But the question remains. An independent GPS unit capable of cm accuracy and costing less than (for example) $500 would command an enormous market, so why isn't it here already?
I don't think the market is that enormous. For people around here, sure, I'm sure you could sell quite a few. But for the guy that wants it for something like car navigation no; like RH314 said you need that data stream from a second, fixed receiver so that would require (most practically) a cellular connection, and that service would get expensive for everyday use. And you don't need cm-level accuracy to find the nearest gas station.
But there is a market, and I think if you built a turnkey system using Navsparks for under $500 you'd get a lot of buyers. For all those companies in construction, farming, etc. using the expensive RTK systems that we've seen thus far you won't be likely to grab that market -- those existing, high prices imply guarantees from the respective manufacturers (Caterpillar, Deere, etc.) that it'll work right and that the integration is seamless. It's similar to how a company like HP can sell a 1 TB disk drive for $300 for a Proliant server; the buyer is expecting guarantees and is thus willing to pay 3x the street price for the product. That's not to say the buyer is making a wise choice, mind you, but that's how the market works.