Spy Agency Bets on IBM for Universal Quantum Computing

A real-life U.S. version of “Q Branch” from the James Bond films has greater ambitions than creating personal spy gadgets such as exploding watches or weaponized Aston Martins. It’s betting on an IBM team to develop the first logical qubits as crucial building blocks for universal quantum computers capable of outperforming today’s classical computers.

Most quantum computing efforts have focused on building ever-larger arrays of quantum bits, called qubits, made from physical components such as superconducting loops of metal or charged atoms trapped within magnetic fields. Qubits can harness the weird power of quantum physics to exist in two states simultaneously and influence distant qubits through quantum entanglement, but the challenge comes from maintaining fragile quantum states long enough to perform computer calculations. As a next step, the U.S.Intelligence Advanced Research Projects Activity (IARPA) has given IBM a five-year research grant to assemble physical qubits into a single logical qubit that lasts long enough to perform complex computer operations.

“The idea is that the encoded logical qubit would last longer than individual physical qubits, so it could be part of a computation in a larger universal quantum computer,” says Jerry Chow, manager of experimental quantum computing at IBM’s Thomas J. Watson Research Center, in Yorktown Heights, N.Y.

A universal quantum computer would represent a much faster version of a classical computer, given its ability to perform many more calculations simultaneously. Building a logical qubit “is an obvious and widely-recognized ‘next major milestone’ along the way to a universal quantum computer, one that many groups are working towards but that none have achieved yet,” Scott Aaronson, a theoretical computer scientist at MIT, told IEEE Spectrum.

“I'm thrilled that we now live in a time when all these groups are racing
toward it as a realistic prospect, rather than the far-off dream that it used to be,” Aaronson says.

Universal quantum computers could even crack the most complex of today’s codes—a most tantalizing prospect for U.S. intelligence agencies involved in spying and code breaking. That’s why IARPA, in charge of funding high-risk, high-payoff intelligence research, is backing IBM’s quantum computing effort through its Logical Qubits (LogiQ) program.

Earlier this year, IBM demonstrated a four-qubit array of superconducting quantum circuits capable of detecting quantum errors. IARPA funded that IBM effort as well, because quantum computers will need error detection and correction systems for their qubits—and logical qubits—in order to reliably perform computer operations. IBM plans to eventually build a logical qubit based on that previous work, but with a larger array containing a double-digit number of qubits (perhaps no more than 30).

But IBM will need to overcome an engineering challenge in connecting more physical qubits together, Chow says. For example, the team will need to ensure that all the wiring connections for input and output signals don’t lead to cross talk errors that unintentionally affect non-targeted qubits. One possible solution might involve moving from a flat, 2-D grid architecture for the qubits and related wiring to a 3-D integrated architecture.

Another challenge comes from boosting the “coherence” of the physical qubits so that they can maintain their quantum states for longer periods of time. Longer coherence times are necessary to ensure that the effort poured into making a logical qubit from many individual physical qubits would be worthwhile. 

“We’re improving coherence times for individual qubits from 50 to 100 microseconds right now, but we want to continue to push for another order of magnitude increase,” Chow says.” That’s the minimum in getting to the realm of a logical qubit being better than individual physical qubits.”

IBM’s goal of building a single logical qubit during the five-year IARPA contract won’t in and of itself enable universal quantum computers. But IBM does hope to demonstrate the computing logic states of “1” and 0” with such a logical qubit in the process. The tech giant will also begin studying concepts for how to connect such logical qubits together within even larger arrays of physical qubits.

But quantum computing need not simply tread water in the meantime. In the next five years, so-called “analog quantum computers” based on arrays of 50 to 100 qubits could begin helping researchers simulate quantum interactions within chemistry or materials science. Google’s Quantum AI Lab has also been working with D-Wave, a Canadian company, on testing more specialized quantum computers, called quantum annealers, that are limited to solving so-called optimization problems.