Even if we assume that companies are willing to support the systems they sold for an undetermined amount of time (something that will _never_ happen), this still won't solve the fact that the hardware itself enabling the "smart" features is bound to turn obsolete way before the appliance exhausts its lifetime.

For instance, I've seen chest freezers running for >30y without breaking; do you expect today's IT technology to remain viable for the next 30 years? What about new protocols that might appear that the hardware may not support, or new highly desirable features that are simply unfeasible to implement on the limited µcontrollers that power most IoT appliances?

You risk ending up in a situation like with LPT printers - when USB became dominant, lots of old printers ended up in landfills due to how unreliable LPT-USB adapters were. The printers themselves were often absolutely fine, but they had become too annoying and keeping an old machine around to use them was much more complicated than just buying a spanking new network printer.

Given the current dire environmental situation I wouldn't push for technologies that might cause perfectly functioning appliances to become obsolete due to a non-necessary secondary features. The only way out of this is for "IoT modules" to be removable, standardized and upgradeable, but thinking anyone would do that is delusional.