The second problem is that efficiency is often applied across many small scopes of operation, ignoring the effects on efficiency at greater scope. You can be tactically efficient at the expense of strategic goals.
The third problem is that efficiency is often over-fitted in a chase for maximalism, which necessarily creates a rigid system that is highly dependent on little or no variation between the past and an expected future. Just-in-time supply chains fell apart due to things like the covid-19 lockdowns, which often cascaded across both vertical levels of specific industries as well as across industries.
From a system design point of view, building a system that has no flexibility in scaling but is instead maximally efficient based on current (and recent past) levels of demand is a disaster waiting to happen. In the old days we'd laugh at a commercial site that got slashdotted because it was incapable of handling burst demand.