IBM has announced a new architecture developed to address one of the biggest technical hurdles of quantum computers, the error correction problem. With this architecture, the company plans to offer a large-scale, fault-tolerant quantum computer to customers in 2029.

IBM solved quantum errors! The revolution begins in 2029

The basis of this system, named Starling, is the qLDPC (quantum low-density parity check) codes developed by IBM as a new approach to quantum error correction. Qubits, the building blocks of quantum computers, are inherently very prone to errors. Therefore, error correction is essential to build large-scale and reliable systems.

The traditional surface code approach requires approximately 1,000 physical qubits for each logical qubit. However, IBM realized that it was unrealistic to produce hardware to support this method and began researching alternatives in 2019. IBM, which introduced the qLDPC approach in an article published in the journal Nature last year, has now announced a new quantum architecture that can implement this method.

The first step of the new architecture will be taken with the Loon processor, which will be introduced towards the end of the year. This chip includes connectors that allow distant qubits on the same hardware to connect to each other. Such non-local interactions offer a more efficient structure than surface code, which only allows communication with neighboring qubits.

According to the roadmap announced by the company, a processor called Kookaburra will follow Loon in 2026. This processor will be the first basic module to contain both a logical processing unit and quantum memory.

In 2027, two of these modules will be combined to form a system called Cockatoo. The Starling system will be completed by 2028 using these modules and will be available for cloud-based access in 2029.

Starling will have a capacity of 200 logical qubits and will be able to perform 100 million quantum operations. Then, IBM plans to move on to the 2,000 logical qubit Blue Jay project, which it positions as the final system.

IBM’s quantum processor leader, Matthias Steffen, said that it is not yet clear how many physical qubits will be required, but they are aiming for a structure of several hundred physical qubits for 10 logical qubits.

One of the important advantages of the new architecture is that the number of auxiliary components used in the hardware will decrease as it requires fewer physical qubits. This makes the installation and management of the systems less complex.

However, there are still many engineering problems to be solved on the infrastructure side that will support the architecture. Steffen stated that elements such as connection units and signal boosters are still in the development phase.

Another fundamental problem is reducing gate error rates. In order for the new architecture to work successfully, error rates need to be reduced by a factor of approximately 10. The way to do this is to extend the “coherence time”, which indicates how long qubits can maintain their quantum states.

It was shared that with the latest improvements made to IBM’s Heron chips, this period has been increased from 150 microseconds to 250 microseconds. In isolated test environments, the average coherence time is reported to have reached 2 milliseconds.

Gartner Vice President and analyst Mark Horvath said that the new architecture offers a significant advancement over IBM’s previous quantum technologies. He stated that the increased connectivity and developments in 3D manufacturing processes make the architecture more powerful.

According to Horvath, if IBM reaches the 200 logical qubit goal with this architecture, quantum computers will approach the threshold of solving practical problems. However, he added that reaching the goal may take longer than expected due to the complexity of the modular structure and the difficulties of the engineering process.

IBM will host the Starling system with its new quantum data center to be established in Poughkeepsie, New York. This facility will also be one of the cornerstones of the company’s long-term quantum computing vision.