Microsoft wants you to build quantum apps in Azure, the cloud that's both up and down

Microsoft believes that solving the world's most intractable problems – reversing climate change, for instance – will ultimately require combining supercomputers, artificial intelligence, and quantum computing.

According to Microsoft's Krysta Svore – who heads up Redmond's quantum development program – it'll take a rather beefy petaflop-class supercomputer capable of interfacing with the quantum elements at 10–100Tbit/sec just to achieve fault tolerance. And at least in Svore's mind, only the cloud offers the scale necessary to achieve that.

While fault tolerant quantum supercomputers remain a ways off – if they're even possible in the first place – Microsoft contends that developers need to start thinking about how quantum algorithms might be used to accelerate their existing workloads. And this week the cloud provider, which is known to tolerate its own faults, made its Azure Quantum platform available to the public to do just that.

"Now researchers can begin developing hybrid applications with a mix of classical and quantum code together that run on today's quantum machines," Svore announced in a blog post.

But unlike some other programs looking to help developers explore the use of quantum algorithms to accelerate traditional computation, Microsoft appears to be doing it on real hardware – as opposed to emulated or simulated environments. Microsoft partners with a slew of quantum computing vendors using a variety of technologies including Quantinuum, IonQ, Quantum Circuits, Rigetti, Pasqal, 1Qloud, and Toshiba, to name a few.

• China leads the world in tech research, could win the future, says think tank
• Assume the superposition: Intel emits SDK to simulate quantum computers
• Google claims milestone in quantum error correction
• China reportedly producing quantum computers – good luck observing one

Svore specifically calls out Quantinuum's systems, but it's not clear if Microsoft's hybrid-quantum program also supports systems from other vendors. The Register reached out for comment and we'll let you know if we hear anything back.

But while the program may be open to the public, the offering isn't aimed at production workloads. Microsoft is still in the early stages here and is targeting exploratory applications run by researchers and scientists – presumably those with fairly hefty budgets.

Cloud-hosted quantum hardware isn't exactly cheap. Using Quantinuum as an example, the standard plan will set you back $125,000 a month – in addition to any Azure infrastructure costs for the conventional compute side of things.

Plenty of competition

Microsoft clearly thinks that its strategy is a winner, but pairing quantum systems with conventional silicon or petaflops of big iron aren't exactly new concepts. 

Several companies and research groups are playing with similar architectures. One of the more obvious is IBM, which has been slinging cloud-based qubits for years now and recently turned on its 433-qubit Osprey system.

Meanwhile, late last year, scientists at Finland's government-sponsored research institute VTT plugged a tiny quantum computer – named Helmi – into Europe's most powerful supercomputer.

The project's goal is to explore the use of quantum computing to accelerate HPC workloads and speed up the development of quantum algorithms and software. And much like Microsoft, VTT believes that hybrid quantum systems will be pivotal to achieving higher performance in HPC simulations.

Then there's Fujitsu, which is exploring similar hybrid-quantum architectures – albeit using simulated qubits running on its Arm-based A64FX processors.

And earlier this month Intel rolled out its own quantum software development kit. Like Fujitsu, Intel is using simulated qubits running on conventional hardware to make developing quantum algorithms for future quantum systems more accessible. ®