Microsoft has made it clear it intends to be a serious competitor in the emerging quantum computing software market. What the company hadn't made so clear, until now, are its ambitions to be a quantum hardware supplier.
At its recent Ignite conference, the company said it is developing a chip capable of running Microsoft quantum software and plans to develop a complete quantum system. The company also unveiled its Azure Quantum Service, now in private preview. Microsoft open sourced its Quantum Development Kit (QDK) and Q# compilers and simulators for developers this past summer.
Julie Love, Ph.D., senior director of Quantum Computing at Microsoft, discussed the new Azure Quantum Service and the progress the company has made toward being a quantum hardware company.
Microsoft has been working on quantum technology for a while now, but recently redoubled its efforts in this arena. Why now?
Julie Love: It's really more continued momentum of the work and investments we've made in quantum computing since the early 2000s. In fact, it was two years ago at Ignite where we announced delivery of Q#, a language meant to be very accessible for programmers.
To realize the potential of quantum computing we needed to bring together the global community of companies building the hardware, their programmers and subject matter experts. I run the early quantum access program where we bring partners and enterprise customers in to meet with our quantum algorithm developers and we've seen momentum built there. It's this momentum that told us it's time to bring all of our work over to Azure, including the applications, tools and algorithms.
Most of your competitors in this space have a hardware-oriented approach to this market and aren't working in concert with a group of partners. Microsoft seems to have a software-oriented approach.
Love: Well, we are also a hardware company. We are building our own quantum hardware, and have made significant investments there.
Let's start with the new chip. It isn't a quantum chip per se, but it has capabilities that support quantum software. Is this based, in part, on the x86 architecture or is it a whole new design?
Love: The promise of quantum computing is it can solve world-changing problems, such as global warming, as well as problems enterprises face every day. To achieve that promise we need a full system that scales, with hundreds and thousands of qubits and those qubits have to be logical qubits and not the noisy qubits that are susceptible to errors. One option is to either turn out a lot of noisy qubits and deal with the error correction, with the issue there being you need 10,000 physical qubits that do proper error correction.
But that's why we have this huge development effort in our labs because we think it's worth it to figure out our own qubit approach. The other part of the challenge is these qubits must work with Azure, where many of our customers are, and then control those qubits at scale in a cryogenic environment.
Compared to your hardware partners and competitors, what is the value you bring to market with this chip?
Love: You see some of these (quantum) systems with swarms of wires to control all these qubits. But the chip we have can control up to 50,000 physical qubits with that one chip and just three wires. We think that is a critical element.
How far along is the Microsoft quantum hardware system development and what internal systems have you been running your quantum software on?
Love: We use a range of quantum systems in-house. We are developing the topological qubit, it's a multi-decade effort to reinvent the approach to building qubits. If we have qubits with lower error rates, which allows us to scale with far fewer resources in this constrained cryogenic environment.
Do you have some idea when you can deliver a complete Microsoft quantum system?
Julie LoveSenior Director, Quantum Computing at Microsoft
Love: We've done a lot of work on our topological qubit and a lot of work on materials development and the manufacturing capabilities to make this system scalable. We feel we are on the brink of something. We will be bringing our hardware to Azure Quantum, which can sit right alongside the systems of our partners.
Will partners like Honeywell and IonQ use your chip, or will they will stand pat with what they already have?
Love: We have produced this chip and it is applicable to our hardware and let's see if we take it to market and make it available to other hardware makers.
Is it your hope to attract other OEMs like an HPE or Dell into using your chip?
Love: We will explore that opportunity at that time.
Is the best and fastest way into corporate IT shops to blend quantum and classical algorithms that can jointly solve specific problems?
Love: Quantum will be a hybrid architecture working alongside classical computers. If you look at the larger areas of opportunity, the most promise for quantum systems is in chemistry, science, optimization and machine learning. So just as you use acceleration options in Azure today, using GPUs or FPGAs for the big workloads, so will you use both quantum and classical systems (algorithms) either together or separately for solving a range of problems depending on their complexity.
Microsoft has focused hard on bridging the delta between Windows and Linux the past several years. Will that continue between Azure Quantum and Linux?
Love: With offerings like Q# and QDK we have assembled a very open environment where you can stand up a Linux VM and use it to call into a quantum system. There won't be any inconsistencies there. Q# and QDK have a set of debugging tools for developers that lets us profile the code and know the runtime. We can discover quickly what is consuming the resources and bringing the runtimes down. We will have something that can run on future quantum systems to solve complex problems in a day and a half compared to tens of thousands of years on a classical system.
This is algorithmic work we can do now. Our teams now can solve hard computational problems, particularly in optimization. We are discovering tricks learned from using quantum methods in Azure we can apply on classical computers, like the work we are doing now with Case Western [Reserve University].
As you said, Microsoft has invested heavily in quantum computing for a while. Is there any pressure from upper management to start bringing in revenues?
Love: Microsoft has the ability to be long-term investors in this technology. We see significant potential in this for years to come with these grand challenge problems, but also the challenges in the here and now that enterprises are facing. What is really exciting is what we call Quantum-inspired algorithms, which are quantum algorithms that work with conventional hardware. We can already see an acceleration there.
As a physicist I know that when you come at a problem from a different angle you come out with different math formulations and a different approach. We already see that these quantum-inspired algorithms in Azure that can transform the way an enterprise can work. And when we get some quantum hardware underneath it for that workload, you will get another quadratic speed up.