Deep Tech Dive #10 | John Levy CEO of Seeqc

Andrew Kirima
From Deep Tech VC to Quantum CEO

For this Deep Tech Dive, I teamed up with Alex Challans, CEO of The Quantum Daily and The Deeptech Insider, to have a conversation with John Levy. John is a former deep tech VC turned quantum computing CEO for Seeqc. At Seeqc, which stands for scalable energy-efficient quantum computing, they are developing the 1st digital quantum computing platform for global businesses. They combine classical and quantum technologies to address the barriers to quantum computing systems.

Unlike most interviews, I and Alex have 2 objectives here: to learn about Seeqc, John Levy, and to understand their motives for their investigation on enterprise investments for deep tech solutions.

Key Takeaways:

  1. Nature isn’t classical, it’s quantum mechanical. Thus, quantum computing could be our only shot as a species to really model and understand the natural world.
  2. Quantum computing isn’t one market, it’s an entire ecosystem. To uplift this community, companies shouldn’t try and do it all — they should find a specialty, and contribute to bringing this technology to life.
  3. From the report, when asked, “What deep tech applications are company’s pursuing?” Climate Change Mitigation was ranked at #6 with Drones/Advanced Robotics. Which is a tad bit worrisome.
  4. From the report, when asked, “When does your company expect to see business results/ROI from your project?” By far the largest outcome was 3–5 years at 46%. 

“Quantum computers can get us to a deeper level of understanding where we can harness that in some useful way. I mean, what could be better than that? What’s more powerful than that? We may not live long enough to see the real use and value of this technology, but at least we can be the pioneers.”

This interview was edited and condensed for clarity.

Quantum Talk, Part 1

Me (Andrew): What’s your story, John?

Years ago, I worked in Silicon Valley for Paul Allen, the co-founder of Microsoft, in a place called Interval Research. He challenged us to invent the industries we’d see 20 years in the future. In 1995, my co-founders and I focused on computer vision and spun out a company, ePlanet, with several applications including one called FaceMe, which was the first version of the virtual background we all use on Zoom today. After that, I became CEO of that company, in which Intel invested and became a development partner. After that, I invested in a series of tech and chip companies that I either sold or brought public. While I was working in venture capital, most of my companies were in the semiconductor and radiofrequency (RF) communications areas.

Me: How did Seeqc start?

For 9 years I was on the chair of a company called HYPRES, which had its roots, many years earlier, at IBM’s superconductive electronics division. Hypres had a dual focus: RF systems and high speed and quantum computing. Eventually, we decided to split the company into two.

Half of the company continued to work with the government on RF systems, the world’s most sensitive and wideband digital receivers. The other side was focused on building high-performance computing, built around the core technology, which was Single Flux Quantum (SFQ) technology. We not only designed SFQ chips, but we also manufactured them in our foundry and integrated them into full systems.

One of the critical decisions was that we decided to spin out Seeqc with HYPRES’s superconductor foundry, which produces some of the most sophisticated superconducting electronics in the world. The only one that can compare is Lincoln Lab’s foundry at MIT, but they aren’t a commercial company. As a result of the spinout, I became Chairman and CEO of Seeqc, which stands for scalable energy-efficient quantum computing.

Me: What is SFQ?

SFQ circuits are built around Josephson junctions, not transistors, which is very different from complementary metal-oxide-semiconductor (CMOS) — what everybody else uses. When Intel, Samsung, or Nvidia wants to build something, they build it all out of CMOS, but the issue with CMOS is that compared to SFQ it’s slow, generates a lot of heat that needs to be dissipated, and is noisy. SFQ — by contrast — has an almost zero noise floor, works at 3 to 5 orders of magnitude lower power (thus lower heat dissipation), and operates with zero resistance which makes it massively faster. We have built individual switches that run at 750 gigahertz while most of our systems run between 20 and 40 gigahertz, which is way faster than anything in classical computing.

Me: What is your mission at Seeqc?

Our mission is to build out our core technology around scaling energy-efficient quantum computing. What I mean by energy efficiency is in comparison to more energetic and hotter microwaves, to operate a fully digital quantum system including reading out and controlling qubits in a reduced power mode. Microwaves are inherently energetic and therefore generate high levels of heat. This works fine with a small number of qubits, but when you want to scale to large numbers, you can’t flood the same place that you’re trying to keep cold at 20 milliKelvin with lots of excess heat. This is why companies such as Intel have built readout and control circuits that must operate at 3 Kelvin, not at 20 milliKelvin. There’s too much heat!

By contrast, we can build multi-chip modules where we take our classical superconductive electronic chips to execute readout and control and other functions critical to the operation of a quantum system, and we put them right on top of qubits, just like a chip sandwich. We use Indium-based superconductive bumps and capacitance to control those qubits. So when you see those massive computers, with tons of wires hanging from them, that kind of look like Frankenstein machines? We don’t have any of that. Our systems are all chip-based.

For example, our SFQ circuitry is used to control qubits but the entire system is chip-based. We can send out SFQ pulses that are operating at picosecond speed and rotate qubits in the same way that microwaves do, except with a lot less energy, much higher speed, and at a fraction of the cost. Since we do this at chip-scale, we don’t need all those racks of equipment you see in our competitor’s quantum systems. Everything we do is on a chip or a series of connected chips.

Our classical co-processors, which are SFQ based and clocked to operate at 10–40GHz, are in the same dilution refrigerator, so we can run algorithms that are both classical and quantum without going to the cloud to access those classical resources. A lot of quantum applications have quantum front ends with a classical back end. And designers want to get a quantum answer and iterate. You don’t want to have to access the cloud to get your classical resource when we can give it to you operating at a massive speed in the same dilution refrigerator with your qubits.

As systems engineers, our approach is not to look at how many qubits we can have; rather, how can we architect an enterprise-grade computer from scratch? Instead of looking to be the 1st-generation quantum computers (QCs), we’re trying to go beyond those gen-1 goals:

  • Can you build a quantum computer?
  • Can that computer run an algorithm?
  • Can it do anything that a classical computer can’t do?

By going to the chip-scale and substituting SFQ for CMOS, we’re able to manage complexity, cost, interference, power dissipation, I/O count, and latency — all the things nobody thought about with 1st-generation quantum computers. So that’s what we’re doing, we are not only designing SFQ chips, these 9-layered circuits, but we also manufacture them in our foundry, integrate them into full systems, and deliver them to customers. It’s a much more complex process, but we have the team, know-how, IP, and infrastructure for all for it, and we are definitely up for the challenge.

Me: What is your roadmap at Seeqc looking like?

Our strategy is not to try to build general-purpose fault-tolerant QCs. We’re trying to build bespoke scalable computers. The idea is to find a specific business problem, build out the chips that are custom designed for that application, and scale it. And we’ve raised capital with tier-1 investors like EQT and BlueYard and industrial partners like Merck, LG, and others who are looking to solve their problems.

We’re developing SFQ to be in control of 2-qubit gates, so we can build Seeqc’s platform. Between the end of Q3 and the beginning of Q4, there will be some exciting announcements, and we’re gonna be doing a story soon about some work that we did with Riverlane.

Alex Challans: What is the differentiator in your system’s architecture?

It all comes back to our platform being chip-scale. We are bringing classical resources into quantum computers at tremendous speed. That speed advantage gives us the ability to examine circuits in a way that other systems simply can’t. Now, with ion traps you’ve got a much longer coherence time, but with much slower gate speeds. I find it funny that if you talk to others in the quantum community, everybody has their own religion for their architecture — they believe there’s some exclusivity and it feels like people think everyone else’s way of doing things is illegitimate. Well, I believe that all religions can coexist. Everyone should find their truth and learn to live with each other. What we’re going to discover is that these different quantum modalities or systems are going to be uniquely good at some things and not with others.

With our superconductive gate-based quantum computers, our coherence times aren’t as good at ion traps, but our gates are great. So we’ll be useful in all kinds of ways that a QC with that system won’t be able to and vice versa. We don’t need to build it all, we just need to focus on what we’re all good at. That way, we can work together to make this ecosystem cohere. And I think it’s going to happen as it did with the classical computing community. There isn’t a winner-take-all. There are chip companies, there are software companies, operating system (OS) companies, system networking companies, memory companies, and so on. Within each region of the market, there are specialists. Companies figured out they have to fine-tune to meet specific customer requirements.

Me: Didn’t IBM start as the outright winner of classical computing though?

I disagree. IBM didn’t invent the transistor, integrated circuit, microprocessor, or Ethernet. Sure, for the Fortune 500 companies that required mainframe computers, they were dominant, but they didn’t build everything.

In the early days, IBM wasn’t so wonderful with software. There were all kinds of companies doing important technical work such as Radio Corporation of America (RCA), National Cash Register (NCR), and Digital Equipment Corporation (DEC). In fact, DEC invented the mini computer and ate IBM’s lunch! So you have to think about the broad evolution of this industry before comparing it to quantum. If you’re not studying history, you’re going to use the VCs of today’s argument and say, “winner take all! It’s all about building the moat.” There are lots of markets where that may be true, but quantum computing is not a single market, it’s an entire ecosystem. We’re gonna have quantum internet, semiconductor, and software companies among many others.

Alex Challans: I have heard that some QPU companies would keep quiet about achieving quantum advantage so they can apply it to their internal development, to create the world’s largest computational moat. It’s a silly thought though. I agree with your points around the value chain, but the question here around the step change and quantum advantage could be creating a whole set of new considerations is the story of history as you’re painting it, with Fairchild through to the various companies today. Do you think that that’s going to play out? Or do you think that there’s a risk that there is some complete change in revenue models and market dynamics?

I think that the revenue model is different than the technological development model. For example, it seems pretty clear that the AWS model makes an awful lot of sense for the quantum community. The winner could be Amazon, Google, IBM, or Microsoft, in terms of being able to distribute quantum resources through a cloud network. It could be! Or it could be somebody we don’t know — someone who is uniquely good at networking, and that can bring unique advantages to delivering quantum resources to wherever they’re needed. The technology development model is different from the business development model.

Me: Well, I guess we need to get to that point where we have an open-source community as we have with classical computing. Right?

Yeah! In some respects, it’s already happening! It’s really interesting to see how IBM is behaving. I think in many respects, IBM is trying to open up — in lots of publicly facing ways, IBM is doing a great job in opening up quantum computing, and others too. I believe people understand how difficult this is going to be so having an open-source notion around it is important.

Let me be clear. We’re not full-stack for a few reasons:

  1. Why should we develop an OS when the best software engineers in world-building OS exist to do it? For instance, companies like Riverlane are doing great work.
  2. Why should we create applications? When you know, what we’re good at is system engineering for scaling complex electronics, and mixing that in with RF and quantum. We have a long history of utilizing our IP which has given us unique advantages.

So we shouldn’t try to do things we’re not good at.

Me: Trying to do too much at the beginning never works well. I assume you feel really strong about concentrating on one thing at the start.

It’s funny because when we started we deeply considered becoming a highly verticalized company — to take on one particular application in quantum chemistry or build a quantum-based climate change model. The problem is that it’s having us do things we’re just not good at versus doing the things that we think we’re maybe among the best at. We decided to put that aside, focus on one thing, and do everything for that application. We’re focusing on the things we’re historically good at, and we’re going to build off of that.

Alex Challans: It sounds like you’ve already overcome some of the main challenges in quantum technologies, particularly scalability. What are the key areas you need to deal with in the upcoming years to build a useful chip?

I wouldn’t say that we’ve overcome them from an execution perspective — we’re developing it. All of the chips we’ve talked about work, but now we need to improve their fidelities and yield — pick and shovel engineering things. I believe our architecture has been designed to scale in a very significant way. But we still have engineering challenges in building these chips, and we will need a lot of quantum resources to make all this work at a commercial level.

We think it’s going to take us 2 to 3 years to get to the point where we have a toy-scale version of a scalable quantum computer. It may have fewer qubits than whatever anybody else has at that time, but the difference is that it’s been designed for a particular purpose, and it’s been designed from the start to scale to a large number. That’s why companies like Merck invested in us — when they were investigating quantum computing, they had a team of 50 people talking to every quantum company out there to solve this problem. It’s one thing to build a really useful prototype to show what things might be. But how can we actually address their complex problems? We had a prototype that could exemplify a roadmap on what we can build. That’s why they chose us.

Alex Challans: What has been the impact of spinning out of a foundry?

I worked as a venture capitalist for 20 years, and I often funded fabless semiconductor companies. When you are not the owner of your foundry, you don’t own the processes, you don’t own the underlying technology for it, you’re dependent on their scheduling and you don’t have control over their yield. We have all of that.

Our foundry is for us, our customers, and for building unique quantum computers. We have a test center that is literally across the street from our chip foundry in Westchester. So, when we have an idea, we tape it out and just do it. At any one moment, we’re running an enormous number of experiments through our foundry, in a rapid cycle of iteration. We’re not a big or high throughput foundry, but we are a state-of-the-art development stage foundry. We have all the things to work with. For instance, we’re experimenting with building fluxonium qubits using Tantalum, which for us, is a new metal for our process.

Alex Challans: Why didn’t you just work as a VC for the rest of your life? What drove you to become a CEO of a deep tech company?

It’s funny because I didn’t want to be the CEO, I was just thinking I’d be the chairman and recruit someone to do the CEO job. But in life, there aren’t so many opportunities to work on truly interesting things. At some point, you realize this is the most consequential thing you have ever come across, and that you have the privilege to work on it. So that’s how I look at it, I feel super lucky to be able to do this — that quantum computing came along, was developed at a certain level, and we were involved. Also, I have been working with my CTO and co-founder, Oleg Mukhanov, for 10 years, and we enjoy working with each other. So it was a way we could continue working with each other, with the bonus of working on the most important new piece of technology that I’ve ever had the privilege to work on.

Alex Challans: I agree with everything you say. Quantum technology is so exciting as it can solve problems that you just couldn’t imagine solving before. What do you think about it?

I often think that as humans, we are really good tool builders. That’s one of the ways we advance as a species — from fire all the way to supercomputers. We now exist in a world where we have figured out how to process nearly everything in zeros and ones, which is miraculous. Everything we consume is effectively down to zero and one. What’s more sophisticated than that?

But Richard Feynman reminded us that nature is not classical, it’s quantum mechanical, so if we want to understand the world we live in, we need to make quantum mechanical computers. Thinking about the complexity difference between the classical world and the quantum world… It’s night and day. It’s an entirely different view of reality.

Every day, we live in a very Newtonian world — where it’s all very explainable, except it doesn’t explain all of nature. Quantum computers can get us to a deeper level of understanding where we can harness that in some useful ways. I mean, what could be better than that? What’s more powerful than that? We may not live long enough to see the real use and value of this technology, but at least we can be the pioneers.

Deep Tech’s Enterprise Future, Part 2

To better understand how companies are managing to investigate deep tech, Seeqc surveyed more than 200 decision-makers at large enterprises (1,000 or more employees) across various verticals who were actively investigating deep tech solutions. The results illustrate deep tech’s burgeoning landscape — one rich with opportunity, but fraught with risk. The report, Out of the Lab, Into the Market: Deep Tech’s Enterprise Future, explores this.

Me: Now getting back to your deep tech research, why did you guys embark on that?

Because we’re interested in application-specific quantum computing at an enterprise level, we wanted to understand:

  • Why do people want to invest?
  • Do people want to invest?
  • Are they interested in this?
  • How would they do it?
  • What would they value?
  • What expectations do they have?

Even though we’re a quantum computing company, we didn’t want to just ask about quantum computers. We wanted to contextualize from a deep tech framework. Frankly, within a given company, it’s competitive because people are saying, “I want to work on this thing,” or “This is important for our business,” so we had to think about the other initiatives that people would take on.

We can’t think of what we’re doing as the only thing, because it’s not. For example, almost everybody is working on something having to do with AI and machine learning. We wanted to broaden out to deep tech, first, to learn what priorities people have, and second being to figure out large enterprises’ deep tech journey by asking questions like:

  • How are you going to navigate the space?
  • Who are you going to believe?
  • Who do people partner with when they want to outsource information?
  • Who are you going to look for and what is it that you are looking for?

Me: What we’re some surprising things you found?

People want to know your intellectual property position. Do you have a good and deep intellectual property in the area and do you have the ability, facilities, and capabilities to do development and testing? They want that. This wasn’t obvious to us.

People are focused on high-value problems. Similar to our situation with Merck and LG, they can tell you exactly what areas they want to focus on with quantum computers, and what’s motivating them is not only the idea of what those opportunities are but FOMO, more specifically fear of what your competitors are doing.

Another very interesting part that has implications for almost everyone is where people get their information. Who do they trust? Overwhelmingly, they trust 3rd-party consulting companies, and I would assume it’s like BCG, McKinsey, Bain, and a few others. We didn’t know if it was academia, government studies, or something else, but consulting companies reign supreme.

About 1/3 of the companies chose quantum computing as one of their areas. We thought it would be more narrow. If we had asked this question 3 to 5 years ago, it would have been single digits. The fact that they do care about quantum computing, that they’re not just looking at developing technology for technology’s sake, was surprising.

Probably our biggest surprise is people think they should be able to get an ROI within 1 to 3 years. We don’t know how they see this regarding quantum computing, but they want results quickly. This timeline is so dependent on an individual technology, though.

Me: It’s shocking that a VC firm, focused on investing in cutting-edge technology, would expect returns in 1–3 years.

Well I mean, if you’re working on databases, there are fewer things you could probably find an application within AI in the next few years. Quantum computing is farther out. Blockchain and cryptocurrency… you could see returns now. I don’t know about drones and advanced robotics. But my point is that some of these things are nearer-term than others.

Me: Regarding robotics, I can’t imagine ROI in 1–3 years.

There are lots of industrial settings where robotics have been well integrated into manufacturing. The data says that 46% of people said 3 to 5 years is when they expect to receive good ROI from a project, but a third said 5 to 10 years, and 10% said 10 years or longer. The 43% that said 5 years or longer have a more realistic timeframe for a lot.

We’re curious if we should do this same survey, again in a year or two, because my guess is we’re going to find even more companies working in quantum.

Leave a Reply
Previous Post
The Spanish Network on Quantum Information: Promoting Research All Across The Bull Skin

The Spanish Network on Quantum Information: Promoting Research All Across The Bull Skin

Next Post
Xanadu Awarded DARPA Grant to Develop Novel Quantum Compiler for NISQ-based Machines

Xanadu Awarded DARPA Grant to Develop Novel Quantum Compiler for NISQ-based Machines

Related Posts