20nm is going to cost significantly more than 28nm for at least a couple years based on NVIDIA's projections here, and the production lines will be busy fulfilling orders for these semiconductor giants, not a tiny company like Avalon regardless of whether or not they receive this alleged $200 M investment. Avalon's 20nm chips would be at least one year away if not more; they would be absurdly expensive and they would not be able to compete on price with the 3 contenders for the 28nm market who will have much lower NRE and R&D costs as well as getting to market faster.
First - that article was published in the Monday edition of the WSJ. It could be wrong, but boy, somebody is getting fired.
If you look at the first graph you linked, 20nm will reach price breakeven with 28nm on a per-transistor basis by 2014 Q3. It might be a strategic move - pay a premium right now for enough hardware to be competitive and get your design issues sorted out (while still being way cheaper than the other guys' gen 1 chips, ASICMiner in particular) - and then once 20nm is really cost-competitive, you can flatten the other guys completely by ramping up production.
That way you won't have to pay the NRE for 20nm later on.