Google LLC today debuted Willow, its latest state-of-the-art quantum computing chip with novel capabilities that the company says solves a key challenge with error correction at large scale.

Quantum computers are important because they have the potential to solve complex problems that could be impossible for classical computing to perform. In the near term, they can solve problems quickly, which would take traditional computers an extremely long time. As a result, they could be used to open up new opportunities in drug discovery, materials science, financial modeling, weather prediction and artificial intelligence.

However, quantum computers have a problem: errors. Qubits, the units of computation in them, tend to lose their information because they can be affected by the environment around them. This makes it very difficult to complete computations. Generally, the more qubits used, the higher the chance that more errors will happen.

Researchers at Google Quantum AI said that with advances in Willow’s architecture, they were able to cut error rates by half each time the number of qubits was increased. The results were published today in a paper in Nature, revealing that when testing ever-larger qubit grids of physical qubits from 3×3 to 5×5 to 7×7 revealed higher error correction rates.

“In other words, we achieved an exponential reduction in the error rate,” said Dr. Hartmut Neven, founder and lead of Google Quantum AI. “This historic accomplishment is known in the field as ‘below threshold’ — being able to drive errors down while scaling up the number of qubits.”

By grouping multiple physical qubits, this allows software to create what is known as a logical qubit. Logical qubits can be used to act as protection against errors to ensure more reliable quantum computations by assisting with error correction. Neven explained that the ability to scale up logical qubits while driving errors down demonstrates significant progress.

This is a big step from 2023 when Google demonstrated the previous milestone breakthrough when it said it had reached the “break-even point” in quantum error correction. Now that Willow has shown an exponential reduction in error rates, Google says it’s confident that its a prototype for a scalable model for running commercially relevant applications that can’t be easily completed by traditional computers.

Beating today’s supercomputers in complex problem performance

Minted in Google’s facility in Santa Barbara, a facility purpose-built for building quantum computing chips, Willow houses 105 qubits, almost double the capacity of the company’s previous chip Sycamore, released in 2019 with 53 qubits.

At the time, Google reported results for Sycamore using the random circuit sampling benchmark, a performance measurement designed to be intractable for classical computers now widely used in the field to test quantum computers. Sycamore completed the task in about three minutes and 20 seconds. Google claimed that the world’s fastest supercomputer at the time, the 200-petaflop Summit system operated by the U.S. Energy Department, would have taken 10,000 years.

According to Google, Willow has completely changed the game on these metrics by performing another computation in under five minutes that would take Frontier, the world’s second fastest supercomputer, 10 septillion years to complete – or written another way, a one with 25 zeros after it. This number exceeds even the age of the universe.

Neven emphasized that RCS isn’t useful for practical applications. “It is an entry point,” he said. “If you can’t win on random circuit sampling you can’t win on any other algorithm either. The next task is to train this massive compute power to a task the mainstream would care for.”

Another metric Google said Willow showed great leaps in performance with is qubit coherence – or how long individual bits retain information. This key resource allows for longer computation times, permitting more complex simulations to happen. According to Google, Willow approaches 100 microseconds, an impressive five times improvement over the previous generation.

“The next immediate challenge for us and the field is to demonstrate a first ‘useful, beyond-classical’ computation on today’s quantum chips that is relevant to a real-world application,” said Neven. “We’re optimistic that the Willow generation of chips can help us achieve this goal.”

Image: Google