The Tiny Startup Racing Google to Build a Quantum Computing Chip
Rigetti Computing is working on designs for quantum-powered chips to perform previously impossible feats that advance chemistry and machine learning.
The airy Berkeley office space of startup Rigetti Computing boasts three refrigerators—but only one of them stores food.
The other two use liquid helium to cool experimental computer chips to a fraction of a degree from absolute zero. The two-year-old company is trying to build the hardware needed to power a quantum computer, which could trounce any conventional machine by tapping into quantum mechanics.
The company aims to produce a prototype chip by the end of 2017 that is significantly more complex than those built by other groups working on fully programmable quantum computers. The following generation of chips should be able to accelerate some kinds of machine learning and run highly accurate chemistry simulations that might unlock new kinds of industrial processes, says Chad Rigetti, the startup’s founder and CEO.
“The chips we roll out will be able to solve very profound problems,” he says. As an example, Rigetti cites the Haber-Bosch process, used to manufacture ammonia for fertilizer production, which has been estimated to consume 2 percent of the world’s energy. Devising a more efficient catalyst for the reaction could be extremely valuable.
Rigetti aims to ultimately set up a kind of quantum-powered cloud computing service, where customers pay to run problems on the company’s superconducting chips. It is also working on software to make it easy for other companies to write code for its quantum hardware.
That plan requires Rigetti to make leaps of science and engineering that have so far eluded government, academic, and corporate labs. Although physicists have sketched out the basics of how quantum computers could be designed and what benefits they might bring, building them is proving tricky.
It involves wiring together devices called qubits, which represent digital bits of data using delicate quantum-mechanical states. Just like the basic components of a conventional computer, they can encode either a 0 or a 1—but they can also enter a state that is effectively both at the same time. When qubits in that state of “superposition” interact, they can take computational shortcuts not available to conventional computers.
Physicists have made qubits in various different ways. But academic and government researchers have gotten only small numbers of qubits working together. A Canadian startup called D-Wave has sold a chip with more than a thousand qubits to clients including Lockheed Martin and Google, but the technology has not yet been conclusively proven to offer a quantum computer’s benefits.
Qubits are difficult to operate in groups because the quantum states they use to represent data are extremely delicate and the devices interfere with one another. Rigetti says his company has worked out a qubit design that should be stable enough to scale up, and that can be made using conventional chip-manufacturing techniques.
The startup is currently testing a three-qubit chip made using aluminum circuits on a silicon wafer, and the design due next year should have 40 qubits. Rigetti says that’s possible thanks to design software his company has created that reduces the number of prototypes that will need to be built on the way to a final design. Versions with 100 or more qubits would be able to improve on ordinary computers when it comes to chemistry simulations and machine learning, he says.
Others working on quantum computing share the belief that qubit technology has finally reached a point where the devices can be combined in much larger numbers. The leader of Google’s quantum computing lab, which like Rigetti uses superconducting qubits, has predicted that he can build chips with about 100 reliable qubits in a couple of years (see “Google’s Quantum Dream Machine”). Researchers at IBM, MIT Lincoln Lab, and elsewhere have also developed superconducting qubits of high quality (see “IBM Shows Off a Quantum Computing Chip”). Rigetti previously worked in IBM’s quantum computing research group.
“This is a very exciting time,” says Daniel Lidar, director of the Center for Quantum Information Science and Technology at the University of Southern California. “This is not incremental; we’re really starting to see various groups working with superconducting qubits taking big strides forward.”
However, it is still far from clear when useful, large-scale quantum chips might be made, says Lidar. A serious attempt at building one remains an expensive undertaking, he says.
The profits from Google’s online ad business make it free to dedicate enormous resources to its quantum computing lab should it wish. D-Wave, the only company to offer a large-scale superconducting qubit chip, employs over 100 people and has received more than $120 million from various investors, including Amazon founder Jeff Bezos and the CIA.
Rigetti’s company has so far raised $5 million in funding and employs about 15 people. He argues that the intense confines of a startup provide the best environment for solving the big challenge of scaling up qubit technology, and that the company will raise more money and add employees as needed.
Keep up with the latest in AI chips at EmTech Digital.
The Countdown has begun.
March 25-26, 2019
San Francisco, CA