Yes. Physicists have only proved two fundamentally different features of quantum mechanics and Einsteinian mechanics such that the latter cannot reproduce the former: quantum non-locality and quantum advantage. Clearly, we do not see quantum effects all around us on macroscopic scales, and Einstein’s GR is not even mathematically compatible with QM, so QM must somehow converge to the classical picture on larger scales. We have tested quantum non-locality over large, macroscopic distances, so the only thing left that could possibly break down on macroscopic scales is quantum advantage.

This fact people all gloss over seems to call quantum computing into question, and has led to the proposal of several models, like Tim Palmer’s recent rational quantum mechanics, post-quantum gravity, and historically some other models like Ghirardi-Rimini-Weber theory, where quantum advantage is fundamentally not scalable, because this creates a natural convergence between quantum theory on microscopic scales and classical theories on macroscopic scales.

These models are inspired by decoherence, which shows that environmental noise, if you cannot track the environment, leads the system to appear like it acquires classical dynamics. Decoherence fails to actually explain the emergence of the macroscopic world, however, because macroscopic phenomenon, like gravity, do not attach to subjective “appearances” but attach themselves to real physical properties of the system, given by the stress-energy tensor.

However, if you presuppose that there exists some noise fundamental to the dynamics itself, not simply in the environment, then it causes a quantum to classical convergence naturally. The consequence, however, is that quantum advantage is not scalable. As you try to entangle more and more qubits, you’d predict it’d also introduce more noise, and so you’d need more error-correcting qubits. This may be manageable for some time, but eventually there’d be so much noise that adding an additional qubit would require so many noise-correcting qubits that it would outweigh any possible quantum advantage.

Tim Palmer’s model for example makes a very explicit quantitative prediction that 1000 noiseless qubits will entirely saturate any advantage you could get from a quantum computer. The “noise” in his model is not fundamental but an emergent feature of an assumption that the Bloch sphere of a qubit is discretized, and since it is discrete, it is finite and can be saturated.

There is good reason to not see quantum advantage as an actual proven technology, because it is based upon a theory which we already know should break down on large scales, and the only way that hasn’t been tested that it could break down is in terms of quantum advantage. The only other possibility is that our understanding of gravity is totally wrong and so QM never actually breaks down, but the only consistent mathematical formulation of that approach is String Theory, but String Theory has been empirically ruled out, because it needs an anti-de Sitter space to work, but cosmologists have empirically measured our universe to be a de Sitter space, so String Theory is no longer a possibility.

These claims of thousands of qubits in quantum computers are never thousands of effectively noiseless qubits that could be used for universal quantum computation. There is always some huge caveat and so in practice they never actually show a real advantage.