Especially when compared to memory or processor chips, the current way of producing LEDs is ridiculously slow and expensive.
I'm not so sure about that. Even memory chips are priced in the ballpark of $1-10. Processor chips can exceed $100.
Bulk LED's sell for pennies. You can buy high power (1+ W) Cree LED's in moderate quantities from places like Digikey for as little as 50 cents each. The lamp manufacturers who are buying more than one real at a time should be getting even better pricing. An 800 lumen bulb takes 8-12 of them. There's some room for further cost reduction, but not a huge amount.
A significant part of their cost is related to the tiny drivers needed to convert effectively unlimited 110 VAC to a stable 3.3 VDC, current limited, and deal with the havoc most dimmers cause. Then there's the relatively complicated bulb bodies, and usually a non-trival amount of aluminum formed into a big heat sink that the design revolves around.
Increasing the efficacy may actually help more with cost than further reductions in the price of each LED chip. Get the heat down low enough and designers more or less stop designing their bulbs around dissipating it.
All things considered, I'd be a bit surprised if 60W-equivalent LED's ever get below $5 a bulb. They'll never reach the $0.50 a bulb price of incandescents, and probably not the $2/bulb price of CFL's. I'm really not worried about that, though. Even at $10 a bulb they're a good value.
My main interest is seeing color rendering and luminous efficacy go up. These two goals go well together, since increasing the CRI almost inherently means reducing the efficacy, so improving the latter reduces the incentive not to improve the former.
Also, another decade or so down the road, bulbs might become a minority product. With the lifespans these lights are potentially capable of, we might move towards the LED's being built in to the fixtures. I've already seen some office-style recessed troffers designed like that.