IBM has introduced the first quantum computer with more than 1,000 qubits — the equivalent of a computer's digital bits. However, the business has stated that it would now turn its focus to making its machines more error-resistant rather than larger.
For years, IBM has followed a quantum-computing roadmap that has seen the number of qubits nearly quadruple every year. The Condor chip, which was introduced on December 4th, includes 1,121 superconducting qubits stacked in a honeycomb arrangement. It follows its previous record-breaking, bird-named devices, which included a 127-qubit processor in 2021 and a 433-qubit chip last year.
Quantum computers have the potential to execute computations that traditional computers cannot. They will accomplish this by taking use of uniquely quantum phenomena like as entanglement and superposition, which allow numerous qubits to exist in multiple collective states at the same time.
However, these quantum states are notoriously volatile and prone to mistake. Physicists have tried to get around this by enticing numerous physical qubits — each encoded in a superconducting circuit, example, or an individual ion — to work together to represent one qubit of information, or 'logical qubit'.
As part of its new strategy, the company also announced Heron, a chip with 133 qubits but a record-low error rate, three times lower than its prior quantum processor.
According to researchers, state-of-the-art error-correction techniques will necessitate more than 1,000 physical qubits for each logical qubit. A machine capable of useful computations would then require millions of physical qubits.
However, in recent months, scientists have been enthused about a new error-correction system known as quantum low-density parity check (qLDPC). According to a preprint by IBM researchers1, it promises to reduce that number by a factor of ten or more. The business says it will now concentrate on developing circuits that can carry a few qLDPC-corrected qubits in as little as 400 physical qubits, and then networking those chips together.
The IBM preprint is described as "excellent theoretical work" by Mikhail Lukin, a physicist at Harvard University in Cambridge, Massachusetts. "That being said, implementing this approach with superconducting qubits seem to be extremely challenging and it will likely take years before even a proof-of-concept experiment can be tried in this platform," Lukin added. Lukin and his colleagues conducted a similar investigation on the possibility of implementing qLDPC utilising individual atoms rather than superconducting loops2.
The hitch is that each qubit in the qLDPC approach must be directly coupled to at least six others. Each qubit in a typical superconducting chip is only coupled to two or three neighbours. However, Oliver Dial, a condensed-matter physicist and IBM Quantum's chief technology officer, says the company has a plan: it will add a layer to the design of its quantum chips to allow the extra connections required by the qLDPC scheme.
A new IBM road map for quantum research, published today, envisions the company attaining meaningful computations — such as replicating the workings of catalyst molecules — by the end of the decade. "It's always been the dream, and it's always been a distant dream," Dial said. "Actually having it come close enough that we can see the path from where we are today for me is enormous."
