Quantum information supremacy - so what?
A recent preprint has got people excited...
Excitement about quantum technology, and quantum computing in particular has reached fever pitch in 2025, the International Year of Quantum. Every announcement and press release on the topic seems to generate hype about how the age of “quantum supremacy” has either arrived or is imminent. By trying to put these things into perspective, I often come across as a sceptic, but I think of myself as more of a realist. Someone else who works hard to oppose irresponsible claims about near-term practical applications of quantum computing is Scott Aaronson at the University of Texas.
Therefore, it’s a bit ironic that a paper he co-authored with other University of Texas researchers on “Demonstrating an unconditional separation between quantum and classical information resources” seems to have generated its own share of over-excitement and hype. Commentary has included claims that it is some sort of “amazing result”, “shows that quantum supremacy has arrived” and somehow has accelerated the likely timeline to quantum computer breaking RSA2048 encryption. (Spoiler alert - none of this is true).
To give due credit to the authors, they have done some ground-breaking work to find a computational problem where they can prove - unambiguously using mathematical theory - that the number of qubits required to solve the problem using a quantum computer is less than the number of bits required to solve it using a traditional computer. This proof doesn’t rely on general empirical observations of running algorithms on traditional computers, and since it uses some clever maths known as “information theory” the advantage of number of qubits vs number will still hold even if people come up with better algorithms to run on traditional computers. They also proved that they can use one of today’s noisy, imperfect quantum computers to actually perform such a calculation at a small scale. The particular hardware they used was a Quantinuum quantum computer which can achieve accuracy of over 99.9% for two-qubit operations. This level of accuracy was key to the success of their calculation, and something that wasn’t achievable a couple of years ago.
However, this narrow claim of “quantum information supremacy” (ie you need less qubits than bits to solve a problem) hasn’t demonstrated any real economically relevant “quantum advantage”. As the authors admit, even if they’ve proved you need less qubits, there is no proof that it takes the same, or less number of steps that the classical computer; even before we account for the likely slower clock rates of quantum computers, let alone the likely massively greater costs of qubits compared to bits. And that’s before whether we debate whether performing this particular computation has any value to anyone, other than as a contrived construct to find a way to prove an advantage of quantum computing. When I reached out to Scott, he said “we made no claim in our paper that our quantum information supremacy task has any economic value whatsoever, and I’d oppose anyone saying that it did.”
In my view, comparing qubit count to bit count is a bit like comparing apples to oranges. There is no equivalence between the two in terms of speed, cost, power consumption or anything else. Intuitively, we would expect a qubit which can encode a continuously varying complex number to be able to store more information than a standard bit that can only encode one of two states (0 or 1). The fact that this can be proven even in the presence of noise shows that attempts to build a quantum computer are not a total dead end, which is great news. The accuracy rate for 2-bit operations was vital, and shows significant and critical progress from what was achievable a couple of years back. However, it’s far from the “game changing transformation” that some have suggested.
It also has zero relevance to any attempt to use a quantum computer to run Shor’s algorithm to crack modern public key encryption scheme (where you may wish to note even the latest advances suggest a million qubits will be needed to calculate a 2048-bit value - so this “information advantage” is far from universal). In fact, I’d suggest anyone claiming a link is delusional, trying to frighten you, or trying to sell you something.
Just to be clear, however, organisations should worry about migrating to quantum-resistant cryptography, and making plans to do so. However, there is no imminent “Q-Day” when everything will be instantly broken; instead priority should be given to starting upgrades with long lead times, and to systems that transport highly sensitive data across networks, where that data may need to be protected for long periods. But please let’s stop the FUD and hype - before we create inflated expectations and rapid disillusionment with the prospects for what quantum computing can offer us in the long term.