Listen: Next in Tech | Episode 122: Cloud delivered quantum

View Full Transcript

Eric Hanselman

Welcome to Next in Tech, an S&P Global Market Intelligence podcast, where the world of emerging tech lives. I'm your host, Eric Hanselman, chief analyst for technology, media and telecom at S&P Global Market Intelligence.

And today, we're going to be discussing cloud-delivered quantum. It's a continuation of a conversation we started a while back with Gabriella Brown, one of the analysts who covers not only cloud but also quantum. Ellie, welcome back to the podcast.

Gabriella Brown

Awesome. Thanks for having me.

Question and Answer

Eric Hanselman

Well, it's great to have you here. There's been a lot that's going on in terms of being able to expand access to quantum that if you don't have a large Dura of liquid helium sitting next to your desk, that there are ways to be able to get to quantum capacity outside of owning your own system, that not surprisingly, the various cloud providers and now have a whole set of additional folks are looking to be able to address what's going on out there in the market.

Gabriella Brown

All right. So we're looking specifically at cloud-delivered quantum, right? So as you mentioned, we have a lot of cloud providers that also are running quantum operations. So some of these are actually developing their own hardware. They're developing quantum systems themselves. And some of them are taking quantum systems that have been developed by quantum hardware companies.

And they're allowing access to users over the cloud. Because as you mentioned, these are huge systems with a lot of technical requirements. It just doesn't make sense for everybody to have a quantum computer in the back next to the printer.

So these cloud-delivered companies that they're allowing users to have access to quantum computers without having to have the hardware on-premises, and there's a lot of different ways that they're doing that, a lot of different services, and then the big cloud providers as well as a bunch of startups that we've seen as well.

Eric Hanselman

It seems like in the everything old is new again kind of theme that we see so often in tech, this is kind of like quantum time-sharing if we want to go way back to the mainframe days but taking all the advantages of cloud's ubiquitous access, and now being able to direct that at quantum resources in whatever way they're getting actually provided in the end system.

Gabriella Brown

Yes, exactly. And a lot of research has been done this way, right? So they've opened it up previously to researchers and universities so that they could really accelerate this drive towards quantum, accelerating the drive towards the bigger quantum systems, how we use quantum computers. And recently, we started to see kind of a bit of a shift away from just access to researchers and just access to everyday business.

So if you have a business problem, you can access a quantum computer on the cloud, you don't have to be a researcher, you don't have to be a professor at a university. You can actually access a quantum computer, share some time on it and run a couple of problems, whether you're experimenting or whether you're running it off on a quantum program within your company.

Eric Hanselman

So what kind of offerings are out there? And how do people actually put them to work?

Gabriella Brown

Yes. So we have, as you mentioned, kind of this time-sharing methodology. We've seen an increase in colocation services, so offerings where users can actually rent basically a quantum computer, rent space in a colocation data center for quantum hardware and access it that way.

We have the traditional kind of cloud as a service model, where people can access quantum as a service through cloud providers or they're just accessing that service there. And then we also have access to quantum simulators, which aren't actually quantum computers, they're classical computers that use high GPU levels in order to simulate the action of a quantum computer.

Eric Hanselman

So not actually a quantum computer but a simulation through a classical computer in terms of what it could actually do and what it would be doing if you're actually running it on a qubit. I guess, these are things that are probably lower qubit density environments, things that are probably simpler problems to solve or ones that you don't need run times that are quite so long or quite so capable. Those are sort of things that simulators typically tackle?

Gabriella Brown

Yes, exactly. So the largest simulator right now can tackle up to 256 qubits, which is actually smaller than our largest general-purpose quantum computer that we've seen. So we've already reached the point where we have quantum computers that are more capable than what classical computers can simulate. But simulators are still useful, especially in the early days of quantum, where we're trying really hard to get used to this new technology.

We're developing. We're trying to run a bunch of new development techniques. And if you did that on a quantum computer, first off, it's very resource-intensive. There's very few quantum computers physically around the world, so you'd have to take a lot of time. And so getting used to using a quantum computer without using one, right, that's what simulators are really useful for.

It's also really great for bug-fixing. If you're developing a quantum algorithm, you can run it on a simulator for small-scale problems, right, and break it down. So part of what makes quantum interesting and kind of difficult to work with is if you look at what's happening during a computation, it actually destroys the process.

And you can't observe a quantum computer operating. It's against the laws of quantum mechanics. And so a simulator is classical computing. You can pause the program while it's running. You can dive into what's happening without breaking it, right? So simulators are really useful for small-scale problems because of that act.

Eric Hanselman

That's a fascinating point. Because it's not just that they're a bit less expensive because you're not actually running on the real quantum computer, but you actually get to look inside the process, take a look at how the circuit is actually operating, take a look at what's actually being evaluated in ways that will help you prove out what you're trying to achieve. And then when you get it upscaled, you run it on a real quantum computer to get the scale and the end effect that you're shooting for?

Gabriella Brown

Yes, exactly. So we imagine that simulators will probably have a key role to play even moving forward, even being qubit-constrained just because they'll allow you to have the access to dig into even a small piece of a bigger problem, put the problem together and then run it on a quantum computer later.

Eric Hanselman

Well, thinking about what actually is getting run, what's the situation with development tools? I know when you were here last time, we talked a little bit about development environments and really what tools are available. Does that change in cloud environments? Is that -- is there more or less difference that you have to work with when you're dealing with a cloud environment?

Gabriella Brown

Yes, I'd say that there's more. Again, a lot of these cloud environments have been built out for everyday users to access quantum, right? So the people interacting directly with a quantum computer, they're trained on quantum mechanics. They're usually physicists and certainly quantum engineers.

And so they work on building out these quantum systems, right? Whereas people who are accessing a quantum computer over the cloud, they're usually not physicists, they're usually not trained in quantum. They're trying to learn about it. Hopefully, they'll be trained in quantum in the future.

But the goal is to try and make it easier for them to interact with this quantum system in a way that they're familiar with. So you see a lot of kind of analogs to classical computing, right? We have programming languages that allow you to talk to a quantum computer. We have software development kits, simulators, hardware platforms, stuff like that.

Eric Hanselman

Ways to be able to maybe do some translation into quantum environments. I know there are things like Python tools that will allow you to take what is a classical language and start to manage that in much the same way that we've leveraged software development kits and language accelerators to be able to actually translate between the world of existing code and this very new and different way in which we actually start to evaluate problems.

Gabriella Brown

Exactly. And Python is usually the language that is used the most. It talks really well with Q#, which is the quantum programming equivalent of C#, right? It lets you interact with that programming language. We've also been seeing just recently kind of an uptick in natural language models that use AI and actually let you type in natural language through AI, allows you to interact with a quantum computer with no experience either with Python or with any quantum computing programming language.

Eric Hanselman

We've hit two buzzy topics lighting together, quantum and generative AI. Well, we're actually going to be talking -- Nick Patience is going to be our following guest talking about some of the generative AI models and aspects of sort of what those markets look like.

But that's a really fascinating thing in that so much of what we're talking about, the use of generative AI, is to help with programming languages. And here's an example in which we can actually start to take the generative AI capability of understanding how to form questions in a quantum way to be able to actually move between the classical and the quantum worlds.

Gabriella Brown

And that was the big announcement actually that came out this week from Azure. Anybody can pop on to Azure Quantum in the browser now, and you can interact with AI to develop your own quantum code and run it on a quantum simulator. So no account required, nothing like that. You could just pop on and diddle around with a quantum computer with AI to kind of help guide you through that process. So that was a very interesting development that came out just this week.

Eric Hanselman

So it sounds like part of the race in the cloud environments today is trying to ensure that there are better on-ramps so that you've got better understandings of what you -- or how you can use the quantum capabilities, but that at its root, it gets back to that same cloud provider, fundamental task of, "We want to make it easier for people to consume our capacity, to increase utilization and to be able to leverage it and pay." They're doing it now with quantum.

Gabriella Brown

Exactly. Yes, it's just another one of those services that cloud providers can provide. And it happens to be very exciting and new technology. But making it easy to access, that's kind of the key right now.

Eric Hanselman

Well, I look at other providers who've got various development environments and transition tools and those sorts of things. I mean, all of the various cloud providers have got one of their own, whether that happens to be IBM's environment, Amazon's Braket, Google's environment for their quantum setup. Each of them are trying to ensure that it says that on-ramp is as simple as possible to be able to get there.

Gabriella Brown

Yes. I mean we're dealing with a new technology, right? And even though it's not fully built out commercially, you don't want to wait until it is available commercially and broadly being used to have familiarity with the tool. So we've seen a lot of cooperation amongst quantum providers and then some of these more classical cloud providers, classical computing providers, just to really build out the research, the public access, the familiarity with the tool.

Because no one wants to be caught flat-footed when it actually hits quantum advantage and there's quantum computers being used in everyday business. You want to train the workforce. You want to train your business to recognize which problems would be ideal for quantum.

And so that's one of the great uses of the cloud actually, it's really good from the quantum provider side, is through the cloud that training can happen, that education can happen, that familiarity can happen while these systems are being built bigger.

Eric Hanselman

And it's also that cloud paradigm of being able to give you access to new technologies in ways that are much easier to consume. The latest hardware -- and hey, in this case, the latest hardware just happens to be a large quantum computer.

Gabriella Brown

Yes, exactly.

Eric Hanselman

Well, you mentioned colocation. Because there are also -- colocation providers are starting to get into some of this as being sort of the intermediaries. And there are a number of groups that are starting to provide this as well. So it seems like there are a lot of different angles to how we can actually provide quantum functionality in all of these different forms in different access mechanisms, depending upon how people actually want to get to them.

Gabriella Brown

Yes, exactly. So the nice thing is there's a lot of options now that just weren't around a couple of years ago. So depending on what your use cases are for quantum, whether you need full access to a lot of quantum systems because you're planning out your business proposals and you're trying to figure out how quantum can solve a specific business problem.

Or whether you're just trying to fiddle around and first testing something out, there's a lot of different ways that you can access quantum over the cloud and just depending on what works best for you and where you're at in your quantum journey, I suppose.

Eric Hanselman

Again, as we start solving more complex problems, people get concerned about proprietary data and where it gets exposed. So do you want to be loading up your quantum experiments on a public cloud? Do you want to be able to do something that's a little more private? The big question then, of course, is how do you get that data moved back and forth?

And we've got quantum networks that are coming into the conversation here as well, I mean, not required to be able to work with quantum computers, but that are offering some interesting benefits based on quantum networking capabilities.

Gabriella Brown

Yes, exactly. I mean, any time you're dealing with any sort of hardware or computer, the next question naturally is, "Okay, well, how do these computers talk to each other, right? How do we share information in a safe and secure way?"

Especially with quantum, that's one of the big questions is, "Well, what about quantum security? What about encryption? Is quantum going to break all of our codes? Is this a problem?" Well, the nice thing is the flip side of this technology is there's ways to actually increase the security of the system using quantum mechanical properties.

And in quantum networks, it's actually using those properties to help communicate information very securely from one point to another without dealing with all of the issues about hacking, security encryption, stuff like that. So it's just based on the actual quantum mechanics, the properties of these particles that we're working with, enables quantum security in networks.

Eric Hanselman

Well, there's that interesting aspect that you actually touched on it, which is that one of the challenges in debugging quantum operations is that the active observing actually fixes the state and breaks whatever the process is as well in quantum communications, the act of observing the communications, as in eavesdropping, actually breaks the communications. And you know when you've actually been eavesdropped on.

Gabriella Brown

Yes, exactly. So that's one of the, I guess, turning lemons into lemonade versions of quantum computing, right? It's you can't dig into the system. But that means no one else can dig into it either, right? So that allows for much greater communication security. So when it comes to quantum networks, we're actually seeing a really big push, less from the commercial side and more from governments that are really interested in using this to enable secure communication at cities, towns.

I just talked to the city of Chattanooga, which they launched the first commercial quantum network in the city of Chattanooga. So that's going to be really interesting to see quantum networks really get built out. And it's probably going to be more at a government and city level than necessarily at a commercial level.

Eric Hanselman

And still experimenting to start. Well, it does mean that you've got to have specialized capabilities within the network that aren't going to mangle photons in transit. Because it's photons, it's the base of that communication piece, you've actually got -- it's not just a regular optical network.

You actually have to ensure that, that same photon that you -- that starts at the beginning actually can make it all the way through and make sure that you've actually got a network that can ensure that it gets carried and not -- because a lot of times when you start going through optical networks, you wind up transiting back and forth between the optical realm and the electrical realm.

You may have switches and repeaters and other things that are in there. But this, in a quantum network, you've actually got to make sure that it is not only all optical, but that it is that same photon that's going from one side to the other.

Gabriella Brown

Yes, exactly. So we have several companies that have to work together in order to make quantum networks possible, right? So we have quantum vendors that sell optical components. Because that is still part of the network, not the main part, right?

You're dealing with photons, and you have to modify everything so that it actually works as you pointed out. But we have quantum control plane software. And we need photon detectors that detect just the photons that we're carrying about and not all the photons that happen to be around. And we have optical switches.

So quantum networks are really unique because there's this really big level of cooperation. There's no really just one company that can build out a quantum network. It's a bunch of smaller quantum companies moving into the network space and really trying to build it out in a way that's useful.

Eric Hanselman

It will be interesting to see what Chattanooga is able to do with it. I guess, it's their power company that was looking at actually stringing up the network itself but looking at that as a secure vehicle, but yet opening it up to experimentation to folks who want to work with it and actually explore what quantum networks can do for them.

Gabriella Brown

Yes, exactly. And it's a very future-looking mindset, right? They mentioned that they're taking the long bet. They want to be prepared. The big focus actually was on load, so how much data is being carried on their networks as a city and electric vehicles and self-driving vehicles, right? So anytime you have a new car, it has a huge data pool. It builds a bunch of data.

And as these smart cars, as these smart grids, as these smart everything build out, we're going to have all of this data which requires computing, which requires really robust networks. And at least Chattanooga believes that, that future is going to be in quantum networks as opposed to classical networks to help balance that load and help keep things running smoothly.

Eric Hanselman

Well, it's one of those things that you sort of figure the complex problems that you have to solve for running an electric grid is that sort of balancing problem. It's the same kind of things that quantum computing is particularly good at, multi-solvers around finding minimum and all those kinds of things that it can do well. And now they've got the communications to actually move that data to the aim for faster and better ways to be able to manage the computational piece as well.

Gabriella Brown

Exactly.

Eric Hanselman

So what should organizations be thinking about when they're looking at leveraging quantum services? You talked a little bit about getting to the point at which we're experimenting. But what should they be considering?

Gabriella Brown

Well, I mean, that's the big question, right, is how do we transition from science and scientific discovery into building this out commercially for organizations? Where might these services be best leveraged? There's been a lot of focus on, as you mentioned, optimization problems at natural modeling.

So any services, any organizations that provide chemistry services, material science, anything that's very physics-heavy, you can actually get a lot of really easy heavy lift out of quantum computers just based on the nature of the problems that you're dealing with, right? So there's a lot of thought going into how can these be applied to medicine, how can we use quantum computers to try and help optimize drug development, help optimize fuel efficiency.

Any natural system that you're looking into, quantum computing can really help boost that. So if you're in really complex situations, if you have a problem that is really complex as an organization, as a business that just takes too long to solve on classical computers or is not solvable using current classical supercomputers, quantum probably is going to be a good fit.

So we kind of have a two-pronged approach, right? Some industries are really looking at quantum and saying, "Well, how would this be useful for us?" And other industries are like, "Yes, this is definitely going to be useful." Now we need to figure out how we build these problems out so that as soon as we hit quantum computing capacity that's high enough, we can just hit go and run these problems and just really accelerate forward.

So it's kind of a bit of a two-pronged approach, depending on the industry you're in, whether you're still testing out those use cases or whether the use cases have been established and now you're just building out that algorithm so that as soon as the quantum computer hits the scale you need it to, you can run it and go.

Eric Hanselman

It sounds like cloud-delivered quantum capabilities could be a useful on-ramp to help people figure out how they approach those problems and whether or not practically, there are things they can actually master.

Gabriella Brown

Yes, exactly. It lets you get kind of that front-seat, interactive view where you can fiddle all around, you can test that out without having to own a quantum computer, having to wait for it to be small enough, temperature-stable enough to exist on-premise. You can access it now.

Eric Hanselman

Probably won't be long before we've got quantum services included in the cloud price index, I would guess.

Gabriella Brown

We actually already do, so...

Eric Hanselman

Oh, shoot. All right, well then.

Gabriella Brown

Yes, we track those on a month-to-month basis, where we look at the new services being added, any price changes to those services. So yes, we're already on it.

Eric Hanselman

All right then, cool. All right. Well, one more thing to add to the value chain in the Cloud Price Index, good stuff. Well, this has been great, Ellie. Many thanks, appreciate all the insights.

Gabriella Brown

Yes. Thank you for having me.

Eric Hanselman

And that is it for this episode of Next in Tech. Thanks to our audience for staying with us. And thanks to our production team, including Caroline Wright and Ethan Zimman on the marketing and events teams and our agency partner, the One Nine Nine.

I hope you'll join us for our next episode, where we'll be talking about generative AI, the next buzzy topic, with Nick Patience, who will back on to talk about a new market monitor assessment of the market. And I hope you'll join us then because there is always something next in tech.