Quantum Supremacy: What Google’s Sycamore Experiment Really Proved

01 What “quantum supremacy” was supposed to prove

Google’s 2019 announcement used a deliberately narrow benchmark. Quantum supremacy — a term John Preskill coined in 2011 — means demonstrating a programmable quantum device on a task that no classical computer can complete in feasible time. It does not mean that the machine is generally more useful, more accurate or ready to replace a laptop.

The distinction matters because the benchmark was random-circuit sampling. The experiment generated samples from a probability distribution produced by a quantum circuit; researchers then checked those samples against classical calculations on smaller circuits and statistical tests. The task was chosen to expose quantum control at scale, not to solve a commercial optimization problem.

02 Why qubits are not tiny ordinary bits

A classical bit is 0 or 1. A qubit can be prepared in a coherent superposition of basis states, then entangled with other qubits. Gates rotate and couple those states; measurement returns ordinary classical bits. The advantage is not that a qubit stores an arbitrary list of answers. It is that interference can amplify some amplitudes and cancel others.

Sycamore used superconducting transmon qubits cooled near absolute zero. Microwave pulses implemented single-qubit and two-qubit gates, while the chip’s wiring, packaging and calibration had to preserve coherence long enough to execute a circuit. Every extra qubit adds control challenges and more ways for an error to enter.

03 The experiment Google demonstrated

Google’s team programmed a grid of qubits with layers of one- and two-qubit gates, then measured the output repeatedly. The circuit depth and gate pattern were varied. At lower depths, classical simulation was possible and provided a reference. At the target depth, Google reported that Sycamore produced a million samples in about 200 seconds.

The headline comparison was dramatic: Google estimated that the best classical approach would require roughly 10,000 years for the corresponding calculation. That estimate depended on the chosen circuit, the classical hardware model and the algorithm used to simulate it.

04 The numbers behind the headline

Sycamore’s achievement was a stack of engineering details: 53 qubits, repeated calibration, a chosen circuit family, about 20 cycles in the headline demonstration and a very large sample set. The output distribution is extremely uneven — some bitstrings are much more likely than others — which lets researchers estimate how faithfully the processor followed the intended circuit.

05 Supremacy became a measurement dispute

The word “supremacy” attracted criticism because it can sound like a broad victory over classical computing. IBM researchers challenged Google’s estimate, arguing that a Summit-class supercomputer could simulate the circuit in about 2.5 days rather than 10,000 years. Google’s team responded that the benchmark was designed around a specific best-known classical method and that faster classical simulation did not erase the hardware demonstration.

Both sides illustrate a rule for evaluating quantum claims: the task definition is part of the result. One must specify the circuit, fidelity threshold, sampling method, memory limit, processor, runtime and what counts as a successful verification. A quantum advantage can be real for a carefully selected task while remaining irrelevant to everyday workloads.

06 Why useful quantum computing is harder

Random-circuit sampling is a stress test, not an application. Useful quantum algorithms require logical qubits protected by error correction, long computations, reliable connectivity and a problem whose structure survives noise. Physical qubits are imperfect; increasing their number can increase the error-correction overhead rather than deliver a linear speedup.

That is why the 2019 result should be read as a scientific milestone. It showed that a programmable processor could enter a regime where classical verification becomes difficult. It did not demonstrate a fault-tolerant machine, a new drug, a broken encryption system or a general-purpose quantum computer.

07 The sober meaning of the milestone

Quantum supremacy is best understood as a boundary marker. Before Sycamore, the question was whether a controlled quantum processor could produce a classically hard-to-simulate distribution. After Sycamore, the questions became more exacting: can the result be independently verified, can error rates be reduced, can logical qubits scale, and can a useful algorithm outperform the best classical alternative?