The point is that, for 99.9% of all use cases, a quantum computer is going to be vastly slower than an "equivalent" classical computer. Think of it as the computing equivalent of an idiot savant. Sure it'll factor numbers exponentially faster, but it's not like it'll improve the graphics of a video game.

Assuming it's even possible to mass produce miniaturized room temperature quantum co-processors suitable for consumer devices, it's not currently obvious (to me, at least) what that might be useful for. I'm not saying we won't ever find a use case, but I just don't know what that would be. Especially when we weigh the pros and cons against a cloud compute solution. My best guess is that there might be potential applications related to military/weapons technology and/or space travel, or other contexts without fast and reliable Internet access, but that's just speculation on my part.