IBM published its roadmap for the future of its quantum computer hardware for the first time today. There’s a lot to digest here, but the main news in the short term is that the company believes it is well on its way to building a quantum processor with more than 1,000 qubits – and somewhere between 10 and 50 logical qubits – by the end of 2023.
Currently, the company’s quantum processors are reaching 65 qubits. There are plans to launch a 127-qubit processor and a 433-qubit machine in 2022 next year. To achieve this, IBM is also building a completely new dilution refrigerator for these larger chips, as well as the technology for connecting multiple chips. These units form a system that is similar to today̵
Dario Gil from IBM tells me the company made a conscious choice in announcing this roadmap and compared it to the birth of the semiconductor industry.
“If you Let’s look at the difference between what it takes to build an industry versus a project or scientific experiment and advancing a field. Our philosophy was that we had to build a team that did three things well. in terms of cultures that need to come together. And that was a culture of science, a culture of the roadmap and a culture of agility, ”Gil said.
He argues that the company could have taken two different routes to achieve the ultimate goal of the quantum industry, which is to build a large, fault-tolerant quantum computer. The first is more like the Apollo program, where everyone comes together, works on one problem for a decade, and then all the different parts come together for that one breakthrough.
“Another philosophy is to say,” What can you do today? “And the ability to use,” he said. “And then you have user-driven feedback, a culture of agility, as a mechanism to keep delivering to a community and that way build a community, and you need to create a roadmap for progress. We firmly believe in this latter model. And to do that you have to do the science, the roadmap and the feedback and get things out there. “
But he also argues that we have now reached a new moment in quantum industry. “We have reached a point where there is enough total investment. This is really important in order to have coordination and signaling mechanisms in place so that we do not grossly misallocate resources and allow everyone to make their contribution.”
He compares it to the early days of the semiconductor industry, when everyone did everything, but over time a third-party ecosystem emerged. As companies adopt new technologies like UV lithography today, the roadmaps that IBM believes are being made for the quantum industry today help everyone coordinate their efforts.
He also argues that the industry has reached a point where the level of complexity has increased so much that individual players can no longer do everything themselves. This in turn means that different actors in the ecosystem can focus on specializing and figuring out what they do best.
“You do that, do you need materials? The position technology? Then you need the device expertise. How do you do the pairing? How do you pack? How do you do the wiring? How do you do the amplifiers, the cryogenics, the room temperature electronics and then the entire software stack from bottom to top? And on and on and on. So you can say, “Well, you know, we’ll do anything.” Okay, well, in the beginning you’ll have to do whatever it takes to integrate, but as time goes on, should we get into coaxial cabling? “
We are already seeing some of these today, for example in the recent collaboration between Q-CTRL and Quantum Machines.
Gil believes 2023 will be a turning point in the industry as the journey to the 1,121 qubit machine brings improvements across the stack. The most important – and most ambitious – of these performance improvements that IBM wants to make is to bring the error rate down from around 1% today to a little closer to 0.0001%. But if you look at the trajectory of the machines a few years ago, this is the number the line is pointing to.
But that’s only part of the problem. As Gil noted, “As you get richer and more sophisticated with this technology, each layer of the innovation stack becomes almost like an infinite field.” This applies to the semiconductor industry and perhaps even more so to the quantum industry. And as these chips get more sophisticated, they get bigger too – which means even the 10-foot refrigerator that IBM is building right now can’t hold more than a million qubits. This is where you need to make the connections between these chambers (because when one chamber alone takes nearly 14 days to cool down, you can’t really experiment and iterate at any appreciable rate). Building such a “quantum intranet,” as Gil calls it, is anything but trivial, but it will be the key to building larger, interconnected machines. And that’s just one of the many areas where inventions are still needed – and it may be a decade before these systems work as expected.
“We are pursuing all of these fronts in parallel,” said Gil. “We’re investing in horizons where the device and features will be available in a decade […]Because if you have this problem and only then start you will never get there. “
While the company – and its competitors – work on building the hardware, there’s also a big effort going on building the software stack for quantum computers. One thing Gil emphasized here is that now is the time to start thinking about quantum algorithms and quantum circuits, even though they still perform worse than classic machines on quantum computers today. In fact, Gil wants developers to think less about qubits than about circuits.
“When [developers] Call up a function and now it goes to the cloud. What’s going to happen behind the scenes? There will be libraries of quantum circuits and there will be a tremendous amount of innovation, creativity and intellectual property on those circuits, ”Gil explained. And then these circuits have to be mapped onto the correct quantum hardware, and in fact it looks like IBM’s vision here is not for a single type of quantum processor, but for those with different layouts and topologies.
“We ourselves already operate over a billion quantum circuits a day from the outside world – over a billion a day,” said Gil. “In the future, billions of quantum circuits will be performed every day on quantum hardware behind the scenes using these cloud-enabled services embedded in software applications. ”