Ideas: Quantum computing redefined with Chetan Nayak

at Microsoft Quantum. Under Chetan’s  leadership, the Microsoft Quantum team   has published a paper that demonstrates  a fundamental operation for a scalable   topological quantum computer. The team also  announced the creation of the world's first   topoconductor—more on that later—and first QPU  architecture with a topological core, called the   Majorana 1. Chetan Nayak, I can’t wait to find  out what all of this is ... welcome to Ideas!

been. You know, just, sort of, you  know, I maybe went there, you know,   looking to pick up the next Lord of the  Rings book, and while I was there, you know,  

wander into a book that was sort of explaining  the theory of relativity to non-scientists. And I   remember leafing through those books and actually  reading about Einstein's discoveries, you know,   most famously E = mc2, but actually a lot of those  books were explaining these thought experiments   that Einstein did where he was thinking about, you  know, if he were on a train that were traveling at  

the speed of light, what would light look like to  him? [LAUGHTER] Would he catch up to it? You know,   and all these incredible thought experiments  that he did to try to figure out, you know,   to really play around with the basic laws as they  were currently understood, of physics, and by,   you know, stretching and pulling them and going  into extreme … taking them to extreme situations,  

you could either find the flaws in them or in some  cases see what the next steps were. And that was,   you know, really inspirational to me.  I, you know, around the same time,   also started leafing through various advanced  math books and a little later picked up a book   on calculus and started flipping through it, used  book with, like, you know, the cover falling apart   and the pages starting to fall out. But there was  a lot of, you know, accidental discovery of topics  

through wandering through bookstores, actually.  I also, you know, went to this great magnet high   school in New York City called Stuyvesant High  School, where I was surrounded by people who   were really interested in science and math and  technology. So I think, you know, for me, that   origin story really starts, you know, maybe even  earlier, but at least in my preteen years when,   you know, I went through a process of learning  new things and trying to understand them in my  

own way. And the more you do that, eventually  you find maybe you're understanding things in   a little different way than anybody else ever  did. And then pretty soon, you know, you're   discovering things that no one's ever discovered  before. So that's, sort of, how it started.

love how you say, I was accidentally exposed.  [LAUGHTER] Sounds kind of like radiation of   some kind and you've turned into a scientist. A  little bit more on that. This idea of quantum,   because you've mentioned Albert Einstein,  there's quantum physics, quantum mechanics,   now quantum computing. Do these all go together?  I mean, what came out of what in that initial,   sort of, exploration with you? Where did you start  getting interested in the quantum of things?

head around because it is so counterintuitive.  Quantum mechanics in some sense, or quantum theory   in some sense, for most of what we experience  in the world is down many abstraction layers   away from what we experience. What I find  amazing is that the people who created,   you know, discovered quantum mechanics, they had  nothing but the equations to guide them. You know,  

they didn't really understand what they  were doing. They knew that there were some   holes or gaps in the fundamental theory, and  they kind of stumbled into these equations,   and they gave the right answers, and they just had  to follow it. I was actually just a few weeks ago,   I was in Arosa, which is a small  Swiss town in the Alps. That's   actually the town where Schrödinger  discovered Schrödinger’s equation.

x-rays. And he said, you know, there has to  be a wave equation for this thing and this   wave equation needs to somehow correctly  predict the energy levels in hydrogen.

that he was onto something. And then remarkably,  other people were able to build on what he'd done,   were able to say, no, there must be a grain of  truth here, if not the whole story, and let's   build on this, and let's make something that is  richer and encompasses more and try to understand   the connections between this and other things. And  Heisenberg was, around the same time, developing   his what's called matrix mechanics, a different  way of thinking about quantum computing, and  

then people realize the connections between those,  like Dirac. So it's a remarkable story how people,   how scientists, took these things they understood,  you know, imposed on it a certain level of   mathematical consistency and a need for the math  to predict things that you could observe, and once   you had, sort of, the internal mathematical  consistency and it was correctly explaining   a couple of data points about the world, you  could build this huge edifice based on that. And  

so that was really impressive to me as I learned  that. And that's 100 years ago! It was 1925.

positrons. So there was a rich, you know,  decades of development and understanding,   and then eventually it got combined with  relativity, at least to some extent. So   there was a lot to do there to really understand  and build upon the early discoveries of quantum  

mechanics. One of those directions, which  was kicked off by Feynman around, I think,   1982 and independently by a Russian mathematician  named Yuri Manin was, OK, great, you know, today's   computers, again, is many abstraction layers away  from anything quantum mechanical, and in fact,   it's sort of separated from the quantum world  by many classical abstraction layers. But what  

if we built a technology that didn't do that?  Like, that's a choice. It was a choice. It was   a choice that was partially forced on us just  because of the scale of the things we could   build. But as computers get smaller and smaller  and the way Moore's law is heading, you know,   at some point, you're going to get very close  to that point at which you cannot abstract away   quantum mechanics, [LAUGHTER] where you must  deal with quantum mechanics, and it's part  

and parcel of everything. You are not in the  fortunate case where, out of quantum theory   has emerged the classical world that behaves the  way we expect it to intuitively. And, you know,   once we go past that, that potentially is really  catastrophic and scary because, you know, you're   trying to make things smaller for the sake of, you  know, Moore's law and for making computers faster  

and potentially more energy efficient. But, you  know, if you get down to this place where the   momentum and position of things, of the electrons,  you know, or of the currents that you're relying   on for computation, if they're not simultaneously  well-defined, how are you going to compute with   that? It looks like this is all going to break  down. And so it looks like a real crisis. But,   you know, what they realized and what Feynman  realized was actually it's an opportunity. It's  

actually not just a crisis. Because if you do it  the right way, then actually it gives you way more   computational power than you would otherwise have.  And so rather than looking at it as a crisis,   it's an opportunity. And it's an opportunity to do  something that would be otherwise unimaginable.

Are there any particular, sort of, modern-day  21st-century or 20th-century people that have   influenced you in such a way that it's  like, I really want to go deep here?

maybe Feynman's the third wave, I guess? I'm not  sure if he's second or third wave, but anyway,   he's post-war and was really instrumental in the  founding of quantum computing as a field. He had   a famous statement, which is, you know, in his  lectures, “There's always room at the bottom.”   And, you know, what he was thinking about there  was, you can go to these extreme conditions,   like very low temperatures and in some cases very  high magnetic fields, and new phenomena emerge  

when you go there, phenomena that you wouldn't  otherwise observe. And in a lot of ways,   many of the early quantum theorists, to some  extent, were extreme reductionists because,   you know, they were really trying to understand  smaller and smaller things and things that in   some ways are more and more basic. At the same  time, you know, some of them, if not all of them,   at the same time held in their mind the idea  that, you know, actually, more complex behaviors  

emerge out of simple constituents. Einstein  famously, in his miracle year of 1905, one of the   things he did was he discovered ... he proposed  the theory of Brownian motion, which is an   emergent behavior that relies on underlying atomic  theory, but it is several layers of abstraction   away from the underlying atoms and molecules and  it's a macroscopic thing. So Schrödinger famously,   among the other things, he's the person who  came up with the concept of entanglement …

underlying theory of electrons and photons and  how they interact. But he also thought a lot   about liquid helium and ultimately about  quantum computing. Motivation for him in   quantum computing was, you have these complex  systems with many underlying constituents   and it's really hard to solve the equation.  The equations are basically unsolvable.

is you actually can't even put it on a computer  and solve a problem like that. And in fact,   it's not just that with Avogadro's number  you can't; you can't put it on a computer   and solve it with a thousand, you know,  [LAUGHTER] atoms, right? And actually,  

you aren't even going to be able to do  it with a hundred, right. And when I say   you can't do that on a computer, it's not  that, well, datacenters are getting bigger,   and we're going to have gigawatt datacenters, and  then that's the point at which we'll be able to   see—no, the fact is the amazing thing about  quantum theory is if, you know, you go from,   let's say, you're trying to solve a problem with  1,000 atoms in it. You know, if you go to 1,001,  

you're doubling the size of the problem. As far  as if you were to store it on a cloud, just to   store the problem on the classical computer, just  to store the answer, I should say, on a classical  

computer, you'd have to double the size. So  there's no chance of getting to 100, even if, you   know, with all the buildout of datacenters that's  happening at this amazing pace, which is fantastic   and is driving all these amazing advances in  AI, that buildout is never going to lead to a   classical computer that can even store the answer  to a difficult quantum mechanical problem.

we are today. But before we do that, you've also  alluded to something that's super interesting to   me, which is in light of all the recent advances  and claims in AI, especially generative AI,   that are making claims like we'll be able to  shorten the timeline on scientific discovery   and things like that, why then, do we need  quantum computing? Why do we need it?

and there was a lot of data on it, right. And so  it was clearly a big data problem that's basically   classical. As far as we know, there's no classical  way to simulate or mimic quantum systems at scale,  

that there's a clean separation between the  classical and quantum worlds. And so, you know,   that the quantum theory is the fundamental theory  of the world, and there is no hidden classical   model that is lurking [LAUGHTER] in the background  behind it, and people sometimes would call these   things like hidden variable theories, you know,  which Einstein actually really was hoping, late in   his life, that there was. That there was, hiding  behind quantum mechanics, some hidden classical  

theory that was just obscured from our view. We  didn't know enough about it, and the quantum thing   was just our best approximation. If that's true,  then, yeah, maybe an AI can actually discover that   classical theory that's hiding behind the quantum  world and therefore would be able to discover it  

and answer the problems we need to answer. But  that's almost certainly not the case. You know,   there's just so much experimental evidence about  the correctness of quantum mechanics and quantum   theory and many experiments that really, kind  of, rule out many aspects of such a classical   theory that I think we're fairly confident there  isn't going to be some classical approximation  

or underlying theory hiding behind quantum  mechanics. And therefore, an AI model, which   at the end of the day is some kind of very large  matrix—you know, a neural network is some very   large classical model obeying some very classical  rules about, you take inputs and you produce  

outputs through many layers—that that's not going  to produce, you know, a quantum theory. Now,   on the other hand, if you have a quantum computer  and you can use that quantum computer to train an   AI model, then the AI model is learning—you're  teaching it quantum mechanics—and at least   within a certain realm of quantum problems,  it can interpolate what we've learned about   quantum mechanics and quantum problems to solve  new problems that, you know, you hadn't already  

solved. Actually, you know, like I said, in the  early days, I was reading these books and flipping   through these bookstores, and I'd sometimes figure  out my own ways to solve problems different from   how it was in the books. And then eventually I  ended up solving problems that hadn't been solved.   Well, that's sort of what an AI does, right? It  trains off of the internet or off of playing chess  

against itself many times. You know, it learns and  then takes that and eventually by learning its own   way to do things, you know, it learns things  that we as humans haven't discovered yet.

to talk about. And this question includes,  what's a Majorana and why is it important?

is that the most pressing thing in the world?  You know, simulating quantum systems? It's great   to understand more about physics, understand more  about materials, understand more about chemistry,   but we weren't even at that stage, I think, there  where, hey, that's the limiting thing that's   limiting progress for society. And then, secondly,  there was also this feeling that, you know,  

what you're really doing is some kind of analog  computing. You know, this doesn't feel digital,   and if it doesn't feel digital, there's this  question about error correction and how reliable   is it going to be. So Peter Shor actually, you  know, did two amazing things, one of which is a   little more famous in the general public but one  of which is probably more important technically,  

is he did these two amazing things in the  mid-’90s. He first came up with Shor's algorithm,   where he said, if you have a quantum computer,  yeah, great for simulating quantum systems, but   actually you can also factor large numbers. You  can find the prime factors of large numbers, and   the difficulty of that problem is the underlying  security feature under RSA [encryption],   and many of these public key cryptography  systems rely on certain types of problems  

that are really hard. It's easy to multiply two  large primes together and get the output, and you   can use that to encrypt data. But to decrypt  it, you need to know those two numbers, and   it's hard to find those factors. What Peter Shor  discovered is that ideally, a quantum computer,   an ideal quantum computer, would be really good  at this, OK. So that was the first discovery.  

And at that point, what seemed at the time an  academic problem of simulating quantum systems,   which seemed like in Feynman's vision, that's what  quantum computers are for, that seemingly academic   problem, all of a sudden, also, you know, it turns  out there's this very important both financially   and … economically and national security-wise  other application of a quantum computer. And   a lot of people sat up and took notice at that  point. So that's huge. But then there's a second  

thing that he, you know, discovered, which was  quantum error correction. Because everyone,   when he first discovered it, said, sure, ideally  that's how a quantum computer works. But quantum   error correction, you know, this thing sounds like  an analog system. How are you going to correct   errors? This thing will never work because it'll  never operate perfectly. Schrödinger's problem  

with the cat’s going to happen, is that you're  going to have entanglement. The thing is going to   just end up being basically classical, and you'll  lose all the supposed gains you're getting from   quantum mechanics. And quantum error correction,  that second discovery of Peter Shor’s, really,   you know, suddenly made it look like, OK, at least  in principle, this thing can happen. And people  

built on that. Peter Shor’s original quantum error  correction, I would say, it was based on a lot of   ideas from classical error correction. Because you  have the same problem with classical communication   and classical computing. Alexei Kitaev then came  up with, you know, a new set of quantum error   correction procedures, which really don't rely in  the same way on classical error correction. Or if  

they do, it's more indirect and in many ways  rely on ideas in topology and physics. And,   you know, those ideas, which lead to quantum  error correcting codes, but also ideas about   what kind of underlying physical systems would  have built-in hardware error protection, led to   what we now call topological quantum computing and  topological qubits, because it's this idea that,   you know, just like people went from the  early days of computers from vacuum tubes  

to silicon, actually, initially germanium  transistors and then silicon transistors,   that similarly that you had to have the right  underlying material in order to make qubits.

at least to some extent. Nothing's perfect, but,   you know, in a controllable way so that you can  make it better as needed and good enough that any   subsequent error correction that you might call  software-level error correction would not be so  

cumbersome and introduce so much overhead as to  make a quantum computer impractical. I would say,   you know, there were these … the field had a, I  would say, a reboot or a rebirth in the mid-1990s,   and pretty quickly those ideas, in addition  to the applications and algorithms, you know,  

coalesced around error correction and what's  called fault tolerance. And many of those ideas   came, you know, freely interchanged between ideas  in topology and the physics of what are called   topological phases and, you know, gave birth  to this, I would say, to the set of ideas on   which Microsoft's program has been based, which  is to look for the right material … create the   right material and qubits based on it so that  you can get to a quantum computer at scale.  

Because there's a number of constraints there.  And the work that we're really excited about   right now is about getting the right material  and harnessing that material for qubits.

underlies it is that the electrons form up into  pairs. These things called Cooper pairs. And those   Cooper pairs, their wave functions kind of lock up  and go in lockstep, and as a result, actually the   number of them fluctuates wildly, you know, in any  place locally. And that enables them to, you know,   to move easily and carry current. But also, a  fundamental feature, because they form pairs,  

is that there's a big difference between an even  and odd number of electrons. Because if there's an   odd electron, then actually there's some electron  that's unpaired somewhere, and there's an energy  

penalty associated, an energy cost to that. It  turns out that that's not always true. There's   actually a subclass of superconductors called  topological superconductors, or topoconductors, as   we call them, and topoconductors have this amazing  property that actually they're perfectly OK with   an odd number of electrons! In fact, when there's  an odd number of electrons, there isn't any   unpaired electron floating around. But actually,  topological superconductors, they don't have that.  

That's the remarkable thing about it. I've been  warned not to say what I'm about to say, but   I'll just go ahead [LAUGHTER] and say it anyway.  I guess that's bad way to introduce something …

subatomic particles that carry no charge. And so,  you know, he was thinking about something very,   very different from quantum computing, actually,  right. So Majorana—didn't know anything about   quantum computing, didn't know anything about  topological superconductors, maybe even didn't   know much about superconductivity at all—was  thinking about subatomic particles, but he  

wrote down this equation for neutral objects, or  some things that don't carry any charge. And so   when people started, you know, in the ’90s and  2000s looking at topological superconductors,   they realized that there are these things  called Majorana zero modes. So, as I said,   and let me explain how they enter the story,  so Majorana zero modes are … I just said that   topological superconductors, there's no place you  can find that even or odd number of electrons.  

There's no penalty. Now superconductors, they do  have a penalty—and it's called the energy gap—for   breaking a pair. Even topological superconductors.  You take a pair, a Cooper pair, you break it,   you have to pay that energy cost, OK. And it's,  like, double the energy, in a sense, of having   an unpaired electron because you've created two  unpaired electrons and you break that pair. Now,   somehow a topological superconductor has to  accommodate that unpaired electron. It turns  

out the way it accommodates it is it can absorb or  emit one of these at the ends of the wire. If you   have a topological superconductor, a topoconductor  wire, at the ends, it can absorb or emit one of   these things. And once it goes into one end,  then it's totally delocalized over the system,   and you can't find it anywhere. You can say, oh,  it got absorbed at this end, and you can look and  

there's nothing you can tell. Nothing has changed  about the other end. It's now a global property   of the whole thing that you actually need to  somehow figure out, and I'll come to this,   somehow figure out how to connect the two ends  and actually measure the whole thing collectively  

to see if there's an even or odd number of  electrons. Which is why it's so great as a qubit   because the reason it's hard for Schrödinger's  cat to be both dead and alive is because you're   going to look at it, and then you look at it,  photons are going to bounce off it and you're  

going to know if it's dead or alive. And the thing  is, the thing that was slightly paradoxical is   actually a person doesn't have to perceive it.  If there's anything in the environment that,   you know, if a photon bounces off, it's sort  of like if a tree falls in the forest …

does that “even or oddness” which we call parity,  whether something's even or odd is parity. And,   you know, these are wires with, you know, 100  million electrons in them. And it's a difference   between 100 million and 100 million and one. You  know, because one's an even or odd number. And  

that difference, you have to be able to, like,  the environment can't detect it. So it doesn't get   entangled with anything, and so it can actually be  dead and alive at the same time, you know, unlike   Schrödinger's cat, and that's what you need to  make a qubit, is to create those superpositions.   And so Majorana zero modes are these features of  the system that actually don't actually carry an  

electrical charge. But they are a place where a  single unpaired electron can enter the system and   then disappear. And so they are this remarkable  thing where you can hide stuff. [LAUGHS]

of those cases where we actually had to engineer  the material. And so engineering the material is,   it turns out to be a challenge. People had ideas  early on that they could put some combination of   semiconductors and superconductors. But, you know,  for us to really make progress, we realized that,   you know, it's a very particular combination. And  we had to develop—and we did develop—simulation  

capabilities, classical. Unfortunately, we don't  have a quantum computer, so we had to do this   classically with classical computers. We had to  classically simulate various kinds of materials   combinations to find one, or find a class, that  would get us into the topological phase. And it   turned out lots of details mattered there,  OK. It involves a semiconductor, which is   indium arsenide. It's not silicon, and it's not  the second most common semiconductor, which is  

gallium nitride, which is used in LED lights.  It's something called indium arsenide. It has some   uses as an infrared detector, but it's a different  semiconductor. And we're using it in a nonstandard   way, putting it into contact with aluminum and  getting, kind of, the best of both worlds of a   superconductor and a semiconductor so that we  can control it and get into this topological   phase. And that's a previously published paper in  American Physical [Society] journal. But that's  

great. So that enables … that shows that you  can create this state of matter. Now we need   to then build on it; we have to harness it,  and we have to, as I said, we have to make one   of these wires or, in many cases, multiple  wires, qubits, et cetera, complex devices,   and we need to figure out, how do we measure  whether we have 100 million or 100 million and one  

electrons in one of these wires? And that was the  problem we solved, which is we made a device where   we took something called a quantum dot—you should  think of [it] as a tiny little capacitor—and that   quantum dot is coupled to the wire in such a way  that the coupling … that an electron—it's kind of   remarkable—an electron can quantum mechanically  tunnel from … you know, this is like an electron,  

you don't know where it is at any given time. You  know, its momentum and its position aren't well   defined. So it’s, you know, an electron whose,  let’s say, energy is well defined … actually,   there is some probability amplitude that it's on  the wire and not on the dot. Even though it should   be on the dot, it actually can, kind of, leak out  or quantum mechanically end up on the wire and  

come back. And because of that fact—the simple  fact that its quantum mechanical wave function   can actually have it be on the wire—it actually  becomes sensitive to that even or oddness.

That that tiny change in the capacitance,  if we put it into a larger resonant circuit,   then that larger resonant circuit shows  a small shift in its resonant frequency,   which we can detect. And so what we demonstrated  is we can detect the difference, that one electron   difference, that even or oddness, which is, again,  it's not local property of anywhere in the wire,  

that we can nevertheless detect. And that's, kind  of, the fundamental thing you have to have if you   want to be able to use these things for quantum  information processing, you know, this parity,   you have to be able to measure what that parity  is, right. That's a fundamental thing. Because   ultimately, the information you need is classical  information. You're going to want to know the  

answer to some problem. It's going to be a string  of zeros and ones. You have to measure that. But   moreover, the particular architecture we're using,  the basic operations for us are measurements of   this type, which is a ... it’s a very digital  process. The process … I mentioned, sort of, how   quantum computing looks a little analog in some  ways, but it's not really analog. Well, that's   very manifestly true in our architecture, that  our operations are a succession of measurements  

that we turn on and off, but different kinds  of measurements. And so what the paper shows   is that we can do these measurements. We can  do them fast. We can do them accurately.

but quantum computers can solve and classical  computers have no chance of solving. On the other   hand, there are lots of things that classical  computers are good at that quantum computers  

aren't going to be good at, because it's not  going to give you any big scale up. Like a   lot of big data problems where you have lots of  classical data, you know, a quantum computer with,   let's say, let’s call it 1,000 qubits, and here  I mean 1,000 logical qubits, and we come back   to what that means, but 1,000 error-corrected  qubits can solve problems that you have no chance  

of solving with a classical computer, even  with all the world's computing. But in fact,   if it were a 1,000 qubits, you would have to take  every single atom in the entire universe, OK, and   turn that into a transistor, and it still wouldn't  be big enough. You don't have enough bytes,   even if every single atom in the universe were  a byte. So that's how big these quantum problems   are when you try to store them on a classical  computer, just to store the answer, let's say.

should definitely not view quantum computers  as replacing classical computers. There's lots   of things that classical computers are already  good at and we're not trying to do those things.   But there many things that classical computers  are not good at all. Quantum computer we should   think of as a complimentary thing, an accelerator  for those types of problems. It will have to work   in collaboration with a classical computer  that is going to do the classical steps,  

and the quantum computer will do the quantum  steps. So that's one thing to just keep in mind.   When we talk about a quantum computer, it is part  of a larger computing, you know, framework where   there are many classical elements. It might be  CPUs, it might be GPUs, might be custom ASICs   for certain things, and then quantum computer,  you know, a quantum processor, as well. So …

key design features, that the qubits be small and  small and manufacturable on a single wafer. And   then the QPU also has to enable that quantum  world to talk to the classical world …

we ultimately have to get back a string of zeros  and ones. But that initially is these measurements   where we're getting, you know, phase shifts on  microwaves, and … which are in turn telling us   about small capacitance shifts, which are in turn  telling us the parity of electrons in a wire.

And the whole thing has to be inside of  a cryogenic environment. So it's a very   special environment in which we … in which, A,  it’s kept cold because that's what you need in   order to have a topoconductor, and that's also  what you need in order just in general for the   qubits to be very stable. So that … when we talk  about the full stack, just on the hardware side,   there are many layers to this. And then of course,  you know, there is the classical firmware that  

takes instructions and turns them into the  physical things that need to happen. And then,   of course, we have algorithms and  then ultimately applications.

together, it's an interesting question, what  that world looks like. Can you just take a   brief second to say that you're thinking about  what could happen to cryptography, to, you know,   just all kinds of things that we might be  wondering about in a post-quantum world?

and you have to bake that in. That ultimately, you  know, if you think of a datacenter, ultimately,   like there's a certain amount of energy, and  there's a certain amount of cooling power you   have. And you can say, well, this datacenter is  100 megawatts, and then in the future, we'll have   a gigawatt to use it. But ultimately, then that  energy has to come from somewhere, and you've   got some hard physical constraints. So similarly,  you could ask, you know, with quantum computers,  

what are the hard physical constraints? What are  the things that just … because you can't make a   perpetual motion machine; you can't violate, I  think, laws of quantum mechanics. And I think in   the early days, there was this concern that, you  know, this idea relies on violating something.  

You're doing something that's not going to work.  You know, I'd say the theory of quantum error   correction, the theory of fault tolerance, you  know, many of the algorithms have been developed,   they really do show that there is no fundamental  physical constraint saying that this isn't going   to happen, you know. That, you know, that somehow  you would need to have either more power than you  

can really generate or you would need to go much  colder than you can actually get. That, you know,   there's no physical, you know, no-go result. So  that's an important thing to keep in mind. Now,   the thing is, some people might then be tempted  to say, well, OK, now it's just an engineering   problem because we know this in principle can  work, and we just have to figure out how to work.   But the truth is, there isn't any such, like, hard  barrier where you say, well, oh, up until here,  

it's fundamental physics, and then beyond this,  it's just an engineering problem. The reality is,   you know, new difficulties and challenges arise  every step along the way. And one person might   call it an engineering or an implementation  challenge, and one person may call it a   fundamental, you know, barrier obstruction, and  I think people will probably profitably disagree,  

you know, agree to disagree on, like, where that  goes. I think for us, like, it was really crucial,   you know, as we look out at a scale to  realize quantum computers are going to   really make an impact. We're going to need  thousands, you know, hundreds to thousands   of logical qubits. That is error-corrected  qubits. And when you look at what that means,   that means really million physical qubits.  That is a very large scale in a world in which  

people have mostly learned what we know about  these things from 10 to 100 qubits. To project   out from that to a million, you know, it would  surprise me if the solutions that are optimal for   10 to 100 qubits are the same solutions that  are optimal for a million qubits, right.

small and be able to control these things. So, you  know, I think that's one key thing. And, you know,   what we've demonstrated now is that we can harness  this; that we've got a qubit. And that's why we   have a lot of confidence that, you know, these  are things that aren't going to be decades away.   That these things are going to be years away.  And that was the basis for our interaction   with DARPA [Defense Advanced Research Projects  Agency]. We've just been … signed a contract with  

DARPA to go into the next phase of the DARPA US2QC  program. And, you know, DARPA, the US government,   wants to see a fault-tolerant quantum computer.  And … because they do not want any surprises.

there was no guarantee we would even be able  to do the things we've done so far. I mean,   you know, that's the nature of it. If you're  going to try to do something really different,   you're going to have to take risks. And we did  take risks by really starting at, you know, the   ground floor and trying to redesign and  re-engineer these things. So that was a   necessary part of this journey and the story, was  for us to re-engineer these things in a high-risk  

way. What that leads to is, you know, potentially  changing that timeline. And so in that context,   it's really important to make this transition  to post-quantum crypto because, you know,   the cryptography systems in use up until now are  things that are not safe from quantum attacks if   you have a utility-scale quantum computer. We  do know that there are crypto systems which,   at least as far as we know, appear to be  safe from quantum attacks. That's what's  

called post-quantum cryptography. You know, they  rely on different types of hard math problems,   which quantum computers aren't probably  good at. And so, you know, and changing   over to a new crypto standard isn't something  that happens at the flip of a switch.

professor Sankar Das Sarma. He suggested that  with the emergence of Majorana zero modes,   you had reached the end of the beginning and that  you were now sort of embarking on the beginning of   the end in this work. Well, maybe that's a sort  of romanticized vision of what it is. But could   you give us a little bit of a hint on what are  the next milestones on your road to a scalable,   reliable quantum computer, and what's on  your research roadmap to reach them?

engineering phase, OK. Again, as I said, it's a  matter of debate and even taste of what exactly   you want to call scientific discovery versus  engineering, but—which will be hotly debated,   I'm sure—but it is definitely a continuum that's  going more towards … from one towards the other.  

And I would say, you know, at a high level,  logical qubits, you know, error-corrected,   reliable qubits, are, you know, the basis of  quantum computation at scale and developing,   demonstrating, and building those logical  qubits and logic qubits at scale is kind of   a big thing that—for us and for the whole  industry—is, I would say, is, sort of,  

the next level of quantum computing. Jason Zander  wrote this blog where he talked about level one,   level two, level three, where level one was  this NISQ—noisy intermediate-scale quantum—era;   level two is foundations of, you know, reliable  and logical qubits; and level three is the,   you know, at-scale logical qubits. I think we're  heading towards level two, and so in my mind,  

that's sort of, you know, the next North Star  is really around that. I think there will be   a lot of very interesting and important  things that are more technical and maybe   are not as accessible to a big audience. But  I'd say that's, kind of, the … I would say,   if you're, you know, a thing to keep in mind as  a big, exciting thing happening in the field.