The current production of 1.9 billion tons of steel per year is something you consider insufficient?

I don't know how much steel we need per square meter of PV (e.g. frames can be made from wood), but I do know the area we need for the current global electrical demand of 2 TW even after accounting for capacity factor and not just cell efficiency, and that our current production in each year is sufficient to put a contiguous 2 mm layer behind all of it:

http://www.wolframalpha.com/input/?i=%281.9e9%20tons%20%2F%2...

Given the panels are supposed to last 25 years, even at steady-state replacement rates, and assuming zero growth in the steel sector, and assuming none of that steel gets recycled when the cells themselves need refurbishment or replacement, that doesn't seem to be a real problem to me.