There are many companies doing just this (sunrun, solarcity, etc). There's currently some loophole that as a business they get paid more for their electricity than a homeowner would (due to subsidies to encourage businesses to use solar). So they get a price that you alone could not get. I think that's why it works.

bigethan -- that's not necessarily how it works. It does work differently in different state & local markets, so it's possible that there's a market that works that way. I don't know of any, personally.

Usually the deal is something called a PPA -- Power Purchase Agreement. It's usually set up that you agree to pay some fixed price-per-kWh, and that price escalates by some regular way over the life of the contract. That price could be higher, or lower, than the prevailing rates for electricity in that market, and that escalator could be higher or lower than rates at which energy prices increase -- but either way, they're locked in from the start.

There's no "loophole" - but the US tax code doesn't all you as a homeowner to use the accelerated depreciation deduction the way a third party for-profit business can if they own the system and lease it to you or sell you the power it produces.

It's really well established that home solar PV installations are currently substantially more expensive in dollars per kilowatt hour than grid energy.

I was following through some of Google's references and eventually found a graph without units perhaps explaining their reasoning behind this claim. It shows the cost of grid electricity undergoing massive exponential hockey stick growth from 2020 to 2030. So, if that happens, and you have a fixed percent 20 year mortgage on these panels, then sure, you might end up doing great compared to the suckers paying $100 a kilowatt hour (or whatever their exact number is since the y axis is not labelled) in 2030. On the other hand, an analysis that simply shows such apocalyptic increases in the cost of energy and considers no other factors is overly simplistic since in such an end times scenario mass-starvation, cannibalism, revolution, and the overthrow of the government would be likely accompanied by this situation and those homeowners with PV panels who did not have their own private armies and weapons stash would have long been dispatched and robbed of their valuable panels by those who do.

I figure it's a risky bet for them to make an investment based on the cost of electricity going up that much. Someone could just figure out a way to burn coal and capture the carbon and use it as fertilizer and the whole investment falls apart. That's probably why they are not sinking more money into this.

Well the loans are contracts to pay back with interest backed by collateral (probably even including the house equity), so it's guaranteed profit for Google. Since the savings are placed so forward in time and are not guaranteed at all, by the time the problem becomes a concern the money will already have been made. Let's say a homeowner starts a blog showing that they are paying much more for power with their new system than from the grid and questions it. Google then points out that the savings will come in 2030 at the end of the loan. If a journalist looking into this is persistent and good they will get a commitment to show exact numbers, which can then be directly discussed as to likelihood. But more likely the graphs will continue to have unmarked y axes, and those doing an analysis of how much the systems cost and what it means to charge 11% interest for 20 years will be discounted as not having the right facts, but no details as to what facts are incorrect and what the correct facts are will be forthcoming. Then, in 2030 the articles will either say how forward thinking the purchasers were to have bought in before electricity skyrocketed, or there will be a retrospective on how they got ripped off. In both cases, Google will have made their money back on their investment in collateral backed AAA rated debt securities.