New quantum computer outperforms Google’s previous record by 100-fold

Quantinuum, a quantum computing startup backed by JP Morgan Chase, has claimed the crown of quantum supremacy from Google’s Sycamore machine, which was introduced five years ago. The new system, with 56 trapped-ion qubits, demonstrated a 100-fold better score.

The startup claims that its quantum computer is already impossible to emulate with classical supercomputers, “as it would take up the entire memory of the world’s best supercomputers.”

In an industry benchmark called the Random Circuit Sampling (RCS) algorithm, Quantinuum claims to have achieved 100 times improvement over the previous record holder, Google’s Sycamore quantum processor with 53 qubits. Quantinuum’s new 56-qubit computer, called H2-1, ran simulations without a single error about 35% of the time, while Google’s – 0.2% of the time.

The startup also claims that a quantum computer uses 30,000 times less power for the same work compared to classical supercomputers.

“Thanks to a run of recent achievements, we have helped to lead the entire quantum computing industry into a new, post-classical era,” Quantinuum said in a press release.

The RCS algorithm doesn’t seem to have practical applications. It’s used to demonstrate “quantum supremacy” by performing computations in a specific way that’s significantly faster on quantum computers than on classical ones.

Google first made this test famous in 2019. The tech giant claimed that a classical state-of-the-art computer would take 10,000 years to complete the same task, which took its Sycamore just 200 seconds. IBM later countered those arguments by claiming that the task could be performed on a classical system in 2.5 days and with far greater fidelity (accuracy).

Quantinuum says that its new 56-qubit computer, called H2-1, is now impossible for a classical computer to fully simulate, “without any loss of the fidelity, all-to-all qubit connectivity, mid-circuit measurement, qubit reuse, and feed-forward.”

“We have been able to run circuits on all 56 qubits in H2-1 that are deep enough to challenge high-fidelity classical simulation while achieving an estimated XEB (cross entropy benchmark) score of ~0.35. This >100x improvement implies the following: even for circuits large and complex enough to frustrate all known classical simulation methods, the H2 quantum computer produces results without making even a single error about 35% of the time,” the company said.

In April, the startup, in collaboration with Microsoft, achieved another milestone by demonstrating four reliable logical qubits, achieving an 800-fold reduction in error rates on a smaller 32-qubit machine.

Quantinuum focuses on “highly reliable” and error-corrected qubits, instead of chasing large numbers of physical qubits.

The company hopes to create more logical qubits with even lower error rates from rising numbers of physical qubits.

A qubit is a basic unit of information in quantum computing, representing a superposition of two possible states at the same time instead of 1 or 0 in classical computing. The power of quantum computers increases exponentially with each additional qubit, so they are expected to revolutionize intensive workloads such as simulating molecules and material structures, cryptography, financial simulation, and others.

“As our quantum computers scale beyond today’s 56 qubits to hundreds, thousands, and eventually millions of high-quality qubits, classical supercomputers will quickly fall behind,” Quantinuum expects.

The paper suggests that a combination of more qubits, lower physical error rates, higher connectivity, and faster clock speed are needed for quantum computers to overtake the classical ones.