There is a multitude of technologies used to convert solar radiation into electricity. From an economic point of view, it makes sense to separate them into two categories: (classic) photovoltaic (PV) power, in which solar cells of various materials are exposed directly to sunlight; and concentrated solar power (CSP), in which mirrors concentrate sunlight on a receiver. If I got the categories right, CSP includes the more mature solar-thermal power technologies, in which a heat-absorbing medium passing through the receiver drives a generator; as well as concentrated photovoltaic (CPV), in which the concentrated sunlight is absorbed by a high-quality multijunction solar cell. (Update [2011-9-1 1:36:55 by DoDo]: Discussion in the comments indicates that CPV is not, after all, part of CSP.)

For photovoltaic power, the bulk of the cost is in the manufacture of the solar cells, thus the cost of a single large project is not much higher than the cost of several small projects that together generate the same amount of electricity. This makes the use on any free surfaces greater than a few square metres viable.

The most expensive parts of CSP are the receiver and the generator (turbine), thus there are significant economies of scale to realise by adding more mirrors. That means a viability as utility-scale project, and also means a need for cheap non-arable land. In addition, solar-thermal technologies can also store energy and thus continue to generate by night, reducing intermittency and even enabling some controlled variable output.

With these different characteristics, how does the spread of PV and CSP compare?

First let me note that there is a common misconception about scale: the notion that for the installation of large amounts of generating capacity, a technology needs single projects to have large power. However, if the technology and regulations are such that small projects can draw in more investors (i.e. even homeowners) and can be realised by mastering less bureaucratic hurdles, the opposite can be true.

In spite of a higher cost of electricity produced, PV took off earlier than CSP: annual new PV installations were in the gigawatts worldwide every year since 2004, with 14 GW added last year. Booms in Germany and Italy included mostly rooftop installations, but large on-ground projects also spread, and were the bulk of the 2008 boom in Spain.

In the last few years, however, CSP also entered the series production phase, with dozens of plants built or being built in Spain alone. Capacity and production data for Spain by the end of June 2011 from Comisión Nacional de Energía:

(Note that due to the collapse of the PV market with the 2008 FIT changes, we can estimate the overall PV capacity factor for 2009: a respectable 19.9%. For CSP, my guess from the numbers is that it was similar or somewhat lower.)

What about support schemes?

In Spain, there are two schemes: one giving a surcharge on the market price, another giving a guaranteed price. For simplicity, I look at the latter only.

Currently (Q3/2011), the guaranteed feed-in tariff for newly installed PV ranges from 13.0324 c€/kWh for (large) on-ground farms to 28.1271 c€/kWh for rooftop installations below 20 kW. If I read the laws right, these rates are now fixed for the first 25-30 years of operation of a plant (like in the German feed-in law).

The rates are re-set every quarter since late 2008 (when they ranged between 32-34 c€/kWh), and have been cut strongly this year, especially for on-ground installations (their FIT was practically halved). It does appear that the measure again throttled the PV market in Spain, however, not to a standstill: even with this quarter's super-low FIT, 39 MW of new on-ground PV capacity was added.

For solar-thermal, there is still one FIT for all plants (old and new) which is re-set regularly, and this year it is set to 29.0916 c€/kWh. That's actually higher than in earlier years. It appears no price cuts were achieved yet that would warrant FIT reductions in Spain.

Although the support schemes differ too much to say for sure, it appears that utility-scale on-ground PV is now cheaper than CSP, in spite of the latter entering series production. The impression is confirmed by a decision two weeks ago to covert the first two units of a monster 4x250 MW project in the USA from CSP to PV. This was reported by melo in the Salon, but I quote the company press release:

Erlangen, 19 August 2011 With respect to the development of the world's largest solar energy location with a capacity of up to 1,000 MW in the Californian city of Blythe, U.S., Solar Millennium AG (ISIN DE0007218406) decided to take a new path in the last week. Instead of the previously projected concentrated solar power (CSP) technology in the form of parabolic trough power plants, the first two 250 MW plants are now to be equipped with photovoltaics (PV). With this decision, the managements of Solar Millennium and Solar Trust of America are reacting to the changed market conditions in the U.S. featuring far more profitable prospects for the use of PV in solar power plants that do not require storage.

They are quite frank about PV being the cheaper option – even accounting for special US support schemes benefiting CSP:

The decision to revert to the significantly less costly PV variant also means that Solar Millennium will finance the power plants on the free capital market and forego the loan guarantee by the U.S. Department of Energy for Blythe 1 and 2.

Then they insist that they haven't given up on the technology outside the USA, underlining the non-financial benefits of CSP:

"The decision to use photovoltaics in Blythe does not mean the Solar Millennium Group is turning away from its core technology of solar-thermal power plants. According to our understanding, the market differentiates between base load solar-thermal electricity generation or concentrated solar power on the one hand and photovoltaics for peak load demand on the other. Whereas the electricity from solar-thermal power plants was more economic only less than two years ago, this relation has changed completely due to the sharp drop in PV module prices, particularly from Asia. Due to its suitability for base load supply, many regions still attach great value to CSP in their energy mix, thus supporting the Solar Millennium Group's growth opportunities."

Putting it more bluntly, there is no overall energy plan and market short-termism rules in the USA. However, I think the more important difference is the way regulations (support schemes) are set up. As much as a surprise it is that PV can be cheaper than CSP, the latter is still at its infancy, so it's not necessarily a valuation of baseload, but fostering an industry in the hope of future lower prices that justifies higher supports.

Update [2011-9-1 4:54:18 by DoDo]: I also looked at the supply chain.

The generous support system in Spain did lead to the development of domestic companies, but not yet across the whole supply chain:

• Receivers: SCHOTT Solar (Germany) and Siemens subsidiary Solel (Israel; successor of Luz, which made the plants in Califormia two decades ago) dominate the market, there is also Archimede Solar Energy (Italy; Siemens has 45% stake).
• Mirrors: manufacturers include Flabeg (Germany; already supplied the plants in California), Abengoa/Rioglass (Spain; took market leadership from Flabeg last year) and Saint-Gobain Cristalería (Spanish subsidiary of a French company). Both Flabeg and Rioglass established factories in the USA; now I wonder how much work those will get.
• Pumps: Ensival-Moret (France; market leader), Flowserve (USA), Friatec (Germany).
• Turbines: Siemens and MAN Turbo (both Germany) are the only suppliers I could find.
• Mayor developers include the Spain-based companies COBRA (300 MW on-line), Acciona (264 MW) and Abengoa (181 MW); as well as Germany-based Solar Millennium (initial partner in 200 MW of projects in Spain) and Japan-based Mitsubishi (partner of Acciona in most of its projects).