Amazon’s big roar…or was it just a meow?

As with every Big Tech announcement in an emerging field, the tech ecosystem has been buzzing since Amazon’s latest article unveiling its brand-new Ocelot quantum chip. Releases from major players — Google, Microsoft, IBM, and others — are becoming increasingly frequent in this space, which is good news — but as always, buzz comes with noise. So, we’d like to offer our modest contribution to decrypting this undoubtedly significant news.

In short, Ocelot is a prototype quantum chip designed to implement bosonic quantum error correction (QEC) using cat qubits. QEC is a vital piece of the puzzle for achieving useful, scalable quantum computers. Every breakthrough in this area is closely scrutinized, making it important to understand what truly changes this time.

By leveraging cat qubits — a special type of superconducting qubit strongly protected against bit-flip errors (one of the two types of quantum errors) — AWS demonstrates efficient QEC with far fewer qubits. Specifically, Ocelot requires five times fewer qubits than Google’s latest chip, which is based on a surface code architecture, to achieve a similar level of redundancy — an indicator of error correction efficiency. Such results indicate that this approach could, at scale, significantly reduce the resource overhead required for QEC, making it a remarkable step toward scalable quantum computers.

Yet, AWS is far from the first to explore bosonic QEC — this field has been under development for more than two decades, with early theoretical works dating back to 1999 (such as the GKP code). Moreover, breakthroughs in QEC are becoming more frequent as the quantum race accelerates. Google’s Willow chip recently made headlines for demonstrating exponential suppression of errors on a 105-qubit processor.

And France is not to be outdone in the race — let’s not miss the chance for a little patriotism. While several French companies are leading the charge across multiple qubit platforms, it’s impossible to discuss AWS’s announcement without mentioning our national cat qubit champion, Alice & Bob. Armed with a technological roadmap that inspired AWS, the 120-employee company based in Paris is a key player in bosonic QEC. They are also a pure player in cat qubits: while Ocelot hybridizes cat qubits with transmons — the most common and better-mastered type of superconducting qubits — Alice & Bob bets on an all-cat qubits architecture, which is less mature as of today but far more efficient than the hybrid approach. Recent Alice & Bob breakthroughs include a technique that reduces the number of required physical qubits by a factor of 200 compared to surface code architectures and the release of the world’s first cloud-accessible cat qubits chip, achieving outstanding performance with a single qubit.

Finally, let’s not mistake an important prototype for an industrialized, deployable product. AWS’s claim that Ocelot could reduce qubit overhead by up to 90% is certainly intriguing and a promising step toward hardware-efficient quantum computers, but it remains to be seen whether this approach can truly scale — and how long that will take. Indeed, moving from prototype to fault-tolerant machine demands major advances in hardware stability, control precision, and qubit connectivity.

As John Preskill, one of the founding fathers of bosonic QEC, wisely put it:

And Laurent Prost, product manager at Alice & Bob, adds:

A fitting reminder that, while using AWS’s announcement to pinpoint exactly how much closer we are to FTQC may be a risky leap, every step forward matters — and this one is certainly a big one.