That's debatable - all the actual figures from the real construction that was actually happening were within the estimates provided, and they had factored in efficiency savings that were being made along the way (as has been shown to happen in other rollouts internationally).
It wasn't 'massively understated' until Malcolm Turnbull's theoretical 'Strategic Review' that estimated that the cost was going to suddenly double because of [redacted] (the alleged reasons that the previous downward trend suddenly stops and then jumps up is blacked out in the report), as well as things like inflating the peak funding because they counted the full cost but stopped counting future revenue after something like 2022 (adding about $12bn to the peak funding), and ignore like $6bn of savings from design changes that had already been signed off.
It was all pretty dodgy, but still with all the fudging only came to $56bn (instead of the old $44bn - the $14bn additional to the Government spend was to be private equity).
The second paragraph of your post is a misconception. They required a $90 per month ARPU - when you factor in business use (many who would be paying in the region of $400 - $2000 because of higher traffic classes, lower contended CVC, business SLAs, etc), home users would only need to average about $60-70 a month (that's more than I pay for ADSL that doesn't hit 4Mb/s), and that's an average which is skewed by the power users (as per the Pareto principle).
So the average household would only be paying $30 to $60 a month, which is more than competitive with ADSL when you factor in Telstra line rental. And this is across all services, not necessarily just internet, so add in phone and pay TV, which people already pay $100pm for, and it's easily conceivable that prices would have dropped.
Even now, if I were in an NBN area I could have 100Mb/s internet and VoIP for just $10 more than I pay for my slow ADSL2+ with phone (I'm on the $79p/m iiNet bundle)