"The technology is not useful", at least in enterprise contexts, is what this comes out to. Which is really where the money is, because some vibecoder paying $20/mo for Claude really doesn't matter (especially when it costs $100/mo to run inference for his queries in the first place). Enterprise is the only place this could possibly make money.

Think about it: MS has a giant advantage over every other AI vendor, that they can directly insert the product into the OS and LOB apps without the business needing to onboard a new vendor. This is best case scenario, and by far the easiest sell for these tools. Given how badly they're failing, yeah, turns out orgs just don't see the value in it.

Next year will be interesting too: I suspect a large portion of the meager sales they managed to make will not renew, it'll be a bloodbath.