Dissecting the Solar Shakeout

A rather obscure article

Great piece. Based on my conversations with installers in Europe, this is EXACTLY what I’ve been witnessing lately. The balance of power has shifted - it used to be with the module manufacturers, but starting in Q308, terms have been dictated by the downstream end - project developers and the like.

One question for you, Daniel - I see you’ve actually built these supply and demand curves for modules. How on earth did you go around doing that?

Hey Daniel, thanks for your work. I’m very familiar with the research put out by some of the other big name research companies.  I won’t name any names, but I think this report blows theirs out of the water.  I’ve gone through your report now three times with a fine tooth comb, and despite the rough condition head for manufactures, I am very optimistic that I’ve discovered some areas of real profit growth. It’s so valuable to see the entire picture. Great stuff - Keep up the good work!

Danny Boy,  Nice to see you didn’t fall of the edge of the earth.  Our research of the current PV shakeout has led to a very similar finding, with the exception of module ASP.  We project module ASP based on module demand which is based and on “project investment return”, with the view demand spikes substantially as return on investment increases and PV becomes a replacement for tradition power sources (i.e. grid parity).

The two primary problems I see with your report are:  FIRST;  At the ASP’s you project, the market would likely be supply (not demand) driven as there would be zero c-Si produced at your 2012 projected PV ASP.  SECOND:  Demand would be be much higher than you project as investment return on projects in multiple markets would be very high due to the fact the ASP’s you project would provide grid parity in multiple markets. I’ll give you a “C” grade for your report with extra credit for the sexy Sarah Palin glasses.

lc58harvard - Thanks. We did something really novel here. We independently constructed supply and demand curves in our two newest reports - PV Technology, Production and Cost and 2009 Global PV Demand Analysis and Forecast. The curves were built from the bottom up. On the supply side Shyam Mehta built company-by-company, technology-by-technology, region-by-region, supply stacks for every year from 2008 through 2012. This let us assign each company a place on the supply stack representative of where their manufacturing costs allowed them to bid into the demand markets.

On the demand side, I built bottom-up demand curves utilizing a multivariate demand function using a the results of returns-based analysis of individual project economics in major markets as inputs in the function. I built these curves for the 16 major PV markets and aggregated them up to get global PV demand curves for 2009 through 2012. They represent that demand potential at each module price point along an evolving price range for each year.

To derive our market-clearing module ASPs and demand volumes for each year, we reconciled the independently-constructed curves. This allowed for a very precise analysis of market states and generated a ton of outcomes related to companies, technologies, regions, and market segments.

Steve-O - Glad to see you haven’t lost any of your weird obsessions. There’s a significant problem with your analysis here. You assume that demand operates independently of financing structure. While grid parity is a demand driver, the extent to which project developers are willing to build projects is related primarily to their ability to access cheap capital to finance these projects. In price supported markets like Germany, grid parity is still a number of years off. So, projects in this market will be built where developers can get financing to build these projects. In Germany specifically, this can happen at mid-range module prices - and probably will, given German lenders inclination toward c-Si module with dependable warranties. So, predominantly European multicrystalline.

PV is now a buyer’s market. Access to finance and projects are now a much greater determinant for growth than access to poly. Because of this, the ability of project developers to bid down module prices in order to get the best returns possible will determine the suppliers actually able to sell into the 2009 market. Those unable to endure significant margin compression will be in some trouble.

With dry credit markets and lots of illiquidity among previously enthusiastic debt lenders, getting good returns is going to be relatively more difficult. Demand will expand in price-sensitive markets like the U.S. and Japan, though off of relatively small bases, meaning we won’t the outsized effects of this until 2010 or 2011. In 2009, lack of financing clarity will soften demand irrespective of some projects in some segments of these markets reaching grid parity.

Danny-boy, to the contrary, we don’t assume “demand operates independently of financing structure”.  We assume demand is based in part on financing, which is turn is based on investment return.  Higher return = more access to financing.
My issue with your report is your ASP projections.  If I read your chart correctly, your 2012 ASP’s are below production costs for all c-Si PV.  Therefore, only FSLR and the as yet to be determined emerging PV technologies will be supplying PV modules at your 2012 projected ASP’s.  The new PV tech won’t have a reliable warranty making financing much more difficult.  This leaves FSLR as the lone, easily financable survivor.  FSLR will be supply constrained, [as will the other lo-cost emerging PV technologies] making the market *supply* not demand driven as your report suggests.  On many other issues, I think we agree based on the little I’ve seen of your report.  A key difference being I didn’t drop any cash at Haaaavaad to figure this out.  And, those of us who can’t grow face hair don’t.

Daniel,

Am confused. The recent ?PV Technology, Production and Cost, 2009 Forecast? by Greentech Media & Prometheus Institute mentioned 2007 production capacity at 5.7GW, however, here the figure for 2007 is 3.2GW…? Is the 3.2GW simply growth between ‘06 & ‘07 or are the numbers confused?

Daniel,

Am confused. The recent ?PV Technology, Production and Cost, 2009 Forecast? by Greentech Media & Prometheus Institute mentioned 2007 production capacity at 5.7GW, however, here the figure for 2007 is 3.2GW…? Is the 3.2GW simply growth between ‘06 & ‘07 or are the numbers confused?

Hi Adam - Thanks for bringing this up. Production capacity - the nameplate on the manufacturing facility - was 5.7 GW in 2007. In the same year, producible supply was 3.2 GW, which is what I said in the piece and what we really care about. We arrived at producible supply by derating capacity, taking into account ramp-up constraints, financing difficulties, and production delays. We care more about producible supply than production capacity because it’s a more accurate indication of what can actually be produced for the market.

Interesting analysis of the market.  How about this for a simple explanation: Like other tech bubbles, the aggregate investment in supply (VC herds + market exuberance) simply outstripped the market’s ability to consume.

Hard to say we are getting closer to grid parity due to the breathtaking fall of electricity prices. How could solar possibly compete in the US when we have round the clock power prices at $30/Mwh?

I kind of disagree with your assumption that solar is getting closer to grid parity. Power prices have plummeted dramatically (in line with natural gas prices primarily). The drop in power prices has been significantly higher than the drop in ASP’s. Hard to see how solar gets closer to grid parity in the US when round the clock power prices averaged $29/Mwh yesterday in PJM.

Where is PJM?  Where did you get that statistic?  How many end user can you list that pay $29/MWh?  Yesterday was not peak season btw,  Get back to me July 1 for that same price.

Sure, we can talk on July 1, as you know peak power prices in PJM are set by the cost of natural gas which is now trading at under $4/mmbtu

Enter your comment here.Steve, PJM = Pennsylvania New Jersey Maryland Interconection. It looks like c-Si can reach ASPs of $1.50. Get an envelope out and look at the low cost producers. STP list non-silicon production costs at 66 cents/watt. Both LDK and ReneSola list wafer processing costs of 30-40 cents/watt and silicon usage of 6 to 6.5 gram/watt. It seems reasonable to expect these production costs will continue to fall as learning occurs and fixed production costs get divided over large production volumes as utilization rates improve (First Solar isn’t the only one that can do it). STP predicts 50 cent/watt non-silicon costs and LDK is predicting surprisingly low processing costs down the road.

Cooper, the end-user has to pay for the cost of energy ($29/MWh) + Transmission and Distribution expenseses (Typically about $40/MWh) + Congestion + Stranded Capital market restructuring fees etc. Look at the break down on your bill.