Don’t employ quantum computing experts? Just head to the cloud
You may never have direct access to a quantum computer. That doesn't mean you can't use one.
When quantum computers finally arrive, practically no one will run out and buy a quantum computer. Instead, they'll access them via the internet.
Early quantum computers are extremely complicated and extremely expensive, and they'll remain that way for a long time. That's why almost all initial users are going to consume quantum computing as a cloud service, joining the ranks of those who have decided to ditch their own data centers in favor of services from Amazon Web Services, Microsoft and Google.
And according to quantum computing experts, most early users of quantum computers will use the hardware as just a slice of an overall computing footprint that will rely on classical computing for the vast majority of the work.
"We can't really talk about modern computing in isolation from networks and from the cloud," says David Rivas, SVP of Systems & Services at Rigetti Computing, a quantum computing startup.
His company signed a deal in November 2019 to provide access to its computers through Amazon Web Services, as part of the Amazon Braket, the cloud provider's planned service for making quantum computers available to its customers. Rival quantum startups D-Wave and IonQ will also host their devices on the platform. But it's not the only one: Microsoft, as well as developing its own quantum computer, last November announced Azure Quantum, a service that lets developers explore how to build quantum applications alongside classical ones. Azure currently trails AWS for market share in the cloud market, but it is growing strongly.
A small piece of a big puzzle
Several industries are already pushing the boundaries of what classical computing can accomplish because they have pressing needs that can't sit idle waiting for quantum computation. Those types of companies — from financial services and logistics, to health care and pharmaceuticals — will need to integrate their quantum computing activity alongside all of their other classical computing work, and the easiest way to do that is on the cloud, says Ben Porter, director of business development for Microsoft Quantum.
Rivas drew an analogy between quantum computing and present-day applications that might use a graphics processing unit, or GPU, on a cloud service. GPUs are particularly well-suited to machine-learning applications, but unsuited to the vast majority of the applications computing needs, which are handled by conventional cloud servers.
"If you're trying to compute the ground state of a medium-sized molecule, and present that to the user, and integrate it with a more general material-science calculation that you're doing, it's that ground-state calculation that needs to be run in the context of the quantum computer," Rivas says. "All the rest of that stuff can be run normally."
Another way to think about using quantum computers is in the context of microservices, a modern software-development technique that many companies are exploring. Microservices allow developers to break their apps into smaller pieces that can be tweaked without having to mess with the entire code base, and quantum computing can be thought of as just one piece of a complicated application puzzle.
"The types of customers that are best suited for exploring this technology and starting to build solutions are customers that already have complex computational problems," Porter says.
Making quantum computation accessible
Quantum computing will require some developers to think very differently about how they build software. But the industry is also trying to make it as easy as possible to get many of them up and running with quantum computation by allowing them to use existing tools that people already know, Rivas says.
Rigetti has developed a way to let developers write quantum algorithms using Python, a widely used programming language, that it translates into Quil, a quantum-specific assembly language used to program its computers. Much of the history of computer science consists of breakthroughs that make complex systems much easier to use — just like the graphical user interface freed early computer users from the command line. And Rivas says that he would "expect you'll see a lot of abstractions" as quantum computing develops, too.
Rigetti also allows developers targeting applications at its quantum computers to package those applications in containers, which allow developers to create applications that can run across several different systems. "There's no reason to reinvent all this stuff," Rivas says.
More work needs to be done in this area, but quantum computing on the cloud should allow customers to access on-demand quantum resources as well as schedule fixed amounts of those resources, much as they currently do with virtual machines on cloud servers. In the early days, on-demand quantum resources are likely to be limited and expensive, given that even quantum hardware makers will expand slowly. But the plumbing needed to make that work is being installed, Porter says.
"This is a journey," Porter says. "The things that we're learning now about how quantum computers work are influencing the quantum solutions we're building today."