What you miss in this calculation is maintainance. Apart from regular repairs of those 1200km2 photovoltaics (due to weather damage or hostile environment), in the end of the 25 year period you will need to start replacing those that you built in the first year - because they would be on the end of their useful life. So to keep up with the increase you will need either to double the investment of 38bln. or just keep spending 38bln. per year, just to maintain the share of the solar, and this until forever.

Another problem would be land - the panels should be near power lines and access roads and equipment and technological distances between solar batteries will most likely at least double the land requirements.

In conclusion I think it is obvious that to initiate such a program at the current state of the technology would be madness. 38 bln per year is the price tag of good 20-25 nuclear reactors that would replace our whole NG usage just for a couple of years.

Good Point LevinK, but you must also remember the other side of the equation. These panels are up there, soaking up sunshine and generating electricity and therefore creating MONEY.
This money will offset the upkeep costs and a proportion will have to go towards the purchase of the 'next generation' in 25 years time.

However I think the direction of the question is not so much; "Can we afford to do this sort of thing"
but more
"Can we afford NOT to do this sort of thing.."

If the area Chris describes was made up of (primarily) roofing on houses. If the householder was given tax incentives to install PV / CHP / Small Wind - then all of these really big numbers become divided by millions at a stroke, and then begin to make a difference.

Solar in isolation is not the answer (just like Gas or nuclear...) but increasing the mix is (IMHO) a valid initiative.

These panels are up there, soaking up sunshine and generating electricity and therefore creating MONEY

Chris's calculation shows that the cost of the solar energy derived this way would be some 20p/kwth. I just added that this cost will likely never come down (absent a technology breakthrough) because of the maintainance and replacement costs. Whether concentrated in large solar farms or subsidised by us taxpayers individual panels on the rooftops doesn't really matter - someone will be paying for it.

Now I think the real question we should ask the people is are they willing to pay 20p/kwth and more? I suspect that the answer will be "hell no". People would accept some small scale solar plants, but once the bills grow to unbearable heights the pressure will grow towards the cheaper sources and there is the chance that solar will be compromised completely. I think something similar is happening in Germany and Denmark which have already gone well down this path.

The bottom line is that it is not the time yet for this - subsidised small scale to stimulate the development of the technology - yes; large scale plants or "rooftop" programs - I don't think so.

The bottom line is that it is not the time yet for this - subsidised small scale to stimulate the development of the technology - yes; large scale plants or "rooftop" programs - I don't think so.
I agree with that, now isn't the time for large scale photovoltaics. Maybe in 30 years when there's no more gas to burn for electricity, we're making do with far less electricity (and paying more for it) than today and if the PV cost per peak watt has come down by half an order of magnitude or so, PV will make a significant contribution of ~10-20%.

However even then other electricity generation techniques may end up being significantly cheaper permanently relegating PV to niche applications.

It looks like we have similar opinions on that. Of course at this point it is quite speculative because of the uncertainties you pointed out.

Nevertheless I support photovaltaics as much as I support nuclear fusion (and maybe even more). And just like with fusion I would object to spend disproportional share of our limited resources for an unproven technology that can turn out to be a dead end.