>why don't they just under promise and over achieve
I can answer this one!
Because when you underpromise, you are undervalued, and if you are undervalued, then you often won't receive needed investment/trust/means/etc. Meaning that while underpromising and overdelivering is a GREAT plan for an individual most of the time, if you do it habitually as a company, you can wedge yourself into a situation where, for example, your product is THE BEST, but you've underpromised enough that you only get 15 million bucks to hire new engineers, buy up hardware/cloud virtualization space, etc., and that isn't enough - you run out of money early and go broke! Where your competitor, who overpromised, got 30 million in investment and while their product wasn't intially as good as yours, they invest time and money into it, and then they just buy you and your team at a fire-sale price when you go broke.
The market is inept at these optimization functions overall, but it is still tuned to TRY and get them right- to take promises, expected performance, indicators, and all that and rough out a risk vs. reward equation and then plug funds into it. Underpromising and overdelivering is a method by which you can sometimes 'hack' the power of the market to 'price' your efforts, but it can backfire hilariously if you use it in the wrong space, against the wrong opponents.