It will scale better than the traditional web because it is peer to peer and it not reliant on individual servers… (at least 4 copies of any particular piece of data spread throughout the world with many cached copies of popular data near the nodes that are requesting it) The current web needs to engineer in the redundancy and map reduction etc, but SAFE should do it automatically
The biggest hurdle for developers will be the P2P nature in general. For making apps using web technologies, anything that’s a pure client-side framework will work just fine. But all the frameworks for server-side might be totally useless. Designing apps for P2P is something many developers have never done and aren’t used to. Another hurdle will be figuring out ways to make apps usable and popular given the constraint that every bit of data saved must be paid for.