Still I've never heard anybody refer to a 100 mW RF output as a power amplifier weather linear or not.
I would also argue that Class A is really the only linear type since Class B may have both crossover distortion and in some cases a deadband chracteristic. I might also point out that in class C the output is actually a reproduced copy of the input since only a portion of the input is used to kick a resonant circuit. Those all fit my description of non-linear, but that's really semantic hair splitting, and the OP obviously has some purpose in mind which he has not communicated to the forum.
100 mW can generally be produced by modest circuits at that frequency without any undue difficulty. For example an oscillator with 10 mA of output current and a 13.8V supply will have 138 mW of ouput power. A Colpitts or a Heartly with a 2N3904 should do the trick.