Matt Simerson

Ivanpah, the largest Concentrated Solar Power plant in the USA has announced they are closing in early 2026. That followed PG&E announcing they plan to terminate their Ivanpah PPA contracts. The reasoning is simple: the PPA price from Ivanpah is estimated at $0.135/kWh. The PPA price on building a new solar PV plant is about $0.05. Ivanpah finds itself in a similar position as existing coal and nuclear power plants, where continuing to operate them is decreasingly economic.

The demise of Ivanpah is a harbinger for the remaining CSP plants. The first seven phases of SEGS were decommissioned by 2021 and replaced with PV, leaving 2 CSP phases remaining. Crescent Dunes went bankrupt in 2019, was reopened in 2021, and now discharges only at night.

The history of CSP builds in the USA is short:

	Built	Operating	Capacity	Storage MWh	Capacity Factor
Solar One	1981	1982-1986	7 MW	0
SEGS	1983-1990	1984-2021	160 MW	0	19%
Solar Two	1995	1995-1999	10 MW	0
Nevada Solar One	2006	2007-	72 MW	0	18%
Ivanpah	2010	2014-2026	377 MW	0	24%
Genesis	2010	2014-	250 MW	0	28%
Solana	2010	2013-	250 MW	1,500	34%
Mojave	2011	2014-	250 MW	0	26%
Crescent Dunes	2011	2016-2019 2021-	110 MW	1,100	20%

There’s a few fun facts to notice in the table:

1. Our CSP plants were all built before 2011.
2. Only Solana and Crescent Dunes are paired with storage.
3. The capacity factor of CSP, ranging from 18-34% is comparable (not better) than PV solar.

The thing that changed around 2010 that caused the death of new CSP plants was Swanson’s Law. Swanson’s law is the observation that the price of solar photovoltaic modules tends to drop 20 percent for every doubling of cumulative shipped volume. By 2010, the cost to deploy new PV versus new CSP had tipped firmly towards PV:

By 2020, the cost to deploy utility scale CSP was 5 times higher than PV, which nearly guarantees no more CSP will be built. To remain competitive, existing CSP plants need an edge in the marketplace. Solana and Crescent Dunes have storage, so they can sell their electrons at night when the cheaper solar PV is offline. Solana is further East in Arizona where it sells exclusively to APS where they’re still competitive. From out here in the cheap seats, it’s hard to see a long runway for Genesis, Mojave, SEGS 8&9, and NSO.

Nevada Solar One is an interesting case. They have a 20-year PPA that started in 2007 and so expires in 2027. From the NREL project page, their 2023 LCOE of $0.20/kWh isn’t very competitive, especially against their PPA sell price of $0.14. Using their production numbers I was able to calculate their total project revenue of $313M. That repays the $266M in construction costs but will fall short of also covering all their operating costs. I’ve read they have a 40-year lease on the land. The smart money is to raze the site and deploy PV–which has a much lower LCOE, doesn’t consume natural gas, and isn’t a water pig.

Over the years I’ve read enough citations that I’ve accepted it on faith that solar panel performance degrades about 2% in the first year and then 1% every year thereafter. Today I had a chance to test that.

Yesterday afternoon I hauled 6 new solar panels up onto my roof and wired them up. Today they’re producing so I spot checked against the production of my existing 290W panels from 2016. The rule of thumb above plus math suggests that those 8 year old panels have degraded about 9% and likely produce about 290W * 91% = 264W. The new panels are rated at 405W so I’d expect each old panel to produce 65% as much as a new one.

Today is an overcast day, which may affect production differently on the new versus old panels, but when I average the production of 10 old panels to 210W and divide that by the average of the 6 new ones at 320W, I get 65.6%. I find it very satisfying when reality matches theory so closely.