This is in contrast to dedicated lithium ion batteries, which just don't work out economically. Otherwise you're right - the utilities would do it themselves at scale.
Elsewhere someone corrected me that the peak grid usage in a year is actually significantly higher - the figures I was looking at were average daily figures - which may make grid savings quite a bit higher than 10%.
I don't think it necessarily is a clear financial win now; I'm just saying, if you squint just right it kind of makes sense now, and with the march of battery technology and PV, not to mention the inevitable adoption of EVs whether or not this scheme exists, it's going to look better and better from here on in.
(Since we know that the latter isn't economical, then that also indicates a problem with your numbers - I suspect that for one thing, insurance couldn't be economically provided at $12k/pack if all of those packs were being used for daily peaking storage).
The key is probably that the $12k replacement program is an investment in Tesla; they're cash starved, and are willing to sell the batteries at a loss in return for the cash advance. Also, they're betting on the batteries being ~half price in 10 years.
From further reading, they'd be $30k - 2.5 times the price - to buy upfront today.
But that doesn't mean that the subsidy doesn't exist - if 1% of consumers would buy a Model S due to its other advantages, maybe 2-3% would if that purchase returned $1500 a year back to them.
So while you're right that it's not outright going to pay for itself in 2012, it probably halves the gap between a Model S and a BMW in the same class, and it's only going to get better as battery/PV technology improves.