Two problems. Firstly, BIP173 (bech32) is relatively fresh (was designed later than segwit itself) so support for it also is very new. Secondly, inertia. Adoption of P2SH-P2WPKH is fairly slow (and it's fully compatible both ways from the start) and bech32/P2WPKH is partly incompatible with old wallets. But bech32 is very nice, much better than base58 standard and pure P2WPKH save additional transaction size compared to P2SH-P2WPKH. Also P2WSH (for pure segwit multi-signature transactions) are more secure that legacy multisignature.  For 160-bit hash one can find a collision in 2^80 operations (it's doable with a large fraction of the bitcoin hashrate within weeks) if one can craft one of the public keys for the multisig. P2WSH are 256-bit and 2^128 operations are infeasible. Eventually, people will upgrade to bech32 but the adoption will not be fast.