate fidelity. We are in the tens of nanoseconds gate speed, which I'll point out is 10,000 times faster than some other modalities like trapped ion or pure atoms. And we are already deploying real time error correction with low latency. We did that, Google did that with their below chip announcement. So when we look at collectively where superconducting modality based quantum computing modality stands to us, it becomes very clear that this is the modality that's most likely going to win. Now, within that, we have our share of challenges. None of us, none of us are demonstrating quantum advantage yet. And I'll say that across the board for all modalities. I mean, you'll hear all kinds of hype going around and there's a lot of hype going around in quantum computing, but none of us have demonstrated quantum advantage. We are all getting to that point. And at least from the superconducting gate-based quantum computing side, we believe we need to get to several hundred qubits, maybe a thousand qubits. We need to get to like 99.7% median two qubit gate fidelity, maybe 99.8. less than 30 nanosecond gauge speed and real-time error correction to demonstrate quantum advantage. And that's where our roadmap and I believe IBM, Google's roadmap is comparable to ours. We are all looking at roughly about four or five years to demonstrate quantum advantage and commercial business to take off. Now, having said that, how do we get from the current about roughly 100 qubits we are all at to 1,000 qubit? And that's where I think different approaches start coming in. Our view is that Chiplets is a key tool that we are planning on using to scale up. And we did some early work with 40-qubit chips in the, like, about two, three years ago. Then more recently, last year, we did more with 9-qubit chips. We demonstrated a couple of different times that you can tile chips and still maintain all your quantum effects and see no deterioration in performance, which is a huge, important milestone to demonstrate. Having done that twice now, now this is the year we have decided to start deploying it in a more of a volume manner. So our first important milestone is demonstrating four times nine qubit. So that would be 36 qubit by the middle of this year and demonstrate 99.5% or better median two qubit gate fidelity. And assuming we are successful and we are fairly optimistic, we will be successful with that milestone then bump it up to more than 100 qubit by the end of this year. It certainly is a big milestone for Rigetti, but we believe it's a huge milestone for the whole industry because it's the first time anyone of us is going to show a real path to get to 1,000 qubit. Right now, all of us, even though we are at 100 qubit, We know that getting to several hundred qubits from where we are right now with a single monolithic chip is a challenge. We see that in our data. We believe IBM tried to go to 430 qubits a year ago, and they had some challenges, which is why you don't find it deployed right now. And certainly when we look at the other modalities, I don't even think they're anywhere close to what we are talking about, hundreds and hundreds of qubits and stuff like that. So our view is that chiplets is a critical technology. We have shown that it works in quantum computing. Obviously, the CMOS world has shown that chiplets are critical. I mean, if you look at any high-end applications with CMOS today, most of them do use chiplets. And there's a good reason for that, because it's a lot easier to control uniformity and performance over a smaller dimension chip, physical dimension chip, than a larger dimension chip. So there's no reason for us to reinvent the wheel. So we are using all the learnings from the semiconductor industry and CMOS industry in specific and deploying chiplets, we feel pretty good that we will demonstrate four by nine by the middle of this year, and then we'll bump it up to 100 qubit. And assuming we are successful with that, that we believe is a really good way to scale it up to several hundred and several thousands of qubits. Just to give you a feel, our nine qubit chip right now is six millimeter by six millimeter. And we certainly think we can shrink it down by a factor of two. That's fairly standard using conventional semiconductor technologies. We could even get more aggressive and reduce it further. But even with the current dimension, if you take a one meter by one meter panel, you can fit in more than half a million cubits. And certainly we believe dilution refrigeration technology will advance enough for us to maintain cold temperatures across a meter by meter square panel in about five years. And so we feel pretty good that we should be able to get several hundred thousand qubits, maybe even more than half a million qubits in about five years or so by using the chiplet approach. You correctly pointed out that the challenge does become packaging to some extent. So we will have to improve the way we are packaging the chips. When you're doing it only for four or 10 or even 100, you can do manual or semi-manual methods. Certainly when you're dealing with thousands and 10,000, you will need to automate that. But you look at the current state of advanced packaging from the semiconductor industry side, there are several advanced processes that have been developed and we will be able to leverage them and take advantage of them. So putting together several hundred thousand tiles, if you will, and certainly several 10,000 tiles is not that challenging given the state of art in semiconductor industry. So pretty exciting roadmap. We feel pretty good about it. That's our path to get to several hundred thousand qubit and like the utility scale quantum computer that DARPA is challenging all of us with. So hopefully I answered most of your questions. Did I miss any?
You know, it was very helpful, very instructive explanation, Subodh. Thank you so much for that. And just one quick follow up more for Jeff on the P&L and cash flow side of things. In terms of the Quanta computer collaboration and I guess the $100 million that you guys are committed to, I guess, investing from your side, how should we think about the expenditures of that $100 million over five years? Is that really like a cash investment in terms of equipment of some sort or is that just in the course of existing R&D and will it flow through OpEx or CapEx? Any details would be helpful.
Thanks. Yeah, I mean, it really is continuation of our ongoing R&D efforts. So it really will flow through OpEx, you know, in the context of our R&D team and with our capital plans and so on. So I don't think, you know, from our side, you'll really see anything change. too different other than the benefits that we're going to get from partnering with the QANTA and taking advantage of their expertise in areas outside of QPU.
Thanks, Jeff. And just as a follow-up, I guess the incremental $100 million spending, does that represent a step up to sort of annual spending, or are there offsets to account for that?
No. In our side, it really is a continuation of our ongoing R&D efforts. So there really isn't any specific step up per se at all on our side. On their side, they've committed to investing $250 million and furthering our roadmap and so on.
Perfect. Thank you so much.
Thanks, Stuart.
Thank you. Our next question. comes from Richard Shannon of Craig Harlem. Your line is open, Richard.
Great. Thanks, Subodh, Jeff, for letting me ask a couple questions. I guess my first one, Subodh, is following up on the prior discussion here on the roadmap for this year. I specifically wanted to ask about the approach to getting to the 100-plus qubits this year. Is this using kind of the tiling approach, kind of extending on the 4x9 one that you said you're trying to hit mid-year, or is this kind of based on more of the monolithic one, the ANCA-3. Can you kind of help us understand what the scaling approach here is?
Certainly, our roadmap is relying on styling to hit the 4x9 demonstration force by the middle of this year. Assuming we succeed with that, we will go, let's say, 12x9 to get more than 100 qubits, like 108 qubits or something like that. But we certainly have the option of using the monolithic approach, which is what others like IBM and Google are doing right now. And we have done all these years too. So we certainly have the option of bumping up the 84 qubit chip to a higher qubit count, but we believe diving is the right way to go in long term. So we are going to definitely try to get first the four by nine and assuming that succeeds 12 by nine or something like that.
Okay. So, So we're just kind of a one-track approach here in terms of using tiling going forward here. NACA has been more of kind of a demonstration or will be entirely focused on tiling going forward. Is that fair to think?
Yeah, it's fair to think. And the main reason for that is because we can see in our data that going the monolithic chip approach is going to be extremely difficult once you are in the several hundred qubits going to several thousand qubits. As I mentioned in my previous answer, IBM tried a 430 qubit chip and there's a reason why they haven't deployed it, we believe, because they also ran into the same challenges that we are seeing. And I mean, the whole CMOS industry has learned it for a decade now, right? It's very hard to build a large single monolithic chip. So we are finding the same issues. The root causes are exactly the same, uniformity and gears and those kinds of things. So given all the information we already have, we decided that tiling is the right way to leverage the chiplet approach that the CMOS industry has done such a great job and rely on that. Once we have proven that the quantum effects can be sustained across an interposer with chiplets, then the path becomes very clear for us. But it's important to keep demonstrating one at a time. So we will demonstrate four by nine first. We already did the work with two by nine last year and two by 40 a couple of years ago. Although at that time, the fidelities were not as good as what we are dealing with right now. Now we are dealing with 99.5% 9 qubit CHEPs. So we want to make sure that when we tile them, we don't see any deterioration in the fidelity performance. So it's important to demonstrate the 419 and 99.5 by the middle of this year. Assuming we succeed, certainly our roadmap will be very tiling-oriented.
Perfect. Thanks for that, Subodh. My second question is following up on one of the topics mentioned in the press release, as well as, again, the press release on this topic, I think, in December or January, regarding joint research with QFOX and QBLOX here. Maybe if you can just kind of talk us through how this accelerates your scaling and fidelity roadmap, it'd be great to hear about as well.
So, sure. So, it was an exciting announcement we did with QFOX and QBLOX, and it gets into signal in, signal out. Right now, we are using coax cables to send the signals to our chip and get the signals back from our chip. And most of the industry, we believe, does that right now. We are all looking at, but there's a reason why we want to move away from coax cables, and the main reason is cost and the physical dimensions of a coax cable. Right now, we are all in the 100-qubit range. Even 150 or 200-qubit range is not a big deal. We can live with coax cables. But once you go to 1,000 qubits and 10,000 qubits, the cost with coax cables will be prohibitively high. And the bigger issue is physically you don't have enough space in your dilution refrigerator unless you start building monster dilution refrigerators. It should be very expensive. So we need to find a more practical way to get the signal in signal out so we are looking at flex cables we have a lot of ip in that area and we'll certainly start deploying flex cables but these are not your standard of the shell flex cables because we are dealing with superconducting temperatures so there's a lot of materials and process innovation that's going on in flex cables and we are working on that uh some other companies are also working in that area like ibm and google i believe uh so we will first go from coax to flex but beyond that we need to think about Even at 100,000 cubits or above, even flex cables will have their share of challenges in fitting into a DR. And that's where optical signaling comes in. So a lot of work has been going on with converting our microwave signal, if you will, or RF signals into optical signals. The key part about our joint work that we published in Nature in December was using fiber optics. So instead of open-air optics, now we are dealing with fiber optic signal in, signal out. That makes it a lot more practical to build a system instead of just open-air optics. So it's a key technology demonstration that you can use fiber optics to get the signal into a superconducting chip and get the signal out. Now that that is available, we will certainly start investigating fiber optics and at what point should that come into our roadmap. As of today, we haven't decided when to
go from coax to flex to fiber optic but certainly by the time we reach several hundred thousand qubit we believe we will need to be with fiber optic cabling so hopefully that answers your question yeah that is helpful thanks for all that subode my last quick question here is just on the uh darpa benchmarking uh project or opportunity here i think you said you expect some decision here in the not too distant future maybe you can help us understand the process for uh for this award here or is this not even the end
Step here and then what opportunity you see from a revenue perspective over time if you're successful winning part of that Yeah, so DoD's DARPA Agency effectively is has invited Proposals from anyone and we certainly have submitted the goal and you can go to the website and see a lot of the publicly disseminated information the goal is to build a utility scale quantum computer and by 2033. And really what it means is you can practically do anything that your classical computing can do, but much, much faster and much, much cheaper than what your classical computing can do. It's really effectively a moonshot type effort. So this is like the US government's official, let's build the world's best, biggest quantum computer type project, man on the moon kind of a project. We certainly have applied. I'm sure others have too. They have said that they will make a decision here soon, hopefully just yet this month. Typically the way DARPA projects of this kinds of complexity work is they choose a handful number of companies and then they ask those companies to demonstrate the next set of milestones. And then they'll pick a couple of companies, maybe one company All our scope of the project is somewhere in the 300 plus million dollar range. I already told you the timeline is set up before 2033. We certainly, our goal is to be the last company out there and build the computer. So this will be the world's biggest, best quantum computer to be delivered to DARPA by 2033. But there are various hurdles to go through. We certainly believe our technology is in a very good shape to make a compelling case, particularly with the superconducting gate-based modality, as I said, In our view, and based on all the announcements you are hearing from other companies too, it becomes pretty amply evident to everyone, I believe, that the most likely modality to win is superconducting gate-based modality. Within that, with our open modular approach, the chiplet IP that we have, we believe we are fairly well positioned to try to win the DARPA project. So we certainly are excited to be participating in it. We look forward to their selections and we'll continue to work on our technology program. Having said that, we will continue to work on our technology milestones. I mean, DARPA is a huge opportunity and lots of money associated with it. But there's a bigger market out there. I mean, we have already said that the market is, we believe, like a couple billion dollars five years from now for all these national labs and universities. And we believe the market is going to be like 100 plus billion dollars in about 15 years from now. So certainly an exciting opportunity. So as important as the DARPA project is, and we certainly want to be winning that one because that clearly demonstrates technology leadership. The bigger potential, of course, is the commercial world and a hundred plus billion dollar opportunity that we'll continue to look at. So again, I hope I answered your question there.
That was very helpful. That's all from me. Thank you.
Thank you, Richard.
Thank you. Our next question comes from Brian Kingslinger of Alliance Global Partners. Your line is open, Brian.
Right. Thanks so much. First, you mentioned your plan to scale through tiling right now. Given the challenges that all of the superconducting OEMs have, is their approach scaling also through tiling? Are they trying to figure it out? Where are they with tiling compared to you?
so um certainly we are relying on tiling i believe ibm has made some statements suggesting that they are also considering tiling uh like exact dimensions will be different um but i believe they have discussed styling openly we are not quite sure of google um and what exactly their plan is at this point they did indicate in their paper when they published a below resource that they have some share of challenges to go up from where they are. They are using what we call perimeter wiring right now. So all of their circuit is basically designed in 2D. And to increase the qubit count, they have to keep increasing the perimeter, if you will, of the chip, which you can do up to a certain point, but not beyond that. So we believe Google will first move to 3D, which is what we and IBM are doing right now. And beyond that, they will probably consider tiling too, but we are not quite sure because they haven't disclosed all their details. But certainly we are on tiling. IBM is considering tiling. And frankly, we view the three of us as the leaders in this space. We are not quite sure of what the Chinese Academy of Sciences is considering. It's very hard to get information from there. And regarding the rest of the superconducting, I mean, even though there are large tech companies like Amazon and Microsoft and many other smaller companies, I mean, Amazon's most recent announcement, they were still talking eight qubits and same with Microsoft. So even though they are much larger companies than we are in terms of quantum computing, I believe they have some serious work ahead of themselves to get to the hundred qubit type level that we, IBM and Google are at right now. So we certainly view ourselves along with IBM and Google in the leadership position and how we go about solving this scaling of problem. I'm sure the other companies will be looking at us.
Great. My follow up that kind of leads into the next one. With your open source architecture, What does Amazon's announcement about faster and more cost-efficient error correction mean for Regetti versus the rest of the superconducting quantum OEMs? And then my second question is, as organizations like DARPA, DOD, and others evaluate you, do they communicate being more excited about an open-source flexible architecture, or does that not yet come up in the discussion?
It absolutely comes up in discussions when we talk to national labs, not just DARPA, but DOE and other government national labs too. The fact that our architecture is open and modular in nature is a significant plus in our favor because fundamentally it allows creative innovative solutions from other third parties to be incorporated relatively easily. I mean, IBM and Google obviously are doing a great job of building a quantum computer right now along with us, but this is a more mainframe-like approach right now. And again, we don't know what the Chinese Academy of Sciences is doing, but I suspect it's a mainstream-like approach to all. And that's great because you control all aspects of the full stack, but it's very hard to integrate a innovative, creative solution in that kind of an approach, with that kind of an approach. Whereas with our open modular, we can relatively easily do that. Specifically, I mean, River Lane in Cambridge, UK is a company that developed some really good error correction software, and we started integrating that in our systems last year. And that's how we showed real-time low latency error correction, which was a very important milestone in the industry. And this new announcement by Amazon, even though the chip is being built by Amazon itself, the 8-qubit chip, I mean, the real value in that paper is the error correction software. And certainly, we will be open to looking at integrating that error correction software if they choose to decouple it from their chip. And I suspect they may be incentivized to look at other 100-qubit type high-performing chips. So certainly, it opens up the avenue where we can integrate creative, innovative solutions from third parties quickly into our stack.
My last question is a really quick one. I just want to make sure I have the numbers right. If you get to 100 qubits at 2x better error rates, just help me do the math. What is 2x better? Is that 99.75? What is that actual fidelity rate that is 2x better?
So the reason we said 2x and not the exact number is because we have started using two numbers now in our fidelity. I saw.
Confusing.
Yeah, exactly. So we use 99.0% with what's called as an ISWAP or a CZ kind of gate. It gets really geeky at this level, different kind of gates. And then there's a unique gate that we call F-SIM gate where we get 99.5% today. That's what we have today. So when you have multiple gates, you are monitoring fidelity now. That's why we decided to start using the phrase 2x reduction in error rates across both of them. So the 99 will go to 99.5 and the 99.5 will go to 99.75. That's what I said.
Okay. Makes sense. Thank you so much.
Thank you, Brian. Thank you. Our next question comes from David Williams of The Benchmark Company. Please go ahead, David.
Hey, good afternoon. Thanks for taking my question. And I'm jumping on late here, so excuse me if this has already been asked. But, Subodh, you talked about in the past your architectural and your IP differentiation, and you talked about maybe Google and some of the others where you really can differentiate yourself from an IP perspective. Can you talk a little bit about what that means relative to maybe some of your competitors, what?
So within the superconducting gate-based quantum computing companies, certainly we monitor IBM, Google very closely. And we cannot really monitor the Chinese Academy of Sciences that closely. And as I mentioned in an earlier question, Even though Amazon, Microsoft play in this general space, they are at a fairly low qubit count right now, like eight qubits. And the rest of the startups or smaller companies in superconducting gate based are quite a distance behind us right now. So the two companies we basically look at very closely from what exactly they're doing are IBM and Google. And I'm sure they're monitoring us too. And IP becomes a very critical part. We have close to 230 patents right now. That is a core value of the company. How do we differentiate from IBM and Google? The main thing is our architecture. We just were talking about it. We have an open modular stack approach versus they have more like a mainframe approach. The other area where we are, and we believe the open modular approach is better in the long term because it allows us to integrate creative, innovative, third-party solutions much more easily. That's why we continue to invest in that approach. The other area where we have a clear differentiation from them right now is the whole chiplet area. Our plan, we have already demonstrated chiplets in quantum computing, once with 40 qubits, last year with 9 qubits. Our plan is to scale up multiple chiplets, effectively, or tiles to get the 36 qubits first and then more than 100 qubits by the end of this year, and then take that approach to continue to speeding up from there. We believe IBM will do something similar based on some of the statements they have made. Google hasn't made their plans very clear as to how they plan to scale up to thousands and several thousand qubits. So we will look at that. And then there are some other more details, gets into more details of how we design the chip and fabricate the chip and the rest of the stack and the differentiation that comes in. We have talked in the past about a proprietary annealing process called ABAA where we are doing DC pulses effectively through the entire area of the qubit. We believe our competition is doing laser annealing type approaches. We believe the ABAA annealing is a lot faster, easier to scale up compared to, and it's actually more uniform and controls the frequency targeting a lot more precisely than the annealing approaches. And then there are some, we touched on earlier on flex cables and that kind of stuff. So probably 10 other things that how we differentiate ourselves from them really gets into the details and the patterns and how we file the patterns and the scopes that they cover and so on. But hopefully that gave you a feel for how we are differentiating. The main differentiators are our open modular architecture, chiplet approach, and a few other things like annealing.
Very helpful. Thank you for that. And then maybe this is a follow-up. Do you think that your IP is compelling enough that over time others will need to develop around your IP? Or how do you think about maybe consolidation just given your rich patent portfolio and what that would mean for some of the others?
I mean, that's the whole goal of patents, right? I mean, when you come up with creative, innovative ideas, you file patents and you try to get as broad coverage as possible. So when others try to essentially copy that idea, they are forced to either find a completely different route, making it inefficient or have to seek a license from you. I think it's still too early, David, to know all the answers. I mean, clearly IBM and Google are gigantic companies and they have their own patent portfolios. Usually in cases like this, most of us in the R&D field who have dealt with these kinds of complex patent portfolios We find ways to collaborate and cross-license in critical areas. Sometimes it can be licensing terms, and we want to make sure the ecosystem is healthy and stays healthy, and everyone gets fairly compensated for the work they have done. So at this point where we are clearly in R&D, the goal is to get the technical milestones demonstrated, get patent portfolio established. Over time, we'll see how the licensing and cross-licensing scenarios evolve.
Perfect. Thank you. And then just one last question, if I may. Just on the customer demand for QPUs, could you talk a little bit about how that demand is and how you think it may be trend through this year?
Yeah, certainly we see excitement growing about quantum computing. People are beginning to, it's no longer a question of if, it's a question of when. Everyone seems to understand that when we talk to them. But clearly, it's still in R&D stages. Unlike some other companies, we believe we are still four to five years from commercial applications of quantum computing in a meaningful way. I know there are some companies who claim they have quantum advantage or near quantum advantage now, but those are for very select niche kind of applications. There's a lot of overhyping, underhyping going on in quantum computing right now. So our view is that we are still very much in R&D. We need to get to like close to several hundred qubits or a thousand qubits, 99.7 or 99.8%, less than 30 nanosecond gate speed and real-time error correction before we can start showing quantum computers to data center managers and demonstrating ROI. And we think that's at least four or five years from now. And that's more or less consistent from a timing standpoint with what you hear from IBM and Google too. So I don't think we are that different than IBM or Google right now in that sense from a timing standpoint. So our view is all the customer demand, if you will, as you called it, is mostly from academicians and research people, mostly to understand the fundamentals of quantum computing. And that's why the numbers are going to be relatively small. Yes, sales matter and we want to continue to grow, but our focus is clearly on R&D milestones right now and making sure the technology is perfected before we worry too much about customer demand and uptick in sales.
Great. Thanks so much.
Thank you, David. Thank you. Our next question comes from Craig Ellis of B. Riley. Please go ahead, Craig.
Yeah, thanks for taking the follow-up question. It's really just a clarification on some of the things that were part of fourth quarter's announcements. And I suspect it's mostly for Jeff. regarding the two QPUs that were sold, one to the U.S. academic institution and the other to the U.K., did those both fully reprack in the quarter? And then, Jeff, I think you said there was still part of a U.K. system sale that existed in fourth quarter sales. Where are we in fully repracking that system? Does it trail into 2025, and if so, to what extent? Thank you. Sure.
The two Noveras did fully rev rec in Q4. And then regarding the kind of the larger ongoing sale to NQCC, that revenue has been taken over time. And, you know, it will be largely complete in the first quarter, maybe a little bit, you know, moving into the second quarter. But don't really expect anything from that in the latter part of the year.
Got it. Thanks, Jeff.
Thank you. I would now like to turn the conference back to Subodh for closing remarks. Sir?
Thank you for your interesting questions. We look forward to updating you with our progress at the end of Q1. Thanks again.
This concludes today's conference call. Thank you for participating. You may now disconnect.