The point is, they may not need to increase CAPEX. There are a myriad of solutions to help scalability these days, they could rent server time from an ESP (chances are the data centre is run and hosted by one of these anyway), it would likely by on a quarterly basis. All of this can be done dynamically, and often without outages. So they could have extra resources for a short time.
I'll concede that its difficult to do it in the time the servers have been up. Where I work it would take a couple of weeks to get the right change controls in place, but that's just my firm.
My problem is that for the life of me, I can't think of a single thing that should have been a surprise to BW. They should know how many keys have been issued/redeemed etc.. So it looks very much like a planning failure.