Error correction could make these revolutionary machines more practical sooner than expected.
For years, quantum computing has been the preserve of academics. New advances, however, are pushing this potentially revolutionary technology toward practical applications.
At the Q2B conference this month, quantum computer makers Google , IBM , Honeywell , IonQ and Xanadu detailed specific steps they expect by 2024 that will push their machines further down the road of commercial practicality. Those achievements include increasing quantum computers' scale, performance and reliability. Private sector spending on quantum computing products and services will likely more than triple to $830 million in 2024, up from $250 million in 2019, according to a forecast from Hyperion Research.
"We're in the early industrial era of quantum computing," said Seth Lloyd, an MIT professor who helped found the field in the 1990s. He says the "huge advances" are comparable to the early use of steam engines to power factories, ships and trains.
One buzzworthy breakthrough is progress toward error correction, which should let quantum computers perform sustained calculations instead of fleeting spurts of work. That improvement comes through overcoming a fundamental limit with qubits, the basic elements for storing and processing data in a quantum computer. Qubits are easily perturbed by outside forces, but error correction is designed to overcome the finickiness of individual qubits. It'll require bigger machines with many more qubits, but quantum computer makers see progress there, too.
If quantum computer makers succeed, error correction could help the industry realize its promise to dramatically improve on conventional processor performance for some important problems. Quantum computers won't replace classical machines, which also face manufacturing difficulties and rising costs, but they could reach beyond today's limits to design new solar panels, lower airplane fuel usage, speed up artificial intelligence, improve financial investing and cut delivery costs.
Conventional computers store information as bits -- ones or zeros -- and perform calculations using tiny electronic data-processing components called transistors. In contrast, quantum computers' qubits can store a combination of one and zero at the same time thanks to a quantum physics phenomenon called superposition. Qubits can be interlinked by entanglement, another quantum physics phenomenon.
Quantum computing involves a series of manipulations to qubits' states. These manipulations are called quantum gates, and a sequence of gate manipulations is called a circuit. As gate manipulations are added, a circuit becomes "deeper" and capable of more sophisticated quantum computation.
Increasing the number of qubits also exponentially increases the size of the computing problem that's within reach. Adding a single qubit doubles the scale of computation that's possible. Adding two quadruples it, adding three octuples it and so on.
These advances excite computer scientists because, although today's machines have a few dozen qubits, tomorrow's will have thousands, then millions.
Quantum computer makers are all working on different ways to build more stable qubits for a stronger foundation to the qubits themselves and how they're connected. Disturbances to either derail the calculation.
Where makers of conventional silicon chips have settled down on one approach, quantum computer makers are exploring widely different possibilities for their qubits.
Google and IBM use superconducting circuits cooled nearly to absolute zero, colder than outer space. Honeywell's ion trap design makes qubits from electrically charged ytterbium atoms. Intel's qubits are individual electrons distinguished by a quantum mechanical property called spin. Xanadu uses photons, and its quantum processors work at room temperature.
A strong foundation is good, but error correction still is essential as a way to overcome individual qubits' flakiness. The main idea for error correction is yoking multiple qubits together into a single "logical" qubit whose state persists longer. Eric Lucero, who runs Google's quantum computing service, calls them "perfect forever qubits." Error correction is the foundation for what's called a fault tolerant quantum computer.
One logical qubit could require as many as 1,000 physical qubits, and serious quantum computing, like Shor's algorithm used to crack today's encryption, requires thousands of logical qubits. IonQ hopes its approach will require as few as just 13 physical qubits for one logical qubit, IonQ chief scientist and co-founder Chris Monroe said at Q2B.
The approach is moving from the theoretical to the practical.
"We've got the technology today," Lucero said. He expects Google will have its first logical qubits in 2023 and 1,000 of them by the end of the decade.
Error correction is a big incentive for increasing qubit counts.
IBM aims to surpass its current 65-qubit system, Hummingbird, with 127-qubit Eagle next year and 433-qubit Osprey in 2022. Then, in 2023, the 1,121-qubit Condor will be "an important inflection point" in making quantum computing algorithms more useful, said Anthony Annunziata, director of IBM's Q Network
Xanadu has 24 qubits now and expects a 40-qubit chip this year, says Zachary Vernon, the company's hardware chief. In coming years, he forecasts qubit counts should double every six to 12 months.
Although researchers are careful to avoid promises of breakthroughs, quantum computers could be useful before error correction arrives. IBM quantum customers today include JPMorgan, ExxonMobil, Mitsubishi Chemical, Daimler, Delta and Boeing.
Some of these customers are interested in designing materials from the molecule up -- one of the first ideas that famed physicist Richard Feynman described in seminal thinking about quantum computers. The hope is for breakthroughs like more efficient solar panels, batteries that store more energy or fertilizer manufacturing that doesn't need so much power.
European aerospace giant Airbus has an extensive program, Marc Fischer, the company's senior vice president for flight physics, said at Q2B. It's investigating quantum computing for improving aircraft aerodynamics, economizing airplane fuel use during ascent, loading planes more efficiently and designing wings with factors hard to calculate using classical computers, he said.
Honeywell sees using quantum computing for its own businesses, like chemical design, warehouse automation and aerospace. "Honeywell expects to be our own biggest and best customer," said Tony Uttley, president of Honeywell Quantum Solutions.
One of the most bullish voices is Eric Schmidt , who in his former job as Google's chief executive and executive chairman approved that company's long-term quantum computing program. That work produced last year's "quantum supremacy" experiment that showed quantum computers could surpass classical computers for at least one narrow (though not practical) computing chore.
"We know this stuff is going to happen six to eight years from now," Schmidt said. "It's going to be incredible when it happens."