IBM has announced the creation of a modular cooling system designed to be able to maintain the extremely cold temperatures required by large quantum computers.

Realising the potential of quantum computing, which currently remains largely in the realms of scientific theory, requires expanding the size of systems significantly.

IBM’s quantum hardware requires temperatures “colder than outer space”, so advanced cooling is a necessary, it said. Its new modular architecture for cryogenic systems moves from the traditional cylindrical cooling design to large rectangular aluminium cuboids that allow for the connection of multiple chips.

This allows information to move between them, and “extend computations beyond the limitations of a single processor and enable more sophisticated quantum algorithms”, IBM said.

The architecture is a key step in realising IBM’s plan to launch the world's first fault tolerant quantum computer, IBM Quantum Starling, by its planned 2029 date. The company has already successfully coupled two cryogenic cells using the system, it added.

Quantum computers currently require temperatures near absolute zero in order to reduce sources of potential disruption such as excess heat and thermal noise, which can prevent quantum states from lasting long enough for reliable computation.

Currently, IBM achieves this through the use of vacuum insulated containers in which dilution refrigerators cool processors. Now the company believes it is running up against the limits of these cylinders, as they can only hold a single chip. This presents “enormous challenges”, the IBM said, due to spatial constraints, excessive heat generation, and unwanted interference between different operation centres known as qubits.

Its modular system allows multiple chips to be connected together while maintaining the temperatures required for reliable quantum computing. The company claims that cool times and temperature stability remain consistent even as more chips are added, making the system “highly scalable”.

However, this breakthrough is unlikely to be widely applicable. IBM explained in its blog post that this system is designed specifically for IBM hardware as part of its long-term goals for multi-chip quantum systems.

---