When Google published its quantum supremacy claim in Nature in October 2019, the paper had already leaked, IBM had already objected, and the argument about what the result actually meant was already underway.
The claim was specific. Google's Sycamore processor, a 54-qubit quantum chip, had performed a particular calculation in 200 seconds. The same calculation, Google estimated, would take the world's most powerful classical supercomputer approximately 10,000 years. That gap, between 200 seconds and 10,000 years, was the supremacy being claimed.
IBM published a response almost immediately. Their argument was that the 10,000 year estimate was wrong because it did not account for optimisations possible on classical hardware. IBM's estimate was around two and a half days, not 10,000 years. Still much slower than Sycamore, but a very different frame for the achievement. IBM also pushed back on the term quantum supremacy itself, arguing it was misleading because the specific calculation Sycamore had performed had no practical application. It was designed to be easy for a quantum computer and hard for a classical one. That is not the same as solving a useful problem faster.
Both of these things could be true simultaneously. Google's result was a genuine technical milestone demonstrating a quantum processor performing a task beyond the practical reach of classical computing at that moment. IBM's critique was also correct that the framing obscured as much as it revealed.
What quantum supremacy on a contrived benchmark actually demonstrates is that the underlying physics is working. Quantum processors are performing quantum operations with enough fidelity that the results are valid. That is important because maintaining quantum coherence at scale is extraordinarily difficult, and demonstrating that the error rates are manageable enough to outperform classical systems even on a narrow task is a genuine step.
What it does not demonstrate is a path to breaking encryption, accelerating drug discovery, or any of the other applications that make quantum computing exciting in a practical sense. Those require many more qubits with much lower error rates and fundamentally different algorithms.
October 2019 was a milestone. How significant a milestone depends on how much distance there is to everything that matters.