Imagine you're packing boxes into a shipping container. You suspect that neatly stacked rectangular boxes — the Gaussian approach — use space optimally, but you can't rule out that some exotic irregular shape might fit more cargo. For fifty years, optical communication theory has faced exactly this uncertainty: everyone assumed Gaussian signal states were optimal for sending classical bits through noisy optical fibers, but no one could prove that some cleverly entangled or non-Gaussian strategy wouldn't sneak more information through the pipe. Adesso, Lami, and Giovannetti settle this definitively. Their committed claim: for every single-mode bosonic Gaussian channel — the standard mathematical model covering fiber-optic loss, thermal noise, amplification, and their combinations — independently prepared Gaussian states with Gaussian-distributed random displacements achieve the maximum classical information rate. No entanglement between channel uses helps. No non-Gaussian tricks help. The capacity is exactly the Gaussian Holevo information, period. The core technical move is elegant. The authors connect an entanglement inequality (from their companion paper arXiv:2608.01909) to an additive lower bound on output entropy. Additivity of the Holevo information across channel uses follows — this is what prior work couldn't nail down. The result holds at all input energies, which matters because the low-energy regime is where squeezed states could theoretically compete with displacement-based encodings, making the proof harder than the high-energy limit where Gaussian optimality was already suspected. Beyond the capacity formula itself, the paper delivers two sharp results about communication above capacity. Under a strict photon-number cutoff, the probability of correct decoding decays exponentially with channel uses, and they give explicit bounds on the exponent. Under a mean-photon constraint, a clean energy-rescaling formula pins down the entire asymptotic rate-error tradeoff: capacity at mean photon number N with allowed error ε equals the zero-error capacity at mean photon number N/(1−ε). This is unusually tidy for an information-theoretic result — it converts the entire error-rate curve into a single parameter shift. The integrity profile is pure mathematics: the results are proven, not simulated. The validation is a formal proof, which is the gold standard in this subfield. There is no cherry-picking risk because the claim covers every single-mode bosonic Gaussian channel — it is universal within its scope. The companion entanglement inequality paper provides the key lemma, so the result's load-bearing structure spans two papers, but both are by the same authors and available as preprints. Where does this sit on the ladder? The Gaussian optimality conjecture — specifically, the minimum output entropy conjecture and the additivity of the Holevo information — has been the central open problem in bosonic quantum information theory. Giovannetti, Holevo, and García-Patrón proved major special cases over the last two decades, but full generality (all channels, all energies) remained open. This paper closes it. Classical channel capacity formulas from Shannon's era assumed Gaussian noise optimality by fiat; this work proves the quantum analogue rigorously. For practitioners designing optical communication systems — submarine cables, satellite links, deep-space optical comms — this result means the engineering targets won't shift. The capacity formulas already used in system design are now proven tight. No future quantum encoding trick will beat them. The practical ceiling is confirmed, and the focus can stay entirely on approaching it through better codes, modulation, and detection.
