There are strategies that play through late-game. Because resources are finite in StarCraft, after a certain point your income is always zero. At that point if one player has "money in the bank" they're actually ahead not behind, because they can build more units in response to their opponent's current composition which can't be countered.
Ordinarily for example if you're crushing a Protoss and they build a Dark Shrine as a last ditch, no problem get detection and shut DTs down before they cause trouble. But if you have no income and nothing left in the bank then you may just never be able to detect those DTs with what you have, and no way to build a detector. Now you're in a race - win before the DTs kill you.
There are some crazier corner cases, if you lose all workers and can't afford to build more then spending everything also becomes a bad idea for a similar reason and spending down to zero actually increases your risk of not being able to buy workers if somehow you lose them all.
I'm not against goog, appl, ect building warchests, but the analogy does break down.
I'd say the situation is more akin to a Free-for-All situation in a game like Age of Empires 2, where the winning strategy is generally to take control of key resources and suck them dry while fighting as little as possible. And then when you need to fight someone during a tech switch, you bury them with your accumulated bank.
Elon's Starlink broadband provider via very-near-earth satellites could be leveraged to provide a real-world advantage against google's internet dominance.
If that conflict happened google would have to dump money into blue origin and would that be enough to counteract the advantage Starlink would have in dumping it's profits into getting more advantage?
Idk, but just making the point
Cost of failure is a huge decider of optimal behavior.
Corporations are limited liability. It is expected that shareholders have distributed ownership in many companies.
For Starcraft rankings, the possibilities are -1 or +1. For investing, it's more like -1 to infinity. It's a distribution within those bounds. Increasing the likelihood of hitting that -1 roll of the dice is completely acceptable if it is counter-balanced by a greater average benefit at the long end of the tail.
Starcraft is different. Because there are only 2 outcomes (excepting draws), an increase in the chance of the -1 outcome necessarily comes with a decrease in the change of the +1 outcome. It would be more like a public company if they gave ranking bonuses for overkill of the opponent.
I wonder if Facebook is following that approach. Ie, maximizing growth and planning to just barely survive legal scrutiny
Furthermore, if you're already at the unit cap, and all your building are already working at peak efficiency, there may be no good way of spending those resources without wasting them.