NewScientist wrote that IBM scientists were able to assemble the largest Osprey quantum computer out of 433 superconducting qubits. That’s more than three times the company’s previous record of 127 qubits, and more than eight times the size of Google’s 53-bit Sycamore computer.

The number of qubits currently available in quantum devices does not allow solving the applied problems. The quantum advantage demonstrated by the Sycamore superconductor processor means that only on some specially selected tasks the quantum computer is faster than the classical one. It takes tens and hundreds of thousands of qubits to move from abstract tasks to real ones.

In addition to increasing the number of qubits, scientists also have to consider how not to make their processors too bulky, as it turned out, for example, for a group of scientists from China. The photonic computer, in which they demonstrate the quantum advantage, includes an optical interferometer mounted on an optical table using bulk optics (when photons travel through the air and the circuit consists of ordinary mirrors, lenses and other optical elements), just for that. requires a lot of space for mounting and high precision is required for adjustment. Scalability issues can be avoided by transferring interferometers to small chips with optical waveguides. And if photonic circuits on integrated chips have become popular recently, superconducting computers have been printed with them from the very beginning.

IBM researchers aim to scale processors on superconductors, and this time they have succeeded in creating a compact 433-qubit integrated chip. The hardest part of building a chip with a large number of superconducting qubits is preventing or greatly reducing unwanted interactions between the qubits. The more qubits that need to be placed on a chip, the denser they must be placed there, and the more technically difficult it is to deal with cross-talk between qubits.

Physicists took it upon themselves to check whether it would be practical to fit so many superconducting qubits on a chip while maintaining its performance. They have not tried to use the chip for any computation, so it is not yet possible to predict whether they will show a quantum advantage in an implemented problem and how long it will take to install and program the processor. With all this, the company plans to develop an 1121-qubit computer by 2023.