Most quantum computing labs hope to slowly build universal “gate-model” machines that could perform as super-fast versions of today’s classical computers. Such labs have tended to cast a skeptical eye upon D-Wave, the Canadian company that has rapidly developed a more specialized type of quantum computing machine for lease to corporate customers such as Google and Lockheed Martin. In the latest twist, Google has hired an academic team of researchers to help build the first Google quantum computer based on the specialized D-Wave approach rather than on a universal gate-model blueprint.
The Google announcement of its plan to build new quantum computing hardware coincided with its hiring of John Martinis, a professor of physics at the University of California, Santa Barbara, last week. Martinis has led an academic team in developing error correction techniques that can stabilize the quantum bits—called qubits—used by quantum computers to perform many simultaneous calculations by representing both 0 and 1 states at the same time. Many news outlets, including IEEE Spectrum, had initially assumed that Google’s hiring of the Martinis team signaled the technology giant’s intent to develop universal quantum computing hardware as an alternative to D-Wave’s specialized quantum annealing machines.
That turned out to be only partly true. In the long run, Google and Martinis do want to work toward universal gate-model quantum computers capable of solving a wide range of problems. But they have set their immediate sights on building a quantum annealing computer similar to the D-Wave machines that can only solve optimization problems. In the short term, Google wants to use the Martinis team’s expertise to build a more stabilized version of a quantum annealer that can ensure longer coherence times for the system’s fragile qubits.
“We’re taking the approach that if we have longer coherence times, maybe the quantum annealer would work better,” Martinis says. “We know how to make coherent qubits and scale them up.”
Quantum annealing by itself isn’t very controversial. But D-Wave has attracted controversy because it has sacrificed some qubit coherence by scaling up quickly to the 512-qubit D-Wave Two machine—far larger than most experimental quantum computing systems containing just several qubits. Many researchers remain skeptical of whether D-Wave’s quantum annealing machines will ever end up beating classical computers in solving optimization problems. (The D-Wave machines have so far not demonstrated significantly better performance than classical computers.)
Google’s new plan represents a complementary, slow-but-steady approach to building a quantum annealer that could potentially deliver better performance in the long run. The technology giant still plans to work with D-Wave’s scientists as D-Wave scales up to a 1,000-qubit “Washington” processor, says Hartmut Neven, director of engineering at Google’s Quantum Artificial Intelligence Lab, in a blog post. But Google’s own quantum computer plans seek to combine the best of both worlds by ensuring system stability through qubit coherence as the hardware scales up in size.
The Martinis group had previously built quantum computing systems of up to nine qubits based on superconducting quantum circuits—the same type of general hardware design used by D-Wave’s machines. Under the new Google effort, Martinis hopes his team can roughly double the number of qubits every year and eventually work up to 40 or 80 qubits through “brute-force” scaling. “Forty qubits is a large enough number so that you can really tell if the device is going to give any interesting performance,” Martinis says.
At the same time, Martinis and his team will continue developing error-correction codes for Google with the aim of uncovering and fixing errors in universal logic-gate quantum computers. In May, they demonstrated a type of error-correction code called surface code that can work with lower accuracy thresholds for quantum logic operations.
The recruitment of the Martinis group signals Google’s intent to recruit or tap into talent with wide-ranging expertise at the Google Quantum Artificial Intelligence Lab. Such a move also inaugurates a new era of cooperation between academic researchers and D-Wave under the Google umbrella—a scenario that would have seemed unbelievable just several years ago because of the skepticism and heated debates surrounding D-Wave’s machines. Scott Aaronson, a theoretical computer scientist at MIT and a leading D-Wave critic, applauded the Martinis group’s move to Google on his Shtetl-Optimized blog.
As for Martinis, the UCSB researcher described his excitement about moving from the university environment to the application-focused Google lab. He anticipates being able to focus more on the business of building a working quantum computer with his more permanent staff at Google, as opposed to university projects with a rotating cast of students and post-doctoral fellows who typically leave in several years. Many of his senior researchers, including post-doctoral fellows, have moved over to work at Google. The graduate students on his team will continue working at the University of California, Santa Barbara, through a grant from Google. (Martinis currently holds a joint appointment at Google and UCSB.)
“One of the things that really attracted me to Google was Hartmut’s real desire to solve computer science problems with a quantum computer, not in the long term but in an immediate way,” Martinis says. “Now we have the opportunity to work hard and really try to do that.”
Jeremy Hsu has been working as a science and technology journalist in New York City since 2008. He has written on subjects as diverse as supercomputing and wearable electronics for IEEE Spectrum. When he’s not trying to wrap his head around the latest quantum computing news for Spectrum, he also contributes to a variety of publications such as Scientific American, Discover, Popular Science, and others. He is a graduate of New York University’s Science, Health & Environmental Reporting Program.