With that attitude, eventually the impact of economies of scale will end up denying us the ability to afford those devices or to use them how we see fit.
The problem GP refers to is this: making computers is expensive, gets cheaper with economies of scale. But economies of scale are capital-intensive, so they exist only where there's a market for it. The market for people requiring general-purpose computation is small and getting smaller, while the market for constrained computing is only ever bigger. Which means at some point the capital-wielding companies will just leave the market, and the prices of new general-purpose computers will go way, way up.
On top of that, computers today gain most of their value interacting with other computers. Once the mainstream usage - like your e-mail or your bank - gets neatly packaged in sandboxed, trust-computing-enabled environments, the utility of your general-purpose platform will plummet. A dark but entirely possible scenario is that it eventually becomes illegal to use general-purpose computers to connect to such services, because "security".
I could picture a CPU design that had a boot key or other restricted features set by blowing fuses (or similar tricks) during packaging. The "developer" unit leaves them unblown.
If they're selling 10M "regular" units per year, and 5000 "developer" models over the product lifetime, you have to amortize the cost of special-casing the production line for a short run, and stocking and supporting a second SKU, over a relatively small sales volume. It could well be more expensive than the few cents of electricity per unit to blow the fuses in a standard unit.