Johanna Rothman writed about "letting go of BDUF" (Big Design Up Front). One statement she makes is that you can't predict what the architecture needs to be. I can't say I agree with that, since many times you do know something about the project and you do have prior expereice that can give you enough confidence to lay some ground rules. I called this Just Enough Design Up Front or JEDUF for short.

Another statement Johanna made, which I wholeheartedly agree with, is that you should let the architecture evolve. Evolving an architecture sounds very compelling but it is not a simple feat. Architectural decisions tend to have system wide implications which means that changing one too late in the game you'd get a lot of rewrite and/or refactoring to do
.
My strategy to solve that conflict is to:

1. Set the first one or two iterations as architectural ones. Some of the work in these iterations is to spike technological and architectural risk. Nevertheless most of architectural iterations are still about delivering business value and user stories. The difference is that the prioritization of the requirements is also done based on technical risks and not just business ones. By the way, when you write quality attribute requirements as scenarios makes them usable as user stories helps customers understand their business value.
2. Try to think about prior experience to produce the baseline architecture
3. One of the quality attributes that you should bring into the table is flexibility - but be weary of putting too much effort into building this flexibility in
4. Don't try to implement architectural components thoroughly - it is enough to run a thin thread through them and expand then when the need arise. Sometimes it is even enough just to identify them as possible future extensions.
5. Try to postpone architectural decisions to the last responsible moment. However, when that moment comes -make the decision. try to validate the architectural decisions by spiking them out before you introduce them into the project

These steps don't promise that the initial architecture sticks, but in my experience it makes it possible to minimize the number of architectural decisions but still have a relatively solid foundation to base your project on