Quantum is Finally Here! What's Changed?

Monisha Saldanha (00:04) Wait, I'm gonna do that again. I think I was frozen. Was I frozen? Anyway, I'm gonna start again. I'll edit out the beginning. Justin Ging (00:10) I think it stutters because the bandwidth starts to, your CPU starts to Monisha Saldanha (00:15) Yeah. That could be it. Welcome to Colorado Tech People, the podcast where we talk about the people building great products in the state of Colorado. Today I have with me Justin Ging, the Chief Product Officer for Atom Computing. And today we're going to talk about quantum. We're going to react and investigate the question about why quantum now. which is a really hot topic across the whole country, but especially for Colorado. I'm going to say welcome, Justin. Thank you so much for joining us today. Justin Ging (00:54) It's pleasure to be with you. I'm excited to talk about Cornyn. Monisha Saldanha (00:56) Okay, perfect. I wasn't really happy with that intro. I'm just going to do it one more time. Sorry about that. Cause I think it could have been a bit smoother. So I'm going to start again. Welcome to Colorado tech people, the podcast where we investigate the great products being built in Colorado. Today I have with me, Justin Ging, the chief product officer for Atom Computing. And we're going to deep dive into the question, why quantum now? I am your host, Monisha Saldana. I have 15 years of experience in product management and I'm delighted to investigate with you today. Justin, welcome. Justin Ging (01:35) Thank you, it's a pleasure to be with you today. Monisha Saldanha (01:38) So let's kick it off by investigating the problem that atom computing was created to solve and what is the solution. Can you tell us a little bit about that? Justin Ging (01:49) Yeah, so Atom Computing is a quantum computing hardware company. We build quantum computers and every quantum computer company is really racing for the same thing, which is value added types of economically valuable computation that just can't be done other ways. So quantum computing particularly tackles certain kinds of problems that with classical approaches, the scale reaches a point where you just, you could spend billions of years computing and you never come in to a solution Quantum computing's different approach allows some of these things to be tackled. So things like chemical and molecular simulations where there's corollaries between the... energy bonds and the angles and things like this that get very complicated the more atoms and molecules you've joined together it's perfectly suited for what quantum computers can do and so everyone's racing for that ability but a journey to go on to be able to reach that. So Atom Computing is one of those companies and our particular way of making the quantum computer, the modality, is called neutral atoms. There are other approaches that are kind of popular. Some of the big names in the industry are taking those. Our approach is different in that the goal is to be able to scale up faster than these other approaches. So it really is about getting to the finish line sooner and enabling all these rich computations. Monisha Saldanha (03:12) fantastic, and how many people work at Atom Computing? Justin Ging (03:17) We continue to hire aggressively. We're up to about 120 people. We'll probably be about 150 people by the end of the year. We have more than half of our company is located in Boulder, Colorado. A large contingent of PhD researchers who are atomic, molecular, optical physicists who are driving the technology, but increasingly most of the people we hire are actually engineers of various disciplines, mechanical, electrical, control systems, optics engineers, as well as software folks. So really rounding out our team as we move forward. Monisha Saldanha (03:53) Yeah, I'm really interesting that you're based in Boulder. Colorado is known as being a hub for quantum. Can you talk to us a bit about what it's like to build a quantum computing company in Colorado and what that tech ecosystem is like? Justin Ging (04:09) So we as a company actually got our start out in Berkeley, California. And as we expanded, we were making choices. Where's the best place to continue to expand? Colorado and Boulder in particular are, it's a great place, partially because of the workforce. The type of approach that we use is very atom focused, controlling and manipulating atoms. And there's a long rich history within this community in Boulder. NIST and CU Boulder and the JILA, the organization that shares those two resources. Many Nobel Prize winners around these various topics. And there's great PhD programs that produce the kinds of employees that we need. And on top of it, Boulder and Colorado are just beautiful. A great place to have work-life balance. So it's quite easy actually to recruit people who aren't in this area of the country to get them to join. So it is about the talent and the lifestyle. Monisha Saldanha (05:16) Are there also PhD students from the School of Mines, the Colorado School of Mines that you would hire? Justin Ging (05:24) We have PhDs from all over the country and people who are in the field. So there are certain folks from Mines part of our company. Monisha Saldanha (05:39) Fantastic. what about networking? Is there a network within the companies that work in quantum? Like, do you meet regularly with other companies working in quantum? Justin Ging (05:51) At maybe different levels of meeting up. So within Colorado, there's an organization called Elevate Quantum, which has a convening element to it. So it does get everyone together to kind of think about topics as diverse as like workforce, but also kind of like what's going on in the industry. There are national groups, Quantum Economic Development Consortium that is now actually inclusive of international groups as well that kind of gets everyone together but the the other thing is there are a lot of quantum conferences around the globe and to some degree, it's a relatively small community. So you end up seeing the same crew of folks at a lot of these events. Increasingly, there are events that go outside of the quantum domain. Recently there was an oil and gas conference that CalCirawick, that S &P Global put on in Texas, and several quantum companies participated in that to kind of introduce what quantum can do for a particular industry. And that's just one example of that industry. With quantum benefits into the pharmaceuticals, and finance and other types of industries. there's increasing expansion into meeting up with those type of networks. Monisha Saldanha (07:01) And why quantum now? Quantum computing has been five to 10 years away for decades. So what's changed right now in the field of quantum computing? Justin Ging (07:12) Well, quantum computing is a journey. And I think the ideas started in the 80s and really formulated once there was this very real, tangible path towards cracking encryption that got people very interested in how this might work. So there was investment over the years. I think there were some big milestones, inflection points. One was IBM taking it in a lead and getting some of the first systems out around 2015, 2016, making it available to researchers, a five qubit system that they put out. And it was very beneficial to have everyone realize, oh, this is real. Quantum computers aren't theoretical. They are real devices that you can actually work on. Five qubits, five physical qubits, there's not too much you can do, but people got very creative and it inspired many others to follow. Our particular modality came about because a lot of things came together in the industry to make it possible. One of those key things for us is lasers. So in order to control and manipulate atoms, you use various colors of laser depending on the element that you're using, the atom that you're using. And... Prior to 2016 timeframe, 2017, if you were gonna tackle this, you'd have to build each of those lasers yourself. So you'd need to be as much a laser company as a quantum company. But because of advancements in those laser technologies and the commercialization of those, where you can purchase a laser that you turn on and it works, then you can move forward to using that as kind of just one component in the system. And that was an inflection point for neutral atoms particularly. In very recent times, I think the energy around quantum is picking up because, like you were saying, five to 10 years, when people say 10 years out, that's usually like beyond a reasonable window where you're like, sure, anything can happen in 10 years. We're at the point where we're more like five years out. So there's very tangible path at this point to get from where the industry is now and particularly our company. The systems that we will put out and reaching that threshold to utility scale. We're no longer at that point, Currently we're building research systems, systems that people can use to make better quantum computers essentially, test out algorithms at a smaller scale and start showing the technology and moving towards it. But at that point where we're making utility scaling, it actually does an identifiable economically valuable. things change quite a bit. And that inflection point is only five years away or so, where the customers aren't any longer just researchers, but actually enterprises. So when enterprise says, right now, hey, why don't you invest in a quantum computer? Well, we're talking many tens of millions of dollars. And that's a big chunk of a research budget to say, well, what's my ROI? You'll get research, you'll learn some things. It's true, you will, but it's a hard type of number to propose to an enterprise. But if you fast forward to, hey, this quantum computer is going to compute these particular things that you care about, reduce your cycle time on development through simulation or costs in some way. Now it's a very particular ROI and you're paying for computation that helps you do things and you can line up. Okay, I know it's worth paying this much for the computation because it saves me so much on the other side or it helps me make so much more money and that is the point where people get very excited and that's where all of the work in quantum starts to pay off. because there are so many industries that will be affected by quantum. Monisha Saldanha (10:55) And do you think we are about five years away from being able to quantify that ROI? Justin Ging (11:02) I think that's a rough time frame, give or take a year to something like that. We believe we will start to dip into that threshold with our next generation system, which will be about two years from now, into the range where many people have shown theoretically the level of computation that you need and will be hitting those thresholds. So some of the first things, there might be one or two very scientific niche type of applications such as something around magnetic materials or something like that that maybe isn't as widespread and helpful to everyone, but it'll start to cross into that threshold. And then as the quantum compute resources, the performance increases, that'll open up a lot more. that have been thought about. The other wild card in all of this is the algorithm side. So when people started thinking about, what size quantum computer do I need to do some of these problems, they came up with, okay, it's gotta be millions and billions of these qubits and it's gonna have to be quite large. And over the years, we see a very aggressive shrinking of those requirements to news that even just came out this past week of some of the things that are being proposed to, hey, we just figured out, here's a way we could do it with even fewer resources. So at the same time that hardware keeps improving as fast as it can, the software and algorithm side keeps shrinking the, what do we think we need, what size machine do we need to be able to do these things? And so the combination of those two things is really what's pushed it to relatively near term when we talk about years. Monisha Saldanha (12:41) And where is the market for quantum? Is it mostly US-based or is it more international? Justin Ging (12:47) It's a global market right now. There are various, I wouldn't say it's everywhere in the globe, but there are various hotspots around the world where there are great quantum ecosystems. The market for quantum computers is very much selling on premises systems and the customers are usually public private partnerships who are very intent on building out their quantum ecosystem. So they recognize quantum's an important technology, similar to AI. They don't want to get left behind. It's crucial to the future of their economy. And not having quantum is a big disadvantage. And so how best to build their own ecosystem of suppliers and researchers and talent and workforce. Generating those jobs in the future, but to have a quantum system at the foundation of that. So a system that can benefit the local universities, the professors and others, researchers can use it to have dedicated access to quantum computers, but even more importantly, novices who haven't touched or thought about quantum computers, maybe aren't even in the field, or maybe biologists or something like that that say, how can I touch a quantum computer? And having that access to be able to introduce people to it and start training them becomes a crucial part. And so all of that together makes the quantum computer as an ecosystem builder kind of the product at the moment. And so around the world, various regions, states, countries, provinces, et cetera, who want to have quantum. in their area. Those are the customers at the moment. Things will change. Yeah, please go ahead. Monisha Saldanha (14:19) And off the top of your... sorry, was going to say, off the top of your head, could you name the top five centers for quantum globally? Justin Ging (14:29) The US is obviously very strong, in addition to that, the EU and UK, in Asia Pacific there is Japan and Korea and Singapore. Australia has a strong ecosystem and the Middle East is starting to show up as increased interest there. Monisha Saldanha (14:47) Fantastic. And what's the biggest bottleneck to scaling quantum systems today? Justin Ging (14:55) For us it's much more on the engineering side. So we have a technology that scales well, but that scaling is an engineering challenge. So the way we operate our quantum computers, we hold individual atoms in a vacuum chamber and we use laser tweezers, which is a very finely focused laser beam, to hold those atoms in free space. And so once you have it held, you can actually move your qubit around, almost like a tractor beam in Star Trek, you can hold those atoms. And we create many of those spots of life, these tweezers that are holding atoms. Once you're holding that atom, you can use other lasers to send pulses to it to control the quantum information. And you can cause them to interact and have the entanglement and perform these quantum operations. So for us to scale, we add additional atoms, which are the qubits, into this vacuum chamber, and we control them. So that is where the engineering starts to come in, is how do you add more and more of these atoms and control all of them and do it in a way that you're doing it very quickly? So there's the number and the quality of these qubits, but also how fast you are doing the computation, because... even when you're talking things that happen at microsecond scale or millisecond scale, doesn't sound like much, but then you have many, many, of these operations. And so it starts to add up. And so the engineering challenge is how do we control that many spots of light and how do we move them around efficiently. and how do we read out the results of it quickly, things like that. Monisha Saldanha (16:33) And it sounds very highly skilled, the work that needs to be done. You mentioned hiring PhDs. What types of skills are you looking for? What types of backgrounds? Justin Ging (16:45) Well, usually the PhDs are AMO physicists, the Atomic Molecular Optical. They have very often worked in the exact type of work in their PhD program of how do I control and manipulate atoms to do various things. And so that's a large part of it, but a very particular type of physicist. And there's only so many around the world and getting trained. So it is a pretty scarce resource. Monisha Saldanha (17:10) And CU Boulder has a program that specializes in this. Justin Ging (17:15) They have a team, there are others around the world, some in the US and some globally who study these. And usually there are particular networks of people around those experts. Monisha Saldanha (17:33) Great. And could you tell us a little bit about the trends around what people want to use quantum computers to do? Justin Ging (17:43) Yeah, so I would say in the past five to 10 years when quantum computers were relatively low capability types of systems, people were making the most of it. So using those physical qubits and doing what are called MISC algorithms, noisy intermediate scale quantum. like The interactions are not necessarily the highest fidelity interactions, not quite reliable. But people got creative and started showing at least demonstration algorithms. Like here's how you could do some chemical modeling. Here's how you could do some optimization problems. And there was several years where that was the bulk of the activity was showing that quantum computers can do these things. Usually small scale problems where you would kind of say like, well, I could almost do that just with a paper and pencil. They're small problems. But the point was, if you write a quantum algorithm to do this thing on a small scale problem, it actually computes and does the right answer. You can start to build up that credibility that the quantum computer can does work in essence. But the reality is for a quantum computer to do these actual utility scale problems, they have to be very high quality interactions. And so the way that you get that is, there's a limitation on the physical qubit fidelity. And so how do you get across the barrier to making it way, way, way better? You start grouping physical qubits together to each essentially act as one qubit. So it's a logical qubit it's called. So it's like a virtual qubit that has made up of actual physical qubits so that collectively they act like high quality. Those are the ones that you can do these great algorithms with. So the question is, how do you take those physical qubits, make logical qubits and make a lot of them? Well, it's algorithmic based. What is the algorithm for working with these physical qubits to track the errors and correct for them such that they work this way? And so that is where all the research is channeled now. I say all, but a majority of the research around the world, everyone's very interested in how can they make a quantum error correction algorithm that makes, efficiently makes logical qubits and does it in the best way possible to increase the highest quality at the lowest physical qubit resources. And so people are very interested in using the quantum computers now. And that's not as appealing to the end users of like, well, how's that gonna help me in my pharmaceutical or finance? It's not solving those problems. It's actually solving, how do we get the hardware so that it can solve your problem? So there's a little bit of a setting aside those end goals for a moment while. So that's where the heart of research on quantum is right now. Monisha Saldanha (20:27) And for listeners who may not be familiar with quantum computing, what kinds of problems could quantum computers solve that classical computers struggle with today? Justin Ging (20:38) Yeah, so I mentioned some of those. I think the one that we're excited about is chemical molecular modeling. And that is beneficial to materials, to many different things in sustainability aspects. Can you design a more efficient solar panel, for example, or can you improve a process with catalysts? Could you improve, let's say, oil and gas drilling or something like that? All of those types of chemical molecular types of things. And part of the reason that's exciting is because we expect that to be one of the first applications, something that requires smaller resources from the quantum computer. As the quantum computers get bigger, more of these types of things open up of like, what could you do for finance or some of these other things, but that chemistry stuff, the materials seems to be first. And part of the way we see this interaction going is actually in conjunction with AI. So there's a interesting tie in with AI. One is perhaps quantum computers can actually be a co-processor in the training process, which is exciting from the energy consumption point of view. If you could pass off at least some of the computational jobs of training which are using all this energy of data centers. If you could do it more efficiently with quantum that'd be a big win right away. And there is a lot of research into how can quantum computers not tackle the whole training but particular aspects of it. But another interesting aspect is how do you make the AI better? And the concept is essentially using quantum computers to produce data that doesn't exist in the world. So AI models are taking the world's data, compiling that together to be very smart but it's only as good as what you're feeding it. So if you can use a quantum computer to model interactions go at a much more higher resolution and use that as a new data set for AI to train on, the AI will become better at actually predicting what they can do. So you could use the AI to do your simulation based on the data that quantum computers fed it. So really interesting interaction that actually there's been demonstrations at a small scale that that can work. Monisha Saldanha (22:44) Wow, fascinating. And atom computing is building quantum computers based on neutral atom arrays. How does that architecture differ from other quantum approaches? Justin Ging (22:58) Yeah, Several different modalities are out there. Essentially, you need something that has a two-state interaction that is quantum. And so we particularly use the nuclear spin state of spin up or spin down a nucleus. You can use electron states. Superconducting is another approach used by IBM and Google. A super chilled circuit and use the current, the direction of the current as your qubit. For us, the atomic approach provides a lot of benefits. The neutral nature means you can pack a lot of atoms close together, so it allows us to scale. The nuclear spin, because it's in nucleus, it actually has some inherent protection because the electrons kind of provide a shield around that. And so we have very long coherence time. So it holds that quantum information for a long time. It helps you in the computation. And the approach that we have where I was describing with the tweezers and you're moving them around, you can actually, instead of having a fixed topology where if one qubit wants to talk to a qubit that's far away, in superconducting it's printed in circuits. So one qubit has to talk to the next circuit and talk to the next qubit and so forth, whisper down the lane. And so you lose a little bit of fidelity each time. With our approach. can go grab whatever two qubits you want, put them together, have them get entangled, and go put them back. And that allows all to So any qubit can talk directly to any other qubit, which turns out to be a huge win when you're doing error correction schemes so that you can be more efficient with those schemes. So all of these pieces together give you the ability to create these logical qubits and create a lot of them in the near term. so that we can get to the score of utility scale. Monisha Saldanha (24:53) Fascinating. Quantum computing is still an emerging field. How do you make product decisions when the underlying science is evolving so rapidly? Justin Ging (25:05) Yeah, so When you're thinking about a product from the technology, so we're taking research that is very cutting edge and productizing it. There's a balance and there's a lot of negotiation between what do you need from a commercial side? What is the customer expecting? What is that overall experience? And what can we actually build at this time? And so we have to be very realistic about what's possible. I mean, it would be a dream if you could just say, okay, keep working it until I have the ultimate system. But we have to kind of set it out in a staged approach because at least we believe that having systems that are not necessarily full utility scale is actually helping other types of research happen. So if there isn't a system in existence for software and algorithm development, that won't make progress. And so having interim milestones of ever-increasing system sizes, we think is beneficial to pull everyone along to the readiness. If we magically produce that utility-scale hardware today, I think people will be scratching their heads. wait a minute, how do I use this and maybe spending a long time just thinking that through of like, okay, you've given me everything I wanted, but now I don't, I haven't done the prep work of how am gonna take advantage of it. So we believe in putting out almost a trained schedule of products and that trained schedule is one of those kind of product development type of basics. There's a famous business school class I recall from about. that's done for pacemakers. it actually provides the path to is this technology ready for this generation? Are we going to drop it into this model? all the pieces fit into a roadmap that's a technical roadmap of we know what we need to hit utility scale. So how are we going to demonstrate and de-risk each of those technologies as we get to that? Monisha Saldanha (27:06) And where is Adam computing on the commercialization journey? Justin Ging (27:12) Yeah, so we sold our first commercial system last year to the country of Denmark and Novo Nordisk Foundation. We're actively delivering on that and seeking out other customers. We build and deliver quantum systems. Then where we deliver them, we also set up a facility. So first of all, just to support that system, because it's going to be hands on for keeping it running during the whole time. These are not. you know, black boxes that just run forever on their own. So we have support people, but then also it's our goal to participate in each of these ecosystems. So that is our customer goal to build out their ecosystem and it's our goal as well. So if we, if we had a magical buyer who said, I want to put it in a glass case and just look at it because it's beautiful. That's nice. We would like to make a sale, but it's not actually serving our long-term goals of cultivating the right research to move this forward. It's not a finally we deliver this product and we're done. This is just the start of quantum computing as an industry. that will come from it. So the more that we can see these quantum computers used in those ecosystems doing interesting research, helping us move our hardware forward but also moving these applications forward is beneficial. And so wherever we deploy a system we also have folks that will participate in that ecosystem actively make sure there is strong demand for that system and that people are taking advantage of our particular hardware. and the benefits that it provides. Monisha Saldanha (28:45) And how have you found fundraising for the company? Has it been difficult or has it been easy? Justin Ging (28:51) I think the typical thing is we're always raising as a company. Quantum computing is an expensive type of endeavor, but there's enthusiasm in the community for it, especially now as we're getting so much closer. think people can actually see the end goal, see the light at the end of the tunnel. There's definitely interest in that. And I think with companies moving to public markets, we've seen even retail investors get more educated and more involved and excited about quantum. So I think there's many paths to funding at this point. and that's great for the industry. Monisha Saldanha (29:32) And if everything goes according to plan, what is your vision for where atom computing will be, five to ten years from now? Justin Ging (29:41) Well, hopefully in the five year frame, we're talking about utility scale systems. There will come a time when we're no longer selling on-premise systems that we're delivering to customers in various places around the globe, but actually moving back to a data center model. We believe that once the compute itself is what's valuable, then a cloud model actually serves everyone quite well. So once you know exactly what your compute's going to do, you're buying your time and you don't really care as much about the physical system and physically touching it. You can have it happen in the background and even longer term when people say, how's it gonna affect my life? You probably won't know that a quantum computer is doing the computation today. In fact, I would guess that most people don't know when they ask the cloud for something, did a GPU help that? Did a particular ASIC, did my video get encoded by a CPU or transcoding chip in the cloud? Nobody knows. Most people probably don't care. They're like, did it do what I wanted to do? And I think the same thing will happen with quantum, where the quantum portion of a computation will be sorted out within the cloud. It'll have And so People's lives will definitely be affected by the things that people do with quantum to invent new materials and make products better to do particular their applications. But they may not know it in the long Monisha Saldanha (31:10) Do you think quantum computers will ever be like personal computers or do think they'll always be something that's more for commercial purposes? Justin Ging (31:21) I think quantum computers are a bit more of a specialty computation tool. There are certain kinds of problems that are better tackled by quantum. And it turns out one plus one is not one of the things that it does well. You actually can do it with a quantum computer. You can make an adder. But there are many more efficient ways to do that type of a problem. But for the kinds of highly, high number of variables, high computation space problems, quantum computers are the best for it. So does everybody need that? Not necessarily. So not for their daily life. So I think that we'll just continue to see compute kind of be this heterogeneous mix of using the best type of processor for what you're going to do. I mean, think, so I came from, before I was in quantum, I've been in the quantum space for about eight years. Before that, I was in the semiconductor space, particularly all the chips and processors and sensors and cameras and things that are in cell phones. mobile market. In cell phones, the processors are systems on a chip. There's actually more than one CPU in these architectures for those chips. Usually have big CPU cores and little CPU cores. And the reason is sometimes I need a lot of horsepower, but it uses a lot of power and my battery is limited in what I carry around in my cell phone. And sometimes I just need a little bit of processing power and I can do that at very low power. And so within that processor, you're actually diverting the job to here's where I need a big one, here's where I need an energy efficient one, here's where I need a particular function done and there's a little special core for doing that. So I think the same kind of thing happens at a macroscopic scale with compute. Hey, in this case, I just regular cloud CPU, in this case, I need a QP, a quantum processing unit to actually tackle that problem. In some cases, I need a supercomputer that is doing weather analysis for weeks on end. So I think there will continue to be this bifurcation of specialization of processors so that you're always using the most efficient way of doing the computation you want. Monisha Saldanha (33:30) And so with that in mind, do you think that there will be different price points? Like as you mentioned earlier that the current computers are tens of millions. Do you think that there will be quantum computers available for less to do more simple computations? Justin Ging (33:41) Yeah. Yes, I think that as soon as you start to produce them in much higher numbers, there's many ways that costs can be addressed. The goal right now for quantum companies, and computing included, is make it work first. Worry about the costs later. We have many ideas about how to reduce those costs. Once you're producing, you know, we're doing at most a handful per year. But if you increase that... let's say 2x, 3x, 4x, that already would start to show some benefits that we can do from the supply chain point of view. And if you say, I'm to do 10 or 100x the number of quantum computers, there are some really big things that could be done and some technologies that could help that move along. So we strongly believe that the cost will come down for quantum compute as they proliferate. Part of it is to get to the goal of doing something economically valuable. and then cost engineer it down from there. Monisha Saldanha (34:45) Really interesting and exciting future ahead in quantum. What is one book every builder should read and why? Justin Ging (34:56) I am a big fan of Chip and Dan Heath's books. More recently, Power of Moments is one that from the business side and from marketing and telling the story of quantum, the book is essentially about how do you make it special when people are presenting something new? For us, we're certainly in the new space. And how do we make that experience a positive one? Monisha Saldanha (35:25) Fantastic. Well, Justin, thank you so much for joining us. This was a really interesting conversation. I think some really great nuggets in this for people that are very familiar with quantum, but also for people that don't know very much about quantum. So thank you for sharing your expertise. Justin Ging (35:40) Yeah, pleasure to be part of the podcast. Monisha Saldanha (35:42) And I'd like to thank our listeners. If you found this podcast helpful, please share it and subscribe so you can hear the next in our series of people who are building great products in Colorado. Thank you and goodbye.