How would a completely modular phone/tablet/laptop/PC/TV fare?

Modularity is something geeks often dream of. I wonder if it may not be time to realise that dream in response to the rapid proliferation of the various connected gadgets in many of our lives.

The Asus Padfone may perhaps be a good first step in the right direction, but I wonder if it may also be feasible to take it one step further: let customers buy only the brain, a fairly modern SoC complete with basic and crucial connectivity features housed in a practical case. Slot that into a small battery-equipped touchscreen gizmo to get a smartphone. Slot it into a bigger one to get a tablet. Or into other peripherals to get a laptop, or a basic PC (it can automatically switch to eg. Ubuntu-for-Android in those cases). Maybe even an e-reader. You could probably even find a reasonable way to use it as a featurephone.

This sort of a setup comes with the inherent risk of being a huge unbearable hassle (feel free to explore specific issues as well as how they may be addressed). I'd love it, however, because it would let me choose which components to upgrade at any given time. It wouldn't necessarily cut down on my expenses: I imagine I'd be more inclined to buy [i]all[/i] the major accessories, esp. since the total cost would be lower and less scary and be justified by the promise of flexible upgrades.

What I wonder is: would such a system ever really be particularly profitable for manufacturers?

The two most expensive components in modern smartphones and tablets are the screen and the nand storage. Fantastic screens and larger storage may command premiums, but there's only so much customers are willing to pay for a device, and only so often. Would a manufacturer be able to benefit from extreme modularity eg. through greater margins on screens sold separately as accessories? Or from licensing dock connectors to other companies? Or from, er, offloading some of the assembly on consumers? :o

Whatcha reckon? Would near-total modularity be something for you? Might it be interesting to manufacturers? Please help me figure out why this idea might fall flat on its face and what may be required for it to come even close to succeeding :)