Podcast with Rob Hays, Atom Computing

Our guest in this episode is Rob Hays, CEO of Atom Computing, a company building quantum computers with nuclear-spin qubits. Rob and I discuss the uniqueness of their atom-based approach, their go-to-market strategy, his view on diversity in the quantum workforce and much more.

Listen to additional podcasts here

THE FULL TRANSCRIPT IS BELOW

Yuval Boger (Classiq): Hello, Rob and thanks for joining me today.

Rob Hays (Atom Computing): Hi, Yuval. Nice to meet you and glad to be here.

Yuval: So who are you and what do you do?

Rob: I'm Rob Hayes, CEO of Atom Computing. Atom Computing is a quantum computing hardware company on a mission to create the world's most scalable and reliable system so that we can help our customer customers reach their next big discovery.

Yuval: Well, that was short and sweet. So what does that mean in a little bit more detail? What is Atom based computing and how is it different than other modalities of quantum computing?

Rob: Good question. Atom Computing is a quantum computing hardware company, which basically means we're building hardware platforms on which people can run their applications. And for us, the mission is pretty clear, which is really to build bigger machines, as measured by the number of qubits.We need to make sure there's error correction built in so people are getting the right answers.  There's a lot packed into that short sentence and it's a lot of hard work, and at the end of the day, we're using a technology that will allow us to scale up to very large numbers of qubits by using the world's first nuclear-spin qubit that's made out of neutral atoms. Neutral atoms are alkaline earth elements that come out of the second column of the periodic table, and they have a property that is very stable and do not interact with the other atoms, unless you want them to.

We're trapping these atoms in a vacuum chamber using lasers at a certain wavelength of light that allows them to get very cold and still in space so that we can control where they are.  Then we can read and write quantum information to them, using cameras and lasers. That's basically how we're doing it.In order to scale up the machine to more qubits, we simply create more spots of light –each of which is individually controllable, and we can run the gates and circuits using those types of controls. On top of that, we have a software stack, infrastructure software and APIs and ways to program the system, as well as layering in error-correction over time in order to build very useful quantum computers.

Yuval: So obviously building a quantum computer with the software stack to control it is a big undertaking. Where are you on that journey? How far are you from being commercial or maybe customers are using it today? 

Rob: We're a very young company. We were only founded in 2018 and got started building our first quantum computing system in 2019. In two years we were able to build what's really was tied for the largest quantum computer at 100 qubits. We did that with far less resources than anyone else. That's really just a testament to the technology that we've chosen, it’s  very efficient and quick to scale. So what we have is our first system, I'll call it a prototype system, called Phoenix. It's 100 qubits; it's up and running. We're using it right now, internally, to test tune it, test the software stack, get features complete ao customers can use it. We'll start to allow customers to use it in really weeks now, not months. That will start the learning curve of the software, the applications and the hardware evolving together, which is really important as we go. To be perfectly honest, a 100 qubit system, isn't all that useful. It's nice for research and running toy algorithms, but to run real applications with commercial value, we need much larger systems. And that's what we're working on in the background.

Yuval: How large is large? What's beyond 100?

Rob: I mean today semiconductors are billions of transistors, right? So I'm sure someday we'll have billions of qubits as well, and the good news is you don't need billions of qubits to get value the way quantum computers work. By being able to compute lots and lots of information simultaneously inside of the qubits, it’s going to allow us to do useful work probably with thousands or at least tens of thousands of qubits. Ultimately we want to get to millions and beyond.I think everyone pretty much has that same kind of idea in the industry of where we need to go.

Yuval: So you're using atoms as the base modality. I think there are three or four other companies in various parts of the world that, at least at a high level, are doing the same thing. How are you different? Why are you different?

Rob: I'll unpack that and answer it in two ways. One is why are neutral atoms, or what some people call cold atoms, why are they different or better than the other modalities? And then, how are we different or better than other companies working in our modality? First of all, neutral atoms or cold atoms, as they're sometimes called, basically are superior to what the older modalities or the modalities that have been around longer compared to superconductors and trapped ions for a few reasons. One is they're not manufactured. We're not manufacturing chips or devices in order to build these qubits. We're basically capturing a natural element in a vacuum chamber. So there's no imperfections because each atom is identical by nature.

They're very stable. We've demonstrated world record coherence times, over 40 seconds for every qubit. And we've got an arXiv paper out in the public domain, for folks to read on how we did that. These atoms are controlled wirelessly by laser. We can read and write the information without having to cable up each qubit. So that makes it unique from other modalities, where generally on their chips or devices, they're able to get maybe a dozen, or 20 or so qubits per device, but then they have to cable up all these different devices in order to get them to operate as one larger system. We don't have to do that. We just create more spots of light. Our atoms are placed a few microns apart, which is far enough apart that they don't interact with each other in their ground state, but close enough that they can have a lot of interaction in a quantum state.

That allows us to write two qubit-gates and we can do all that in a very small space. So at 100 qubits, the whole system is less than 40 microns on a side. To get to a million qubits, we would still be less than 500 microns on a side. That’s less than a cubic millimeter, far less than a cubic millimeter to get a million qubits. It's quite unique in that way, relative to the other modalities that have to be cabled up and put into dilution refrigerators and things like that. With respect to our position versus our competitors in a similar modality. On one level, we're all kind of in the same place, we all have something around 100 qubits. There's a couple of other companies that are out there, doing something similar, but in some ways we're going after, either different applications or a different go-to-market strategy. I know one of my competitors is more around building systems. I'm more focused on building a cloud service at this point in time. Another competitor is building a simulation machine, not a gate-based machine. Most people believe that gate based machines are what's really needed for the future. So there are differences in our approaches, even if we're using similar technologies.

Yuval: So let's talk about go to market, the customers that you mentioned. First, what is it that they want to do with 100 qubit system? And how do you make it available to them? Is it through your own cloud? Is it through a cloud from like Amazon or Microsoft or one of these guys? Is it in some other way?

Rob: Our 100 qubit based system is contemporary with the size of many of the systems that are out there. Some are dozens of qubits. There's a few that are 100 or so. Right now, most of the customers that would use something like that are classic early adopter customers. Customers that are wanting to build up their internal skill sets on how to program quantum systems and learn how the performance differs versus CPUs,GPUs or other accelerators. And they're willing to invest in people and time in order to go get those learnings. There's not a lot of economic value that gets created out of such small systems today. It's more like a preparation and a learning for the future systems in the next two or three years. The kinds of companies that we're seeing that are making those investments are large banks on Wall Street, for example, who are building up teams to figure out how to do portfolio optimizations or risk management type of problems that they can use quantum computing for.

We're seeing a lot of interest in transportation and logistics, where we've got a different mapping or routing problems that quantum computers can help with. Biotech, for chemical and molecule simulations. There's a lot of interest in the government space worldwide for lots of reasons.Government research, defense usages, as well are just building up sort of the economic and technological capabilities for quantum, because everybody knows it's a huge paradigm shift for the next 50 plus years in computing. Nobody wants to get behind. As far as go-to-market, today on Phoenix, we don't have it available on anyone's any third party cloud. We only will offer it through our own API initially. We'll give customers access through our own private cloud and let them test the code. We'll be launching a public cloud service in the future, but we're not ready to announce the date yet.

Yuval: When you look at a quantum computer, it's a very complex system starting from creating the individual qubits to controlling them and so on and so on. It's a multi-level stack. Do you see yourself as vertically integrated sort of like the car manufacturers of way back then, or do you think that your work stops at some level and then other companies or other organizations will take care of other points in the stack?

Rob: I think as a hardware platform provider, we're going to need to have some skills and capabilities and know-how from top to bottom in the stack, so we can understand how customers want to use the system and how it performs and help guide them. From a business model perspective, we have much more of a horizontal building block approach where we'd like to partner with software providers and cloud service providers, application developers and consultants and system integrators, all of the above in order to basically participate in the ecosystem. Our goal is to build the most scalable quantum computers we can with error-correction and then let our partners invite customers and help the customers program them to do whatever applications they want.

Yuval: You mentioned that the early customers are experimenting, essentially riding toy models, because perhaps with 50 or 100 qubits, they're not going to get anything that they couldn't get done on a classical computer. At what size of a quantum computer, or what other parameters, whether it's fidelity or connectivity or gate set or anything like that. Do you think that people will get true quantum advantage?

Rob: That's the billion dollar question you're asking right there. I don't know that I have all the answers on what is going to be the killer app on 1,000 or 10,000 qubit machines in the future. What I hear from customers is there's lots of ideas and what people think that they can use these systems of these scales for. One of the things that we're doing with partners and others is really trying to map out some of these use- cases of interest that we're hearing from customers onto how many qubits or circuits... How big does a circuit need to be? How deep does it need to be?How performant does it need to be, in order to provide value versus what they could do on a classical HPC cluster, for example. It looks like there are some early use-cases that will have economic value in, I'll say, 1,000 to 4,000 qubits.

But again, these will be early adopter use-cases. They could be big, like big financial services, big pharma, for example, machine learning in cloud services. But as we scale to 100,000 and 1,000,000 and beyond over the coming years and decade, I think a lot more applications are going to get written, a lot more use cases are going to fit in those class of systems or size of systems. We're going to do everything we can with our partners to help our customers discover the next killer app. That probably no one has even thought of yet, to be honest.

Yuval: I think that one of the issues that people talk about is the workforce issue. I saw a slide from McKinsey. I hope I remember it correctly, that in a couple of years, there will be 10,000 open jobs in the quantum computing industry and maybe four or 5,000 people that can do it. So there's going to be a huge shortage. What do you think about that? And what do you think should be done to close that gap if it exists?

Rob: Yeah, I think it is going to be a problem. Today the market and the industry is small enough that it's probably not an immediate problem that we don't have millions of people that have the right skillset, but it is scarce today. There's definitely competition for talent. For us, what are we looking for? Number one, it's technical talent, right? Engineers, physicists, software developers of lots of different disciplines, because we're building a complete system. So think of any engineering discipline, it's probably somewhere embedded in our system. We are attracting, we have over 20 PhDs, we're a small company, but we've got lots of PhDs. We're over 80% engineers, and like I said, multiple different disciplines. We need to, in our company and the industry at large, attract talented engineers who want to come into a nascent industry like quantum computing and bring their talents to help us make it real. Then there's also the business hires and the administrative and all that.

When you're building companies in an industry, you need all disciplines.. It's not just technical. Technical is number one, that's going to be the majority of the jobs, but we also need finance and marketing and business development, sales, all the other disciplines. At Atom, I feel very blessed that we've been able to attract really good talent. We've got an amazing technical team that Ben Bloom and Jonathan King, who founded the company, started to build two, three years ago. We continue to get very high caliber resumes flying in, which is really nice. We've started to build the business team to go to market. We just hired Denise Ruffner, who has experience at IonQ and IBM quantum. She's very well connected in the industry and very talented and Justin Ging, who we hired out of Honeywell Quantum Systems recently, and he's going to be our head of product. He's got a lot of experience at both semiconductor companies, as well as quantum systems companies and defining products that customers love.

We're really excited to have that team on board and start building out the business side as well. I think that the next big challenge for us in addition to just skillset, is diversity. We're not where we want to be on diversity. When you look at the applications that quantum computing is going to have, it's going to touch many different people around the world, many different types of industries and eventually consumer life.I don't know how it will do that exactly, but we know it will. We want those products and services to represent the people. We need more diversity of thought,experience,background and culture in the industry in order to discover those applications and make sure that they have great user experiences that people around the world love.

And so that's something that, we at Atom Computing, are really trying to dedicate ourselves to, is attracting a diverse workforce. That's going to start with education and sponsorship of interns and things before people actually walk in the door and send us a resume.We're going to need to make sure we attract them, a diverse set of people into the industry at large.

Yuval: As we get close to the end of our conversation today, I'm wondering as CEO, what keeps you up at the night? Is it overhyping quantum computers and there'll be a quantum winter, or is it funding or is it people, what's the thing you are most worried about?

Rob: That's a good question. A lot of people talk about this quantum winter. I don't buy it. We're at the ground level of building quantum systems, we're basically capturing atoms and we're turning them into quantum computers. Doesn't get any more elemental than that. I'm very confident in our roadmap and our team in order to go do that. As long as we continue to scale the systems at the pace that we know we can go do, I don't think they'll be a quantum winter, so that doesn't worry me. I don't worry too much about funding. =If we execute to the roadmap, I know the funding will come. What I really worry about is building. Building a team, competing for the scarce talent that's out there, attracting the diversity like I just talked about and executing to our technology roadmap at a pace that really sets the pace for the industry, to be frank. And that's what I get up to do every morning. And that's what I guess maybe keeps me up at night –how can we go faster?

Yuval: Excellent. So Rob, how can people get in touch with you to learn more about your work?

Rob: We’d love to hear from potential customers, partners, employees, anyone that's interested in following our journey. You can follow us at atom-computing.com.

Yuval: Very good. Well, thanks so much for joining me today.

Rob: Thank you, Yuval. It was a lot of fun.