New Quantum Chips Do in 5 Minutes What Would Take 10 Septillion Years. Here's the Catch

In December 2024, Google's Quantum AI team announced a new chip called Willow. It solved a standard benchmark problem, a specific kind of random circuit sampling task, in under five minutes. Google calculated that the world's fastest classical supercomputer would need 10 septillion years to do the same thing, a number written out as a 1 followed by 25 zeros, far longer than the universe has existed.

Willow's real achievement was narrower and, in its own way, more significant than the headline number. For nearly 30 years, quantum researchers had chased something called below-threshold error correction, the ability to add more qubits to a system and have errors go down instead of up. Willow was the first hardware to demonstrate it. Hartmut Neven, who leads Google's Quantum AI lab, wrote in the announcement that the result even lends credence to the idea that we live in a multiverse, since a benchmark that hard is easiest to explain if the computation is somehow happening across parallel universes at once.

That's Google's own interpretive framing, not a settled conclusion. Plenty of physicists studying the same result explain it without invoking a multiverse at all.

There's a reason to be cautious about the headline number specifically, and it has a precedent. When Google made a similar beyond-classical claim in 2019 with an earlier chip called Sycamore, competitors spent the following months narrowing the gap with better classical simulation techniques, cutting what was first described as a 10,000-year classical runtime down to a matter of days. Neven addressed this directly in the Willow announcement, writing that he expects classical computers to keep improving on the new benchmark too, even as the underlying quantum hardware keeps pulling further ahead over time.

What Willow did not do gets less attention than what it did. Neven himself acknowledged that the scientific simulations run on the chip so far remain within reach of classical computers. The next milestones on Google's own roadmap, a long-lived logical qubit and a working logical gate between two of them, have not been publicly demonstrated. Google has not named a successor chip with published specifications either.

Google was not the only company making noise. IBM unveiled Nighthawk, a 120-qubit processor, in late 2025, alongside Loon, an experimental chip meant to prove out the hardware pieces fault tolerance will eventually require. IBM's stated target is specific and close: quantum advantage, meaning a real, commercially useful computation a classical machine genuinely cannot match, by the end of 2026, close enough now to actually check.

A third approach entered the race too. Microsoft, Atom Computing, and QuEra are all building with neutral atoms instead of the superconducting qubits Google and IBM use. Atom Computing reported entangling 24 logical qubits built from 112 physical ones this year. QuEra has described a target it calls the Teraquop regime, roughly one error per trillion logical operations, as a plausible near-term goal.

This has also become explicitly geopolitical. What was a Google announcement in December 2024 is now a multi-country, multi-approach race with several separate groups claiming they're close to something significant, on similar timelines.

The reason any of this matters beyond the lab is a date security researchers call Q-Day, the point at which a quantum computer becomes capable of breaking the encryption that currently protects most digital information, the RSA-based systems underlying online banking, medical records, and government communications. Google's own internal target for that milestone moved from 2030 to 2029 after Willow. That's a real, if modest, acceleration, by the company's own account.

What hasn't moved is the number the people actually planning for this risk are using. Independent security researchers, the ones advising governments and companies on when to migrate to quantum-resistant encryption, have kept their central Q-Day estimate at 2033 to 2035 through Willow, through Nighthawk, and through every claim aimed at 2026. Two genuine hardware breakthroughs and a wave of competing announcements, and the number that determines when this actually becomes everyone's problem hasn't changed.

IBM has made claims like this before and adjusted course when the timeline slipped, so the industry has some recent practice watching a specific quantum date arrive and quietly move. Whether Nighthawk actually delivers a checkable quantum advantage before the end of this year is one of the more testable predictions in technology right now, worth watching resolve in real time rather than taking on faith.

That's the catch sitting underneath the impressive benchmark numbers. A chip beating the age of the universe on a narrow, specifically chosen problem is a real technical result. Whether it moves the date anyone should actually worry about is a different question, and so far, the people whose job is answering it are saying no.