all generalizations are false, including this one.Hint: all major cryptanalytic advances, where governments broke a cypher and general public found out few *decades* later were not of brute-force kind.
most of the WWII advances in computing were to brute-force code engines, not solve them analytically.
but yes - analysis has come a long way, and it is always going to be more cost effective for the NSA to hire mathematical geniuses (at however much it costs) than to build a brute-force cracker at the keysizes available today.
And cheaper still to do an end-run around the crypto and access plaintext on the microsoft-dominated internet.
