By default, every quantum computer will be a hybrid combining quantum and classical computing. Microsoft estimates that a quantum computer that can help solve some of the world’s most pressing questions requires at least a million stable qubits. It will take enormous classical computing power – which is really only available in the cloud – to control a machine like this and handle the error correction algorithms needed to keep it stable. In fact, Microsoft estimates that to achieve the required fault tolerance, a quantum computer must be integrated with a peta-scale computing platform that can manage between 10 and 100 terabits per second of data being transferred between the quantum machine and the classical machine. At the American Physical Society’s March meeting in Las Vegas today, Microsoft is showing some of the work it has done to make this possible and introduce what it calls the “Integrated Hybrid” feature in Azure Quantum.
“With this built-in hybrid capability, you can start using classic code right next to quantum code in your quantum applications,” Krysta Svore, Microsoft’s VP of Advanced Quantum Development, told me. “It’s the mixing of this classical and quantum code that unlocks new types, new styles of quantum algorithms, prototypes, subroutines if you will, where you can control what you do with qubits based on classical information. This is an industry first.”
Photo credit: Microsoft
This, she argued, is a step to bring classical and quantum computing together, but also to enable new error correction protocols. Without this massive amount of classical computing power, it will not be possible – at least in the foreseeable future – to control a quantum machine effectively.
“The only place you can have scale-up quantum machines, scale-up quantum computing is probably going to be on a public cloud because it’s so critical that this scale of classical computing is integrated with the quantum machine,” Svore explained. She describes the process as a dance in which the classic computer helps choreograph a million qubits to work together at the same time, “everyone doing their little square dance — or hexagon dance, whatever it is.” But to do that, you have to talk to all of those qubits at once, which requires those massive computational and bandwidth requirements.
The story goes on
Svore also argues that it takes a lot of classical computing power to create the algorithms that are then sent to the quantum machine – which can then also take weeks to perform a given calculation (and this feedback loop can also occur multiple times).
With this new built-in hybrid capability, Microsoft is giving developers – and researchers – the tools to explore how this combination of quantum and classical looks in practice. In particular, Svore told me that it will allow them, for example, to run a version of the phase estimation algorithm, which is a key algorithm in the quantum computing toolkit. Researchers will soon be able to test this with the Quantinuum hardware available in Azure and then, for example, have the classical computer react to data coming back from the quantum machine. So far much of this has been theoretical, but now it will be possible to do it in hardware.
Over time, the role of classical computing in enabling quantum computing has become better understood in the industry. Microsoft, of course, argues that its massive cloud will allow it to provide the kind of classic computing power needed to control these machines. It’s obviously not the only player in the market, as Amazon, Google, IBM, and others are also poised to integrate quantum processors into their massive data centers.