I think solar efficiency improvements played a very large factor and also the cheaper labour costs in India.

Why can't they just use Nuclear? Surely, there are inhabited places that can be used for it. Correct me if I'm wrong - nuclear power should produce more reliable power at a fraction of the cost of solar.

Nuclear requires huge up front investment, may not be so easy for gvmnt.

Solar farms (I believe) can scale slowly even after being operational.

I may be wrong but it's also likely that deeper analysis of the cost for solar doesn't look that bad as you're probably spending most, if not all of the money on your own people - giving jobs and pouring money back in, not out of the country.

India is building nuclear power too. But it is not as fast or inexpensive as you might guess.

India's most recently grid connected reactor, Kudankulam-2, took 14 years to complete: https://www.iaea.org/PRIS/CountryStatistics/ReactorDetails.a...

Expected costs for the next 2 units at the same site are $5.91 billion: https://en.wikipedia.org/wiki/Kudankulam_Nuclear_Power_Plant

That's $3.58 per real annualized watt assuming 90% capacity factor. If this new solar farm operates at 27% capacity factor (like Topaz did in 2015) that will be $3.88 per real annualized watt. So nuclear is still cheaper than solar in India in terms of construction costs. That might be offset by higher O&M costs; in the US at least, PV O&M costs come to $25/kilowatt/year while nuclear is $198/kilowatt/year: http://www.power-technology.com/features/featurepower-plant-...

Again adjusting for capacity factor (PV 27%, nuclear 90%), that means spending about $220/kilowatt/year on nuclear O&M and $93/kilowatt/year on PV O&M. I wouldn't expect the absolute numbers to be the same in India but the ratios may still be similar.

I'd say that the instantaneous generation costs for solar and nuclear projects that start generating in India now or next year are going to be pretty close. In the future, nuclear still has the advantage of working around the clock. But if the next 14 years see even a modest fraction of the solar cost reductions of the past 14 years, the next Kudankulam unit to come online will be far more expensive per annualized watt than a solar farm completed at the same time. 10 years ago it was a lot easier to figure out the lowest cost mix; nuclear power was cheaper than utility scale solar always and everywhere. It will be an interesting balancing act, in India and elsewhere, to determine just how much cheap-but-intermittent power you can use instead of expensive-but-steady power.

EDIT: I might have overestimated how well nuclear power performs in India. I assumed 90% capacity factor but it looks like all but one of India's nuclear reactors have a cumulative capacity factor below 80%. Kudankulam-1 was at 40% last year: https://www.iaea.org/PRIS/CountryStatistics/ReactorDetails.a...

WTF.

If future Indian reactors continue to operate at dreadful capacity factors like this, a new utility scale PV plant is already cheaper per real annualized watt.

On further reflection, whatever issues cause dreadful under-performance of Indian nuclear reactors might also affect solar plants. Solar needs significantly less maintenance than other kinds of generating capacity but if it gets no maintenance then the output will fall even in areas with excellent solar resources. For the US the Energy Information Administration publishes detailed information about all utility-scale generators so it's easy to track performance over time but I don't know where or if equivalent data is available for India.

The reason for under performance of nuclear is lack of fuel (due to sanctions imposed long back). Its only recently that India is signing up with many nuclear producers so this should improve drastically.