Quantum computers are prone to high error rates, so, to make qubits usable, a lot of redundancy is required. It typically takes hundreds—even thousands—of physical qubits to make one usable, “logical” qubit.

This has been a major obstacle to the development of practical quantum computers. If thousands of qubits are needed for a quantum computer to do anything useful, and it takes a thousand physical qubits to make one logical one, then we’re talking millions of qubits—and, today, the largest qubit array is just more than 6,000 qubits, and that was an experimental demonstration.

The largest general-purpose quantum computers today are IBM’s Condor, with 1,121 superconducting qubits, and Atom Computing’s AC1000 with more than 1,200 neutral atom qubits—and both counts are for physical qubits, not logical qubits.

Last week, quantum computing company QuEra published a paper showing that a logical qubit can be built with just two physical qubits. Neutral atom computers have two types of qubits—memory and computation—and these qubits can be moved from one mode to another. The two-to-one ratio is currently just for the memory qubits.

“The paper does not show—yet—that you can do operations on the qubit,” says Yuval Boger, chief commercial officer at QuEra Computing. “We’ve done it with other codes, but not this particular one. But without quantum memory, the quantum computer won’t work anyway.”

The same algorithm should also have a very significant improvement in error correction on the entangled computation side of the quantum computer as well, he adds. “But we haven’t published that result yet,” he says. “I’m sure that will be forthcoming.”

Like Atom Computing, QuEra makes a neutral atom quantum computer, and its Gemini model has 260 physical qubits and is commercially available.

“We’ve demonstrated a 3,000-qubit machine running continuously,” Boger adds.

That puts a usable quantum computer within sight, he says. “There is still work to be done. But it’s not years and years and years.”

The timeline has also been affected by recent changes in estimates as to how many qubits a quantum computer actually needs to be useful. For example, according to the latest Google research, it could take as few as 1,200 logical qubits for a quantum computer to break elliptic curve cryptography.

The QuEra announcement is also helping position neutral atoms as a viable alternative to the superconducting approach, such as the one used by IBM, says Holger Mueller, analyst at Constellation Research.

“It’s the race for error correction,” he says. The question is whether QuEra’s approach will be the one that works. “How good or record-breaking the result is remains to be seen,” Mueller says.

And the paper is focused on memory, he adds, though the company plans to use the same algorithm for other qubits. “So, it is a key report to keep the neutral atom technology and vendors viable,” he says.

According to Sridhar Tayur, professor of operations management at Carnegie Mellon University’s Tepper School of Business, there are four levels of quantum computing breakthroughs.

“There’s something on paper, then there’s something in a lab, like a proof of concept,” he says. “Then there’s a prototype at scale, and then there’s production scale.”

The QuEra announcement is just at the research paper level, he says. Many people see these theoretical results, and their minds immediately jump to commercial products, he says. “But it’s not like they even have a physical demonstration of this,” Tayur says.