At any given point in time, there is a roughly finite amount of demand, and inching closer to meeting that demand can't hurt.
https://news.ycombinator.com/item?id=170719 https://news.ycombinator.com/item?id=860913 https://news.ycombinator.com/item?id=236853 https://news.ycombinator.com/item?id=1255718
This coincides with when tech companies started popping up in SF, instead of South Bay where they traditionally went to.
That's a 25% increase in 3 years and then another 25% increase over the next 9 years. And there was no, or at least only minor, tech company factor in NYC.
For San Francisco, the rate of new unit construction was always too little, but in the 2008 housing crisis, construction stopped and was slow to restart; that would have been a great time to get ahead of demand -- labor and materials were less expensive (because construction was slowed everywhere), but there was risk that the prices would stay low, and difficulties with construction financing.
In transportation, this well-established response is known in various contexts as the Downs-Thomson Paradox, The Pigou-Knight-Downs Paradox or the Lewis-Mogridge Position: a new road may provide motorists with some level of respite from congestion in the short term, but almost all of the benefit from the road will be lost due to increased demand in the longer term.[0]
[0] https://www.citymetric.com/transport/does-building-more-road...
Considering road or housing construction as something you can do once to 'fix' the situation only works if you are able to limit demand. Most metropolitan areas are encouraging of population growth, though. Areas with limited population growth generally can build enough roads and housing to satisfy demand and have a static demand, so there's not a lot of congestion or rapidly rising housing costs.