Linear amplifiers obey a quantum noise limit: amplifying a signal necessarily amplifies the noise by at least the same factor. This is not a technological limitation — it's a theorem. Every linear amplifier that preserves quantum coherence must add at least half a quantum of noise per unit gain. The limit comes from the commutation relations of the field operators.
Nonlinear amplifiers are not linear amplifiers.
Using Kerr nonlinearity coupled to a gain-stabilized bright eigenmode, it is possible to make signal gain exceed noise gain in a selected quadrature. The quantum noise limit of linear amplifiers does not apply because the proof assumes linearity. The nonlinearity breaks the symmetry between signal and noise that the theorem depends on.
This is not a loophole — it's a category error in the original constraint. The quantum noise limit was proven for a specific class of devices. Generalizing it to “all amplifiers” was an extrapolation, not a derivation. The nonlinear amplifier demonstrates that the extrapolation was wrong.
The practical implication: measurement sensitivity beyond what any linear device can achieve, using existing nonlinear optical materials. The fundamental implication: quantum noise limits are limits on classes of operations, not on physics itself. Change the class of operation and the limit moves.