[ikalugin]

Most stealth aircraft (ie F117A, F22A, JSF series, B2A to lesser extend) have shaping features that are either equal in size or smaller to the wavelength, but are not significantly larger than it.
This leads to a such interaction between the radiowave and the aircraft that the shaping features do not actually matter, the wavelength (by enlarge) ignores those.
RAM coatings wise, you either make those broadband and require RAM coatings/structures thickness to be on par or thicker than the wavelength, or you create narrowband RAM coastings/structures. This is why RAM coatings/structures are defeated by even shorter wavelengh radars of L-band (which also partially defeats shaping).

JSF is too big to fail, even if it's stealth (which comes from way back) is no longer a decisive factor. A purpose built, low level flight platform is another matter however, even though you could fit an L-band radar that could detect 0.005m2 RCS at 600km onto a plane.

In my opinion stealth sort of lost it's thunder back in 1987, with S300V2 (which had mobile long wavelength radar for target detection, surprise) coming around (unless you believe in the 0.00000000000001m2 RCS figures given for F117A and F22A) and rendered tactically (via fully mobile VHF/UHF and other radars) and strategically (via fielding of new long range means - such as beyound horizon radars) irrelevant, especially at high altitudes, as current airborne radar technology is still some ways behind the ground based stuff.