IBM unveils experimental ‘Loon’ quantum computing chip

• The approach reduces the burden on purely quantum error-correction codes but requires more complex chip designs, with qubits supplemented by additional quantum interconnects.
• IBM believes Nighthawk chip could outperform classical computers on select tasks by the end of next year.
• To accelerate validation, IBM is collaborating with startups and academic researchers to openly share code and benchmarks.

IBM announced “Loon,” an experimental quantum computing chip that the company says marks a key step toward building useful, error-managed quantum computers before the end of the decade.

Quantum computers promise to tackle problems that would take classical systems thousands of years to solve, but fragile quantum states make today’s machines highly error-prone.

Tech giants including Alphabet’s Google and Amazon are racing alongside IBM to tame those errors and demonstrate quantum advantage in real-world tasks.

IBM has been pursuing a hybrid error-correction strategy it proposed in 2021, adapting algorithms originally designed to improve cellular signals and running them across both quantum processors and classical chips.

The approach reduces the burden on purely quantum error-correction codes but requires more complex chip designs, with qubits supplemented by additional quantum interconnects.

Quantum advantage

Jay Gambetta, director of IBM Research and an IBM Fellow, said access to the Albany NanoTech Complex in New York—equipped with tools on par with leading-edge semiconductor fabs—was critical to integrating the new quantum connections into Loon’s architecture.

The company did not disclose when external users will be able to test Loon, which remains in early stages. The company also unveiled “Nighthawk,” a separate chip slated to be available by year-end. IBM believes Nighthawk could outperform classical computers on select tasks by the end of next year.

To accelerate validation, IBM is collaborating with startups and academic researchers to openly share code and benchmarks.

“We’re confident there’ll be many examples of quantum advantage,” Gambetta said. “But let’s take it out of headlines and papers and actually make a community where you submit your code, and the community tests things, and they select out which ones are the right ones.”