I think the point was that the big DBs have better ways to deal with scaling than sharding.