Tall Quantum Claims: Microsoft's New Quantum Computing Chip

Topoconductor, Majorana particles, and a clear path to a million qubit processor

Microsoft has a breathless (but difficult to understand) press release about a new quantum computing chip which makes it appear as though they've solved Quantum Computing (QC). They claim to have invented a new type of material (topoconductor) which allows a new type of qubit (Majorana zero modes) which are far more resistant to errors than regular qubits, significantly reducing the need for error correction, and this could allow us to build quantum computers with millions of qubits and this could revolutionize quantum computing.

Microsoft’s Majorana 1 Chip. The “clear path to a million qubit processor”

We don't know if this is indeed true or just marketing hype, but the science of the claims is fascinating. So let me give a high-level explanation of what's being claimed.

Warning: I've oversimplified a lot, and used very loose analogies. If you are an expert and think anything here is seriously misleading please let me know.

This is a rather complex topic: I'll have to explain bits, qubits, why error correction is so important in qubits, Majorana particles and how qubits based on them help with the error correction problem.

Bits and Qubits

Computers work on bits. A bit is any physical thing that can have two states, one of which we interpret as 1 and the other as 0. So, in a transistor, voltage > 1.2V is 1 and voltage < 0.5V is a 0. On a disk, north-south magnetism is 1 and south-north is 0. In an optic fiber, presence of a light pulse is 1, absence is 0. On a punched card, a hole is 1 and no hole is 0. A qubit is a quantum bit. To understand it, you should read my earlier Twitter thread introducing qubits.

Schrödinger's cat is both, dead and alive at the same time (in a “superposition of those two states”) until you open the box. Then if she’s alive, you’re dead—cat’s don’t forive easily.

The short explanation is that a qubit is like a bit but the two states are quantum states and we can do amazing things because of the amazing properties of matter at those scales. A qubit might be an ion (for which high energy state A is 1 and low energy state B is 0), or a photon (horizontally polarized is 1, vertically polarized is 0), or a superconductor (higher energy current and phase difference is 1 and lower is 0). To make things easy to understand, imagine a box containing Schrödinger's cat: live cat is 1, dead cat is 0. That quickly illustrates the cool thing about qubits: they can be in both states at the same time. Until you open the box, the cat is both, alive and dead. The qubit is both 1 and 0. When you open the box, the cat will definitely be in only one state: alive or dead (qubit = 0 or 1). In quantum terminology: before, the qubit is in a superposition of 1 and 0 states. After we "read" the value of the qubit ("observe") it "collapses" to exactly one of the states.

The uncool thing about qubits is that anything that disturbs the qubit can be interpreted by it as an attempt to read/observe it and it collapses to 1 or 0, ending your computation. This is extremely inconvenient: you don't want this to happen until your computation is done! Imagine a bunch of Schrödinger's cats in transparent boxes. Anyone looking at any box will cause that cat to collapse into one state (dead or alive). Now imagine the boxes in the middle of a busy mall. You don't want anyone looking at the cats while you're doing your computation because otherwise it completely messes up your computation and you have to start over again. Even if you walk past one of these boxes with your eyes closed, there's a probability that you slightly shift the probability between the dead and alive states. (Sorry, stretching the analogy but a better explanation will take too long and will make your eyes glaze over.) But maybe you want to read my earlier Twitter thread explaining the basics of Quantum Computing.

Because of these problems, quantum computers are quite difficult to build in real-life. Random disturbances cause the qubits to collapse (ending your computation prematurely) or the probabilities of the states to shift (introducing errors in your results). And hence, you need error correction. Unfortunately, we can't just fix this problem by having multiple copies of each qubit because of another bizarre property of quantum physics: the no-cloning theorem says that making identical copies of qubits is theoretically impossible.

There’s a little bit of hope: by doing advanced quantum things with qubits, we have figured out how to get say 10 to 100 qubits to work together to pretend to be one single qubit which is resistant to disturbances and collapse (decoherence). But here's the problem: to solve any practical problem using QC, we'll need millions of error-corrected qubits, and the best we've managed so far is (*checks notes*) 105. We are not even sure whether it is theoretically possible to have that many error-corrected qubits.

So what's to be done? Enter Majorana. Imagine a bunch of electrons dancing together. If the dance is a Waltz, we treat it as 1, and if the dance is Swing then 0. A Majorana mode is one such configuration of electrons that are locked together in a certain behavioural pattern. (Some terminology: A Majorana particle¹ is a theoretically predicted fundamental particle with certain bizarre quantum properties (first theorized by a guy called Ettore Majorana). A Majorana mode is a configuration of electrons in a superconductor which behaves like a Majorana particle; so it is called a Majorana quasiparticle.)

How can a "configuration of electrons" "behave like a particle"? It's like a wave: it has predictable properties, like speed/direction of movement, and what happens when two waves meet. Even though a wave isn't a physical thing: it is just a configuration of water particles. When a superconductor is built with a specific physical configuration (called a Josephson junction), such Majorana modes form and they can be in one of two states. These can be interpreted as 1 and 0 for quantum computing giving us a Majorana qubit.

And how does that help? Remember how the whole problem with qubits is that any disturbance can cause their state to collapse/change? Well, a Majorana mode is not a single particle but a configuration of particles, and to disturb it, you'll need to disturb all its components in the same way. Random disturbances, which cause problems for normal qubits, cause far fewer problems for Majorana qubits because the components of the Majorana mode are physically not next to each other, so any disturbances they face are uncorrelated. Hence, Majorana qubits (are theorized to be) resistant to errors.

As a result, Majorana qubits—if indeed we could build them—could revolutionize QC by allowing us to build practical quantum computer using far less physical qubits than regular qubits (which need 10 to 100 physical qubits for one logical error-corrected qubit).

The Microsoft announcement yesterday claims that they've managed to build a superconductor of the correct ("topological") shape which causes the formation of Majorana modes. So now they have announced a chip which they claim has 8 Majorana qubits. Huge if true, because so far, Majorana modes had only been theorized, and there have been only a few laboratory experiments which "are consistent with the formation of Majorana modes". In other words, we were still not sure if we could build Majorana modes. We just thought "yes, maybe we did create them".

Now if Microsoft can not only create Majorana modes but it can also use them to create stable qubits, this is a huge advance over the state of the art. But, we don't know for sure whether this is true or just marketing hype. A paper they published in Nature is far less certain than their press release. See this screenshot from Nature:

This is what Microsoft’s own scientists say!

See also this tweet by Brian Skinner a professor of quantum materials at Ohio State University. And this.

So, that's the story. Microsoft claims to have successfully created a new material ("topoconductor") which allows the reliable creation of a new particle ("Majorana particle") which allows the creation of a new qubit that is resistant to errors. If true, then according to Satya Nadella, "we now have a clear path to a million-qubit processor"—something that was not within our reach before this announcement. And which could change the world, possibly far more than ChatGPT

If.

1. I would have linked to the Wikipedia article about Majorana Fermions but it is unlikely to make any sense to you. Better just ask Claude. (No, don’t ask ChatGPT 4o, it did a terrible job.)