No kidding
"Microsoft meanwhile would very much like to find a relationship between AI investment and increasing growth "
Well duh, you don't go investing billions in a pipe dream without expecting to get billions more out of it, now do you ?
Except that Redmond has forgotten (or, more generously, not taken into account) a few small details : AI is not holding up to the hype and, more importantly, computer components are looking to vastly increase in cost in 2026 and beyond - especially if Trump abandones Taiwan to China (which is a possibility no serious publication is ignoring at this point).
Given that "datacenters" are now going to have to be built with their own bespoke nuclear power station, the cost of sourcing their components is going to be multiplied by an amount that any Board is going to find difficult to swallow, compared to the revenue they might generate.
AI investment is starting to look a lot riskier given that existing AI does not measure up to expectations, so more of the same that costs several times more to build and maintain (and you might want to avoid building that in Texas - just a suggestion) is not really looking like a winning business case at this point in time.