Technology, Connected - How IonQ’s Trapped-Ion Quantum Computers Work: Coleman Collins on Qubits, Fidelity and Quantum Advantage

Episode Date: February 28, 2025

Coleman Collins, Director of Product at IonQ, joins Thinking on Paper to explain how trapped-ion quantum computers work and why IonQ believes they offer a credible path towards useful quantum computing.IonQ builds its systems using individual atoms held in place and controlled with lasers. This differs from the superconducting, neutral-atom, photonic and topological approaches being pursued by companies including IBM, Google, Microsoft and D-Wave.In this episode, Coleman discusses how these competing quantum-computing architectures should be evaluated and why headline qubit counts reveal little about a machine’s practical performance.We cover:How trapped-ion quantum computing worksWhy fidelity, connectivity and control matter more than raw qubit countsWhat IonQ means by “algorithmic qubits”How lasers are used to control and connect trapped-ion qubitsThe technical criteria required for a viable quantum computerWhen quantum computers could deliver a practical advantageWhether quantum computing could eventually threaten modern cryptographyHow IonQ’s approach compares with superconducting and topological quantum computingWhat investors misunderstand about quantum-computing performanceHow developers can begin working with quantum systems todayWe also examine Microsoft’s Majorana-based strategy, the limitations of current quantum hardware and the difference between measurable technical progress and corporate claims.This is a practical discussion about IonQ, trapped-ion quantum computing and the race to build commercially useful quantum computers.Please enjoy the show.--Chapters:(00:00) Introduction to Quantum Computing and IonQ(02:17) Understanding Trapped Ion Quantum Computing(04:57) DiVincenzo's 5-Step Criteria for Quantum Computers(07:31) The Natural Aspect of Quantum Computing(10:01) Algorithmic Qubits vs Physical Qubits(12:49) Achieving Quantum Advantage(15:04) Investment Trends in Quantum Computing(17:44) The Role of Education in Quantum Investment(20:40) Hot Buttons(25:57) Topological Quantum(30:47) Microsoft's Majorana Quantum Chip(34:09) Engaging Developers in Quantum(36:59) Hybrid Quantum Computing and Integration(38:55) The Quantum Promised Land(41:19) Can Quantum Hack Bitcoin?(45:09) Could Quantum Currency Exist?--www.thinkingonpaper.xyz

Disruptors and Curious Minds. Welcome to another episode of Thinking on Paper, where we get you our curious listeners behind the scenes backstage on some of the latest and greatest new technologies that are going to change the way we look at the world. We're doing it to offer you a path to see what's coming up, what's going on in the future. Some of these will be a little more hard to listen to,

technological bits and widgets thrown at you, but they will always be interesting. They will always be impactful. and get you using this thing right here. Mark Fielding, how are you today, man? What are we learning about? Yeah, think for yourself at thinking on paper. After a few weeks in space,

we're going back down to the quantum realm today, Jeremy and our listeners are going to be learning about full-stack quantum computing, application-driven quantum, the difference between algorithmic cubits and physical qubits. I have something written on a piece of paper here called quantum imaginary time evolution algorithms might get into that. Whoa, don't go anywhere.

We're talking to IONQ. Coleman Collins, he's a director of product management at IONQ. And Jeremy, I have my bell again, and this is more for me, perhaps, today. Because if we start talking about sending our consciousness into the deepest, darkest depths of the universe, using superposition to time travel there, stop me, stop you. Let's stay in reality a little bit. Well, if I, I need to harness some sort of superposition so I could take control of your bell and ring it because I can't, you know, I can't do that. I don't have a bell on my side.

But yes, let's do that. Well, both ring the bell and not ring the bell. Fair enough. Fair enough. Is the cat alive or dead? All right. So without further ado, Coleman Collins will bring you on. Let's talk about ion Q.

Let's talk about your approach and get into quantum. Good. For our listeners, it's Coleman is joining us from the west coast of America. because it's quite early for him. So maybe we should start off a little bit easily this morning, Jeremy. And what is ion queue and how are you different to our former guest, D-Wave, IBM, Horizon? Yeah. So there are a few answers to that question.

Most practically, we are a company that makes quantum computers, as well as a number of other supporting technologies that we sell and sort of different packages and bundles depending on who's using it, why, and where. We've been around since 2017. We were spent out from university research at University of Maryland and Duke. And we'd make trapped ion quantum computers specifically. So what sort of the two big ways were different. One is our hardware approach and then one is our sort of philosophical approach to solving the problems.

Hardware approach, we use trapped ions. So trapped ion quantum computing, there's a lot of different ways to make a cubit, right? all you really need, there's this list called Devin Chenzo's criteria. You need five things plus two if you want to do networked quantum computing. As long as you have those five, you can use a photon. You can use a non-abillion anion, which is the thing that's in the news with Microsoft recently. You can use neutral atoms. You can use superconducting circuits.

So basically a loop of wire that, I mean on the chip that has this discontinuity called the Joseph's junction that you use to then sort of create atom-like behavior. We use trapped ions. We use physical single ions of uterbium or barium, depending on which generation of system you're talking about. And then we address those with lasers. And that does math, turns out. Okay, hold on. Right. We'll get to Microsoft later. I'm sorry, Mark. I'm going to stop you. Those five steps. We need to know what those five things are. Real quick, real quick, Colma, before that, Mark, have you ever been addressed by a laser before?

I was addressed as Mark Fielding five seconds ago, and that was weird enough, so no, I've never been addressed by a laser. Have you? I have it. Coleman, what, talk about Devencenzas? Yeah, so Devinchenzo. David Devinchenzo is a theoretical physicist. I forget where he works or worked. But basically, this is the list of the things you have to have to have a quantum computer. And so you need a well-characterized cubit. So you need a two-level system, quantum system. that you can either be in like a fiducial one, a zero, or somewhere in between, super position. You need to be able to initialize those qubits to a single state to know where they are. You kind of know where your zero is. And that's just like so you have a reference point to then do all the other stuff. You need long coherence time. So what this means is you may be talked about this before, but many different types of two quantum states sort of relax

over time, right? One of the challenges of quantum computing is that you quite literally need to isolate the quantum state from the universe. And there are many reasons why that stops being true. One is sort of coherence from interference of like electrical fields, magnetic fields, solar radiation, cosmic rays is actually a real problem in superconducting system. And you also just, depending on what kind of quantum state you're using, sometimes it's relax, right? So you start losing the ability to tell the difference between your one and your zero again. and you're going back to basically random. And so you need both kinds of coherence there.

That's three. You need to be able to measure ideally individually, but at least all of them. And then you need what's called a universal set of quantum gates. And so you need an entangling gate and you need a single cubic rotation of some form. And generally that's kind of enough. And that's basically so you can kind of explore the entire what's called the Hilbert space, the sort of the mathematical space that you're trying to compute in kind of efficiently. And so you need all five things.

But there's a lot of different ways to do that. You can do that with photons. You can do that with squeeze states. You can do it so on. We use trapped ions. And the kind of the reason the logic behind this is that we are cubits don't have any fabrication errors, right? They are made by nature. Good to go.

All of the errors that come in are from environmental control, environmental interference, and then basically our inability to actually perfectly, like you can like literally we are trapping ions on a chip. They're floating like, I don't know, it's like 40 microns above a chip that we design and then Fab with a partner, and then we're shooting lasers at them. And they're effectively a single point source, but we need to be able to exactly hit those with exactly the right wavelength, exact the right phase, drive these transitions between multiple ions to entangle them, all of which is that's the engineering challenge, right?

Like our founders like to say that we have no new novel physics to figure out, just a lot of really hard engineering. But to be clear, it's pretty hard engineering, like, getting there. So, yeah, so we use trapped ions, and that's a philosophical choice, basically. There's a lot of, the way I like describe it is there are everyone who is sort of driving, especially kind of in the big commercial space, everyone is picking the hard problems they think are the easiest. And we think that's sort of cubic control rather than cubic fabrication, which is what a lot of your kind of superconducting people are thinking. I think that that goes without saying like a lot of people focus on the easier thing than the hard thing sometimes just to get the things done. Look, we're doing things. We're doing things. A couple quick thoughts. Disruptors and curious minds, don't ever say we didn't give you anything. We actually, thanks to Coleman, you now have the checklist to build a quantum computer. So brought to you by thinking on paper.

Secondly, Coleman, I've always thought we've read a lot about quantum mechanics. I've been a quantum mechanical nerd for a long time. I mean, I'm not a physicist, but I love the idea of science that is emergent and not hierarchical. And it kind of just does stuff because that's what it does. More jazz, less classical. So let's unpack the natural aspect of using ions, right? Because nature does quantum better than any man-made type of quantum. So you're now kind of harnessing a bit of nature and hoping to get results that nature does

because there's no cooling, there's no massive cooling challenges that you have with this type of technology, right? Or they're not as difficult to overcome. They're not as difficult. That's correct. So we do, actually, we just, I think, announced maybe today some of our, our work to get back to room temperatures. We use what we sort of call easy cryo, but the reason we use that is primarily for vacuum pressure purposes because you need to isolate your ions. If you have a collision, then you lose an ion and then you lose your computation. We don't need to be like inherently

in all you can actually cool ions with lasers. It's called Doppler cooling, which is a pretty neat idea. It uses the same effect as like, you know, the Doppler effect, like an ambulance is driving by. Yeah. So specifically what you can do is basically detune a laser such that it only is seeing, the ion is only seeing the laser, so it's only resonant in one kind of emotional direction. And then you can use that to basically bleed off all of this motion over time, because only when it's going towards the laser does it kind of get a photon and then release it, but it's moving randomly. So over time, you can compress that down to, you know, basically a couple of quanta of energy in that ion. And then that's actually better than trying to use a cryostat to do it.

Is this a little path integral kind of thing happening? It's actually the opposite. So based on the species of ion, there is just math you can do that lets you know what that wavelength is. Got it. Yeah. And then to detune it, you're just sort of moving it slightly off. And that's then the Doppler effect, redshifting, then lets you only in sort of one emotional direction. Are you sort of catching that wavelength at the right kind of perceived? The ion sees it now at the right wavelength rather than a slightly wrong one.

Jeremy used to ask this really great question to our guests. And then we learned and then we don't ask it anymore. but you lost me like the quality over quantity that IonQ. Yeah, yeah, yeah. But then you say you don't need these calling mechanisms, which we all as quantum curious, that's what we imagine. So there's like this contradiction there between what we think we know and what you're doing.

Could you explain it as you'd explain it at the family dinner table at Christmas to your uncle when he asks, so Coleman, tell me about INQ, why is it different to IBM? Sure, yeah. this is kind of a, well, I know the question you're actually asking, but I'll note that my uncle works in an accelerator lab. No, pick a different relative. Pick a different relative. I'll pick a different uncle. Okay, what do the rest of the family do when you two start talking to each other? Exactly. So we, yeah, we are ultimately trying to do the same thing, right, which is build large quantum computers and then also create, basically make them very easy to use such that not only are really phenomenal applications,

research team can develop use cases that actually impact the world, but everyone else can too. This is the way in which we are the same. The ways in which we are different primarily is hardware technology. So this is the trapped eye and stuff I'm talking about. And basically rather than choosing to fabricate things that act like atoms, we choose to use atoms. And that comes with a different set of benefits, but also a different set of challenges. Some of the benefits are we think it's going to be better, it's easier to scale. We think it is, it's definitely easier to fabricate the cubits because we can just basically, all you need to do is get a pure source of whatever metal you're using. It's to some extent easier to measure, readout, control. On the flip side, the cross-stock is much better normally,

but that's actually kind of an architectural decision. On the flip side, it is harder to make very, oh, also, sorry, one of the big things about trapped ions, at least the way that we do it is connectivity. So effectively because we're using an ion chain and we're using the emotional modes of the ion chain to drive our entangling gates, we can pick any two on the chain and have them entangled as opposed to a chip layout cubit where you're limited by what you're actually physically connected to by wires. We don't have any wires. The lasers are the wires and physics is the wires. And the lasers can actually be, so you're not hardwiring something in, they can be changed to coordinate the qubits differently. Is that, is that the idea? So we use what's called Acousto Optic Deflector Technology to basically steer laser beams. Wow, super cool. Yeah, I think so too. Like, to be clear, my background is not in physics.

And I'm not one of the people that designs those optomechanical systems, but I love going to the design reviews. Can we go into this idea of algorithmic cubits and physical cubits? And I want to start off by using an analogy that came into my brain as I was learning about these things, right? So qubits, no matter what way they're put together or the type of qubits they are, it's this new way of computing that we're trying to get better at what we do than we do today. But the challenge is knowing whether the cubits are actually useful, right? Would this be akin to testing the capabilities of a race car before you got in and ran a race with the race car? Is that kind of the same idea what algorithmic cubits do is test the usefulness? Is that what we're getting with us?

Yes, yes, exactly. We are, algorithm of qubits is just sort of our shorthand single number metric to try and describe the number of useful cubits we have in our system. And the short version of how we do that is we take a set of benchmarks that were developed by the QEDC, which stands for Quantum Economic Development Council, I'm pretty sure. And then we did a little bit of extra stuff to basically be able to distill that to a single number that describes kind of how big your kind of run box of your quantum circuit can be. as actually measured by specific kind of reference algorithms that we know to have some potential interesting use cases in the future. Things like phase estimation, things like amplitude estimation, these sort of algorithmic quantum Fourier transforms, these algorithmic primitives that then you can kind of use as building blocks to solve interesting problems. And so AQ is just our way of

kind of describing that run box. There are other ways to do it. They all sort of have their pros and cons, right? Like as an example, quantum volume. We have nothing wrong with quantum volume. It just, that is actually trying to describe sort of the latent space run box. So kind of all of the randomness inside of that. And it's, it's a, I blanked for a second. It's a useful test, but it's a less useful test because it's not as focused on what are you going to do with this, right? Because I don't buy a TV to just like look at random static. I buy a TV to look at pictures. And so you could actually, if you allow yourself to, without actually specifically optimizing for any individual algorithm, that's sort of against the rules in the rules that we wrote for ourselves,

I am at least able to optimize for the fact that I'm trying to solve problems rather than just explore randomness. So it's like a tuned television as opposed to a fuzzy screen. So Jeremy, you're very good at summarizing. So summarize for me an algorithmic qubit, how you understand understand it? I just see it is I would kind of lean on my previous analogy and maybe, you know, tie into what Coleman was saying just a little bit before that, you know, you could have, have 500 cubits, but if only like two of them can be put together for you point to useful compute, the volume isn't working. It's a quality over quantity approach and finding a way to define in their own terms what is a question.

quality qubit, I think. Yeah. How did I do? It's really quite well, I think. So volume is sort of an interesting one because it actually, that's a really challenging benchmark too, to be clear. But it's not as useful from like a customer perspective or a user perspective of what can I do with this, aside from like run quantum volume at that size. It's similar to like the cross-entropy benchmarking, some of the stuff that Google has

used to benchmark their systems, which is, to be clear, a really fascinating scientific result, but not doesn't tell you a ton about how useful it will be to. actually solve meaningful problems. All the Willow Chip stuff I don't want to hate on because it's really cool for a bunch of other reasons. Like Lambda, LAMD greater than two is actually a really cool story there in my mind. But it gives you this really big number. It is like a genuine, interesting scientific result. It's just not quite as close to applicability as we like to talk at IonQ.

Can I just read it? I don't know if this is correct. So please correct me if I've missed found this. IonQ expects their 64 tempo system to achieve quantum advantage for certain use cases. and they also anticipate reaching a 5-9.9.99% success in logical two-cubit gate fidelity by the end of this year. Are both of those correct? And if they are, what does that mean for solving these real-world challenges which... Yeah, it's a good question. So the logical qubit thing is, yes, the thing that we're... So as a note, this will be on demonstration systems. This is not...

like commercially available quite yet. That'll happen in 26 sometime. And sort of the capability of commercial advantage, it's sort of a weird thing because commercial advantage, it's like a thing we spent a ton of time thinking about, right? Like a lot of how we do development in our ND at I-NQ is what we kind of call application driven, our application kind of first. And what we mean by that is we are trying to, by working with customers as well as within our own applications research team, which kind of splits their time between customer engagements and internal research. We're trying to find of these promising algorithms and approaches that we think actually can provide commercial advantage in like a NISC regime, in an early small quantum computer regime, because we all know

what we can do with, you know, a few thousand high quality logical qubits. And logical qubits is actually a word worth unpacking too. I'll come back to that. But what can we do with like two or three hundred or even, you know, a hundred? So Tempo is supposed to have 100 qubits, 64, AQ, maybe better, but like, you know, we're trying to set expectations appropriately. And we think there is some stuff. There's definitely stuff in that run box that is not simulatable classically. What that actually means in terms of commercial advantage we're still working on, right? Because it's a much harder problem because like you can do these theoretical calculations.

And at this size, it's hard to like prove a theoretical like definite speed up. But we think there are practical things that will be very close. but the problem is you actually need to look at price, power, performance, time to solution, all these different things, and then find and not just have your sort of compute layer, but also then your application, or sorry, your algorithm layer, and also your application layer as it fits into a real workflow for a real customer. And so we're pretty positive that we can show some benefits from like quantum over classical and at least one of these things in the temporary regime.

But if it's going to be enough to actually drive like a production service, that you can actually kind of trade dollar for dollar for, we're not sure yet. We continue to turn that crank to get closer and closer to that answer. But it's a really hard problem that we may not actually know until we have the system built and we start trying stuff. But that's kind of why we have this whole other piece of like broad access, open access, because we also want to bet on the field. We don't want to just our, you know, 18 really smart applications researchers to be helping

or trying to solve this problem. We want systems in people's hands. So a bunch of people can try and solve that problem. So that's where the SDKs come in and the software that you guys have and you let people jump in and play around and share what they've learned and hopefully build a little bit of a think tank. I think about this a lot with quantum because like we all know like, holy smokes, like the potential is there for quantum systems. Because quantum powers nature. Nature does amazing things, right, that we don't understand even how it works. And for us to harness a little bit of nature's magic and point it to big problems that traditional compute can't solve, I think is super powerful.

but I think here, I want to throw this out, and the question's kind of forming in my head as I'm talking, so bear with me. With how we're tracking who's doing better at quantum, investment powers all of this stuff, right? You have to have money to be able to experiment, to be able to test, to be able to build. So you have to attract investment. Attracting investment requires speaking a common language. The common language is how many qubits do you have, right? And now it's starting to change a little bit. Like, well, are these good, are these usefuls? How do you balance doing really innovative stuff with communicating in a language where the financial side might be like, oh yeah, I'm going to take a bet? Like, because I think that's where the stalemate kind of comes, right? It does. Yeah, it's really hard, right? We are one of only three pure play public quantum companies. And that creates a lot of interesting benefits for us as a business, but also like a lot of challenges. And this like sort of communicating the street one is a big part of it. The thing that I think we found to your point is it's just so much education, right? We have to in like AQ, it was sort of one of our earliest examples of trying to do this, but

there's others too of basically laying out the roadmap and saying, here are the things we, here are the hard problems we need to solve to get to the promised land and then updating on how we're doing on those. Because, yeah, to your point, you can, you can describe kind of the vision for 20, 30, 2035, but then like what's going to happen next year is a much harder question. And so being able to like efficiently communicate, okay, what we need to do is we need to advance networking so that we know how to connect these systems together to scale. We need an advanced cubic count. We need an advanced fidelity, but we need to do both at the same time, right?

We're not just trying to make a really big system. We're also not just trying to make a very small, perfect system because you actually do need both, right? And that's sort of this like volumetric like run box thing. short version it's really hard or yeah it's education mostly we're gonna get to the hotburns in a second don't worry just want to ask on the investment side though

and I think maybe quantum is becoming like blockchain a few years ago and AI now where the the hype the conversation the media exposure has created a point where investment VCs are just throwing money into quantum is that really happening Is the investment 10x or 100 X, a thousand X up on two years ago? So that's an interesting question.

Yeah, the private investment continues to increase. The big thing we've actually seen recently is that we have reached some sort of tipping point from if to when where the public money is actually increasing pretty rapidly too. So this is basically nation states, including the United States, although some recent science funding stuff notwithstanding, including nation states who basically want to make sure that they don't miss this wave. And so are starting to invest at different levels, whether that's sort of trying to build a, what the world economic forum calls like a quantum economy. And you see this even in the states in places like Chicago, in places like Maryland,

where they're trying to build a nexus of people of work of research at different layers. So down the supply chain, people who make lasers, make components, up to people who are trying to make applications, as well as sort of different versions of this where it's just investing in more basic science to support these broader things. So that's actually, I think, where the big boom has come. But also, yes, private investment has, I don't have the numbers offhand, but definitely continued to grow significantly, right? When I started at INQ in 2017, a kind of reasonable check size for a even a quantum hardware

company, right? Which is like very capital intensive was tens of millions of dollars. And now they're, now you're looking at like a hundred million for Q control. They just raise a $100 million around, which like they don't make hardware. And actually they might make some sensing stuff now. But to be clear, and that's not to denigate Q control because they think they're doing a bunch of really incredible stuff. But like the numbers have gone up. I have to draw attention to the dog. Who's dogs? Who's got a dog? What's a dog is it? She's a golden doble. So she's a between a golden and a poodle. Wow. Okay.

I'm picturing something in my head, which may or may not be. Yeah. If you'd like me to, I can pause for one second and get her to calm down. I think there's just someone at the door. Oh, yeah, no worries. Yeah, that's fine. We're good. Hopefully we can see the golden poodle, the golden doodle. Hey, there. Okay. Yes. That is the first dog on thinking on paper. What's her name? That's Daisy.

Daisy, welcome to thinking on paper. same as my niece we just need to get Daisy some opposable thumb so she could write we've got to do that to think on paper poor prints on paper yeah there you go there you go I think it's hot buttons time Mark

what do you think that's a perfect break for some hot buttons and then we get back into the to the quantum yes let's do it so just to explain Colman it's five questions 30 seconds one word answers it's the Daniel Carneman thinking fast of the thinking fast

thinking slow. Are you ready? I'm ready. Timers on. What's more likely? AGI or alien contact? AGI. Jazz or classical?

Jazz. Willow or Majerana. Willow. I'll explain why later. Feynman or Hawking? Feynman. Hard one, though. Sport or beer?

Sport. But again, like both. Well, and beer, boom, boom, boom. That's it. You absolutely aced it, hot buttons. Well done. Well done. That was awesome.

That was awesome. Well, I guess right now, we could fold straight into the news. And we can frame that with Coleman's answer to the hot button, Willow or Marjorana. He said, Willow, which maybe we are surprised about. Jeremy, here's with the news. It actually is great that we've tied together news and hot buttons in such a way. That's why I wanted to go into it quickly. So Microsoft's in the news.

We've all seen it. We've all heard about it. They've got something called margarine. No, I'm just kidding. It's Margarana, right? So they've claimed to do something different in quantum. There's a path to a million cubits. There's a lot of people on both sides of their enthusiasm for their announcement.

Coleman, can you break this down for people that don't really know much about quantum, what this could mean possibly? Yeah. So I think there's kind of two, the short version. as to why I said willows because I think that path is shorter and more concrete, basically, to the, like, the promised land, if you will. So, yeah, Magironic Cubits or topological cubits are this thing that's, like, been around in kind of people's heads for, I think, like, 20 years. And Microsoft has been kind of working on it that whole time. It's sort of been in the background,

you know, we, Microsoft, like, well, I don't think we get to say this anymore. But for a long time, we were, you know, we always are aware of Microsoft. They have a great science team. They have a great platform with Azure Quantum, but they didn't have any qubits. Now they might have one, which is more than zero. And I say might because sort of the interesting, so topological cubits are this slightly different approach to making a cubit that still sort of follow the rules that we talked about earlier.

And if you can actually do them, if you can actually create them and then get them to be able to be interacted with efficiently, they have a bunch of sort of natural benefits because you're using these sort of special what are called non-abellion anions, these things that are neither fermions nor boson. So they're a different kind of quantum particle that the way you have them actually do your gates, your computation is via this thing called braiding. And braiding has this really, this is like the topology thing. Braiding has this really cool quality of being hard to unbrate, harder than sort of a lot of the other ways you would do a gate. And so this is literally like if you think about taking a, a, uh, uh,

well, literally making a braid, right, or tying a knot in a rope, where you have a circle and you put something through it, this is topologically bound. I can't without cutting this, which is a harder thing than just, you know, taking them apart. For the podcast listeners, I'm making little rings with my fingers and then pulling them apart like I'm a magician. That's much harder to do. So there's this really cool quality of it sort of naturally being protected from certain types of noise that really impact other types of quantum computers. The hard part is that to date, we're only pretty sure you can actually create these things in nature. So you can do a simulation. They've actually done a simulation on, I'm not sure if it was Willow or Sycamore.

You can do a simulation to prove that like the state of matter does exist theoretically, but then actually getting it and making it well controlled is very, very hard. And so the Microsoft announcement is basically, hey, we did it. We made a quibut. There is some, I think rightly, a lot of kind of back and forth in the community that is like, kind of above my head in terms of these are like the capital P physicists around there. Basically it's it's hard to prove that you actually did it. And so there's some conversation about the there's because there's a paper in nature, right, and that has some data. And then there's also this

other sort of update that they did in a in a smaller session in this group called Station Q that has some additional data. And so it's this back and forth of did they actually, have they actually sort of done the thing and made a topological key, but even just a single one, which is, like an incredible step forward. And it's obvious that they're way, way closer than they've ever been, but it's like that's your zero to one problem. And then your one end problem is the next one that is again, why I sort of at least, you know, today knowing what I know, even though topological over superconducting, I think has some clear benefits. Actually being able to build that system is pretty hard, right? It took INQ. We're able to double our cubic count roughly every 18 to 24 months.

So if they're at one right now, also topological. Hubets kind of require error correction to be effective, much like superconducting. That's not entirely true. That's a little bit of a gloss on a comp kid statement, but broadly speaking. And so it's really exciting. And if they've done it, you know, goddamn. But I think the jury's still out a little bit. I'm sure they'll get there eventually. And then even when they do, there's still a long road between that and a useful quantum computer. So like, without like, sort of stemping on the excitement at all because it's really exciting. It's just worth noting that it's still pretty far away from actually having used.

useful quantum computer for a couple of different reasons. There's a couple quick things that just, I'm just going to give you full access to what's going on in my head right now, Coleman. So he's got a bell ready. So I think I see this cartoon, right, with a bunch of lion tamers, picture of lion tamer.

And each lion tamer has the Microsoft logo, the IONQ logo, the, you know, pick your quantum. And they're in the corner trying to tame. this cubit and they're trying to wrangle it and like whoever can wrangle it the quickest. I don't know. I'm just sure. There's not a question in there. It was it.

Daisy loves it. Daisy loves my analogy. But that's what I'm thinking because these things are hard to wrangle and point to point to use. And it's not like something's going to happen in three weeks in the garage somewhere in Palo Alto. This is going to be a bit of a line of kind of working through some of the stuff, which is the exciting part to me, I think.

Yeah, I agree. There is actually one garage in Palo Alto. that I'm keeping my eye on, which is cyquantum. But they've been working on it for about a decade, to be clear. It's not. What can you tell me about those guys? Quite a bit more than a garage at this point. So they're doing Photonic quantum computing, which is really interesting because one of the

hard scaling problems is networking systems together and they sort of get it for free. The hard part is that it's a very big bang sort of approach. You're going to need basically to come online with everything you expect to have, all the the cupits, everything. And also there's like a ton of interesting, hard. sort of environmental problems with them. Like their qubits don't need to be cold, but their sensors do. And so as a result, you end up with this huge, like, rack of cryocooled sensors.

But again, that's another one, much like topological with, like, if they can pull it off. And I think it's still an if at this point. Like, I would describe superconducting in ions and neutrals as sort of a when. If they can pull it off, it's going to be amazing. It's going to be really cool. I am, like, officially because I work for Ion Q, I am, you know, an ions fan. And that's true. but I'm like, I want the industry to succeed.

So I have a lot of sort of opinions and love for all the different qubit modalities. Because I also think like to some extent, A, we're all still trying to like to use the classical analogy. We're still trying to build vacuum tubes. Like there's going to be a thing after this. And it's also likely that in the medium term, it's going to be multimodal, right? This is some of the work we're doing with the Air Force Reacher's Lab actually is creating the networking infrastructure to allow for multimodal quantum computing using ions and neutrals

or ions and superconducting, although that one's a hard one. But so, yeah, I'm excited about it all as well. And yeah, there's a lot of qubits out there. It's a callback to Book Club, Quantum Supremacy. Anyone thinking on paper.x, Y, Z, a few months ago, Jeremy and I did a chapter by chapter breakdown of quantum supremacy, and one of the chapters was on the different modalities of that. So we cover some of those.

Head to the website to get that. I love that your passion for the space and you want the space to succeed, you want the technology to succeed, you want quantum to succeed. We don't always see that. I think that's quite clear even just with the recent announcements. Not everybody is not a team sport all the time. But could we take it back to the integration, the classical hybrid integration? When we had IBM, they were speaking about having 100,000 hobbyist coders on there.

I don't know how, on their Kisket platform. Yeah, on Kisket, yeah. Is IANQ doing something different if we have some developers watching this who work in Rust or something else? They want to get involved in quantum algorithms. Yeah, can you speak us through that? Sure. Well, first of all, we support Kisket.

So some of the 600,000 are running on INQ. Okay. And there's other hardware partners that Kiskit uses. Kisket is open source software. We actually, our approaches were sort of strategically promiscuous is the phrase I use And I'll just, I'll use it on the podcast, too, with a lot of these SDKs because we're just not sure what's going to win. And we don't honestly care. We just want to get quantum in people's hands.

And so that's important to us. So we support, you know, Kisket, Cirque, which is the SDK that Google makes and about half a dozen others of smaller and larger user bases. And the access layer, like you're saying, is really important. I think getting started is a really kind of challenging thing right now because every kind of intro to quantum computing or algorithm design development. development kind of textbook or like resource kind of is like, okay, quantum's going to solve these amazing problems. How much linear algebra do you know? That's step one, step two. And right, same. Well, I've got a little bit now, but I didn't, you know, even just a few years ago. And so we are still, and I think this is true of most of the SDKs out there, but you're still missing sort of these

abstraction layers between like raw writing quantum circuits. So gates to be run like physically on systems and actually sort of solving useful commercial problems or even just useful scientific problems. And building those abstraction layers is pretty hard, right, because you mentioned Rust. Like Rust is a really interesting programming language for a bunch of reasons. But one is because like most good programming language is it's trying to solve a specific problem or a set of specific problems, right, which is like memory safety, all these different things as opposed to see. And I don't know that we have enough of the algorithmics or the primitive building blocks to be able to build those abstractions in a useful way right now. So you can't, like, I don't know what the

quantum programming language sort of in the middle is, right? We have these great libraries that people are studying and writing and including the INQ team about, okay, how do we kind of package all this, this all the way into something that solves optimization problems or solves chemistry problems or whatever, but kind of being able to actually operate at that building blocks level and have an intuition about it, I think is just something that no one is doing very well at this point. And And we are trying to find different ways to engage with that community and find these things and build these abstractions, these kind of composable pieces. Our cloud platform now has this set of things called quantum functions and quantum solvers,

which are trying to sort of kind of allow to build these building blocks. But it's a work in progress, I think, is it requires us to have a better picture of what we're trying to build. So that's kind of the access and SDK question. The hybrid and like integration question is maybe a longer one than we have time. for because you have to break it down quite a bit. Because broadly speaking, what you're talking about is, okay, we want to use classical computers to make quantum computers better, or we want to use quantum computers to make just classical or just any sort of computational workflow better. They're both, they're slightly different. Both are sometimes called hybrid quantum computing, and they're also

a virtuous cycle. So they're like, the edges are blurry because you are adding capabilities at the low level, error mitigation, better control, which then allows you to improve your algorithm design to do these like quantum accelerated applications and back and forth like this. So breaking it down into useful pieces of conversation to talk about like how do we integrate with HPC to do quantum accelerated workloads versus how do we expand the instruction set to do these very low level fast feedback like during a quantum circuit while your quantum information is still coherent things like mid-circuit measurement and dynamic circuit execution, which you actually need to do air correction to do all these other really important things as we scale.

There's a bunch of different interesting pieces in that if you want to pick one, or we can just sort of leave it at a gloss, your call. I want to partner with a caricature artist, because now I'm seeing there's a guy, there's a guy doing calculations, and there's a guy spinning a wand holding two cubits together. And the guy in the calculator, the guy with a cubit's like, yo, hurry up and finish, hurry up and finish. They're starting to decouple, right?

There's an urgency to this that is like mind-blowing, and how do you... Yeah. Colman, can we, let's save that. And if publicly invite you back for part two, where we perhaps start there and go from that point because I don't know about you, Jeremy,

but the last four minutes have been incredible. I think it's one of the best conversations we've been part of on quantum, on thinking on paper. So thank you for that. You've mentioned the promised land twice. What is the quantum promised land for you for I and you? Sure. Yeah, it is commercial utility.

It's having someone choose to use a quantum computer because it is technically economically and just sort of practically the right choice, as opposed to right now where most people are interested in quantum computing because it's an interesting technology to explore, which I very much agree with. Like in a sense, and this is like, I don't know if this is an official I-N-Q messaging statement,

but it's a Coleman-Collins one. It's making quantum computing boring, right? It's, I just want it to be part of the toolbox of things I can use to solve problems. That sort of, to me, is the promise line. And in practice, that means being able to solve interesting commercially and scientifically useful problems at meaningful scale. So humans, humans, there is still hope.

Mark and I unpacked. unpacked, Pachy McCormack's, the most human wins paper, talking about how humans survive in the realm of AI and the realm of emerging technologies. Humans, we still have to figure out what questions, what those problems are to find those problems that we can point all these new technologies

like quantum computers to. So I feel that, Coleman, you've most succinctly explained a lot of things related to quantum. That's a call out for our listeners, anyone who's listening, if you have any ideas of what they are, put them in the comments wherever you're watching or listening to this.

I love the Coleman Collinsisms. Can we call the show IONQ is strategically promiscuous? Is that we use that as a title for the podcast? We won't. I'd be fine with it, but I think our comms director might have something to say. Maybe a little heartburn in there. All right, let's get to carryover question from last week.

This is a simple one, well, simple relative to someone who's been. to where that question came from, Jeremy. So we talked about space data centers with a gentleman named Philip yesterday. They're putting data centers on rockets and they're doing compute in lower Earth orbit, which apparently isn't as crowded as everyone thinks it is. So everything's going to be fine up there. We talked about space elevators, all kinds of. But Philip is asking Coleman, when will quantum break up Bitcoin?

So this is such a, I don't want to... Harsh on Phillips vibe, but this is such a common question that we have a shorthand for it, which is Q-Day, which is effectively, the other way to describe this is, when will we have a cryptographically relevant quantum computer? So one of the things we have known for a very long time, in fact, one of the things that is that spurred the initial sort of push into quantum computing technology development in the 90s is this algorithm called Shores algorithm. And Shores algorithm is actually a period-finding algorithm. It lets you see how far apart things. are because you're using this like basically a quantum 4A transform to do some special tricks.

The cool thing about that is you can map that problem to finding the constituent co-primes of a very large number that is made up of two prime numbers. Well, why does that matter? That is the basis for most cryptographic algorithms that are used for, you know, securing Bitcoin wallets as well as regular wallets and banking and telephony and all these different things. And so that's a really, really important question that a lot of people have to be able to have on their mind quite a lot. In the specific case of Bitcoin, the likely attack vector is going to be actually breaking the public keys and then using those to access wallets rather than breaking the

chain itself. And that's just because where is the easiest thing that you can sort of find? That I'm pretty sure I forget exactly. I don't think Bitcoin is RSA, so there are some of these cryptographic algorithms. There's RSA, ECC, AES are the three big ones. I'm pretty sure Bitcoin is either AES or but I forget. In any case, the answer is kind of the same for all of them plus or minus a factor of maybe two, which is we're going to need order. I think it's like, so it's like double the number of bits in the bit string you're trying to solve in logical cubits. Like not just sort of like logical cubits is a bit of a funny word because it really just means it's good enough to solve the problem you need to solve, right? So you can have a logical cubit that isn't like 10-nines or

12-9s of perfection, depending on your approach. But ions can definitely do this. But for this, we are talking about you need like 10, 12, 19, like basically a perfect qubit, which requires error correction, which is a thing that we are just now as industry starting to actually do cool demonstrations on, like, in the past 18 months. And so broadly speaking, kind of the most, like, air corrected or logical cubits anyone has. At this, like, level of quality is zero.

At any level of quality is like maybe 10. and maybe a dozen. And we're going to need them to be way better, and we're going to need like 4,000 of them. So it's not tomorrow. It's probably not the day after that either. In terms of actually predicting a timeline, I'm always really wary to do that

because one of my jokes is that, like, the only thing we know about the future is that it's not going to be like the present and it's not going to be what we expect. And so what about if I make this a little lighter lift? So we had the quantum, the post-quantum cryptography expert from IBM on a few shows ago, or a while ago, actually. And she predicted, what, 2032? I think that was in line with IBM's Bluebird chip.

And I think, but with Bitcoin, because Satoshi's, the original Bitcoin, the early Bitcoin, aren't under the same cryptography as the later. Something changed, didn't it? Okay. We don't want to put a day on that. I just want to move much, much further into the future then. Is a quantum currency possible? I was listening to a podcast of a big idea. It was bankless, shout out to bankless.

And I was speaking about this idea a long way in the future. Quantum money. Is that something you've ever thought about? Yeah, so there's a, oh man, that's a great question. So there's a paper, a sort of a proposal from the nines. And I want to say it's, what's his name? at IBM Charlie do some live

Googling one sec Charlie Bennett Live perplexiting nice well because I just need a data point not an answer that like is described to me I use all those tools

quite a lot yeah so yeah it's like basically a very early idea in quantum information theory which is yeah you could theoretically sort of use

entangled pairs to create some sort of like proof of work proof of ownership doing that in a way that is like again back to our like the Finchenses criteria, like, is long-term coherent, is, like, practical.

I wouldn't, I wouldn't bet on it for this century, but I don't think it's impossible. Far from it. My money is already quantum. Coleman, it's entangled with bills and invoices. There's quantum entanglement there. There's quite a lot of entanglement there, yeah. That's right. My money is there, and then it's not as well.

Oh, man. What a great chat, Coleman. This has been a blast. Can you leave us as a final takeaway, a, carryover question for our next guest, who I believe, Mark, do we have the world folks on? I know that's in two weeks. Next week, we have Diego on AI and blockchain. Got it. Got it. Carry over question. Can be about anything, Coleman, whatever's on your mind that you'd like to send down the line. Okay. So, okay, for someone that is focused on AI and blockchain, maybe an interesting

question would be where do our current approaches break? Can we just keep scaling forever or are we going to have to do something different at some point? Very interesting. And where are those break points and like what's going for what like what dominoes fallen first? Really interesting man. I love that. Yep. Great question.

Great interview. Thank you very much. Coleman Collins, round of applause from the thinking on paper audience there. Yeah. Thanks for having me. Sorry about the dog. She's sweet, but she's loud. Personality, character.

First dog on... First dog on thinking on paper. Coleman, come back and join us again, and let's continue to unpack the future together. Appreciate it. Sounds great. I'd love to. All right.

Take it easy. Bye. Right. All right. Mark Fielding, we are backstage. How's your brain feeling? Scrambled eggs or what? Like, what do you talk to me?

Yes, yes, we are. Jamie Gilmston. So, my... There was something he said. My brain kind of, there was a moment after about three minutes of Coleman speaking. And he said, I think it was on the five steps that you need to have a qubit. And he said something about separating the process from the universe. He said separated from the universe.

And my brain just went, what? And I'm going to go back and listen to exactly in what context he said that. But yeah, it was very informative. The first 15 minutes was very technical and I enjoyed that. I have to go back and listen again to try and understand some of it, but I thought it was a great show. Yeah, separating something. Like there's an edit tool, right, in the quantum algorithm process where it's like,

okay, now separate from universe. Separate from the universe. I mean, what sentence that is. I guess that comes from like wanting to keep noise out of the equation and like wanting to kind of because like back to my analogy of the lion tamers. in the cubit in the corner. They're trying to kind of keep all the noise away and keep it from decohering,

which I think is interesting. I always think about, there's a great analogy in Michi Okaku's book that we unpacked in Book Club, Quantum Supremacy. He talked about the idea of like a radio tuning in, right? And coherence is when it actually lands on the station.

And, you know, decoherence is kind of the in-between stations. And, you know, the in-between... If you think about that, Like the in-between states as you roll through, and friends, like if you go back to where you actually had a radio with a dial, right, that you could literally turn the dial going from. Oh, my gosh. But like, as you run through that, that's basically communication attempts, right, attempting to cohere in between states that we can't actually lock into. So I roll back on that.

It's a really good analogy. You can really see it. and then you extend the frequency out to infinity on each end. It's not just like 92 to 104 on a car radio. You know, it goes forever and you have to fine tune it to that one bit. I thought you did a really good job on the algorithmic qubits versus the physical qubits. Jeremy, yeah. It's obviously all the shows are that something's stuck in your head, well done.

Something's clicking. Yeah, I like the idea of this being like them kind of kind of trying to get the natural process of all of this together because nature beats us in quantum every day of the week, right? And then, you know, by taking these ions that are natural, right, and allowing them to be what they are and trying to work within that system, maybe we can enable the emergent qualities of nature that we all find so fascinating and wonderful, but we don't know how they work. Well, yeah, we've had quite a few conversations about that, haven't we, at thinking on paper. xyz you can go there and so through flows i know that every every week now something rings a bell

from a previous thinking on paper episode and there was two today that he mentioned the first was from yesterday so at the beginning mentioned that solar rays were a problem for quantum essentially they cause errors don't end discaherent and then but yesterday with the data centers in space solar arrays weren't a problem. So solar rays weren't a problem for the big, but there were a problem for the very, very, very small. So that was a nice throwback to that. And the other one was he was speaking about making quantum boring.

And it reminded me of Jonathan, can't remember his surname, we were speaking about NFTs and blockchain and making NFTs boring. And this way of like making the technology that's really hyped, kind of boring, but they don't mean making it boring. They make, they mean making it kind of like, just part of the landscape,

accessible, too. Accessible, easy to use. So the way I think about it, back to Packing McCormack's paper, think about moving to higher levels of abstraction, right? So when these things become easier and more commonplace, they're just Lego blocks for us to piece things together at higher levels of abstraction.

Yeah. I think that works a lot more easily with NFTs than it does with, with quantum computers, because honestly, it was another reminder of just, it is just ridiculously complex. And I wonder maybe some of us just shouldn't really keep going, because we're never going to really go ahead around it,

because it's just so wonderfully complicated. And, yeah, I was reminded of that again today. I love to be confused. It was like, oh, my gosh, I know nothing. That's the beauty and quantum systems in nature. Like, we don't understand it, but they're there and they're beautiful. And it generates wonder in our head. We talked about wonder a couple of episodes ago as well,

and that we're in a deficit of wonder. activate your wonder by listening to Thinking on Paper Thinking on Paper. We read books together We have book clubs We've gone through seven or eight different books Stay tuned for the next book

That's coming up We've got two chapters left of Mr. Harari's book that has been Been Dark And most depressing book on artificial intelligence I hope I will ever read And just because I say it's depressing Doesn't mean it's not good

Because it's very, very good It's very important Amazing And some very, very, very, very vital or things for humanity to think about. We've got two chapters left. Tell us about the guests we have coming up, Mark. We've got some killer guests over the next couple of weeks.

Next week, we're speaking about the intersection of AI and blockchain, how blockchain can perhaps secure AI or perhaps AI can just use blockchain to take over the world. We'll find out after that we've got world ID, the orb. That's going to be amazing. Come scan your retinas. A.J. Patel, I think he's, I can't remember his job job. Ajay. Ajay. Patel is going to be breaking down world ID.

Kevin Kelly is thinking on paper to XYZ. You can win tickets for the screening of that. So, yeah, a lot of good things to come. Be curious. Stay disruptive. Keep thinking on paper.