The issue there is that even at that pricepoint, Microsoft is still operating CoPilot at a loss. If they drop it more, they’ll be making even more of a loss. Which is the standard business model for new products these days, but the losses on AI products dwarf things like Netflix and Uber during their “operate at a loss to drive everybody else out of business” phase.

Of course, that would all be fine if CoPilot was some killer product that people quickly found themselves unable to work without. Instead, the feedback shows that workers find that it’s not useful or reliable enough to be worth using, and Microsoft’s own latest advert for CoPilot in Excel contains data which shows that at best operation it doesn’t work 46% of the time, and that figure can be as high as 80%.

I’m not sure these problems are really surmountable - you’ve got an incredibly expensive-to-run product which doesn’t do much that’s useful and is bad at the things that it actually could be useful for. It’s not just Microsoft, it’s the entire tech industry that’s facing this problem.