Description

Ticker                     L / S                        Price                       Date                                       Name                                    Sector

FSLR                      Short                      $133                       12-8-09                                  First Solar                             Thin Film Solar Modules

Price Target:  $100 (2-3 month timeframe), $70 (9 month timeframe)  

Catalyst:  Dec 16^th introduction of 2010 guidance could be below street estimate of $6.69 (could be $6 or below).  There is a good chance an additional German feed in tariff (subsidy) reduction could be announced by the end of December.  Full negative revisions of EPS on the out quarters will have to wait until Q409 and perhaps Q1110 is reported. 

Quick Pitch:  Although "the short story is out there", it is unappreciated that FSLR is 80% dependent on the German market which has provided an unrepeatable level of subsidies and EBIT margins for their CadTel thin-film modules in the last 3 years.  FSLR's biz is highly dependent on German "free field" builders, the most speculative of the German module demand (and the most at risk in further "feed-in tariff" subsidy cuts.  By definition, FSLR's 35% EBIT margins are due to dramatically reset lower, below street expectations.  This is because FSLR is emphasizing making up for German revenue with the Systems / "EPC" business, where FSLR enters the installation business in the North American markets to drive module sales.  FSLR's efforts to acquire North American projects / customers (its "captive business" strategy) should be taken as a further sign of the coming German weakness and the shift to lower-margin, lower-quality revenues.

Argument Points

1)  FSLR's dependence on Germany's "free field" (ground-mounted) business is underappreciated (~45% of sales).

2)  "Free field" customers are most at risk and thorough Q409 are showing a panic pull-forward of module demand ahead of subsidy cuts.

3)  By definition, FSLR's entrance into EPC / contracting will substantially lower EBIT margins (not fully reflected in street estimates).

4)  FSLR's effort to acquire customers (as a "captive business" strategy) is a low-margin, low-quality "revenue grab" strategy.

5)  FSLR's recent low quality earnings and its stretch to "make the numbers" could be evidence of the earnings danger ahead.  Its                 Mongolian solar farm development also appears to over-promise.

6)  The polysilicon market is finally cracking, which is eroding FSLR's cost advantage as a non PS-based module manufacturer.  There is                 strong evidence that the traditional boom / bust model in the semiconductor business may finally be applicable to the solar                 components business in 2010.

7)  Recent rally from customer Phoenix Solar's PR last week creates a shorting opportunity (+ discussion of other catalysts)

1)  FSLR's dependence on Germany's "free field" business is underappreciated (~45% of sales).

The above shows the falloff in Spain solar demand in the last year since subsidies were cut.  This is not directly applicable to Germany (because Spain enforced a 500 MW cap for subsidies) but it shows that country demand can be very sensitive to subsidies.  Since late September, Germany's politicians have been loudly saying that they would like to increase the size of planned solar subsidy cuts next year ("feed in tariffs").  Incremental German political power shifted to the FDP (Conservatives) on the Sept 27^th election - they promised tax cuts and reduced government spending (specifically naming outsize subsidies to solar).  Germany gives the solar business the highest margins in the world because the FITs are 32-43 euro cents per KW hour (vs 9 cents for wind).  They are spending $2B / year to subsidize solar (85 euros annually per citizen) for 1% of power production.  $2B is not small relative to the $17.5B government deficit for H109.

German politicians might be less sympathetic to foreign companies like FSLR.  FSLR has the highest German exposure:  77% of its Q309 sales (and a likely greater percentage of net income because of the higher margins in Germany).

2)  "Free field" customers are most at risk and through Q409 are showing a panic pull-forward of module demand ahead of German subsidy cuts.

~10% of the German market is "free field" solar construction - large open area fields.  The rest of the market is dominated by rooftops (commercial and residential).  FSLR has commented that up to 60% of its business recently has been "free field" - implying that FSLR is by far the largest module supplier to free fields.

These are projects that have historically been levered 10x and had 20%+ IRRs based on high legacy feed in tariff rebates.  FSLR has admitted that current IRRs are in the 12% range for "free field" equity investors and that IRRs would largely be wiped out after the planned ~10% FiT reduction Jan 1.  We may be witnessing a "cash for clunkers"-style pull forward of demand to beat this subsidy decrease.  Every solar company has noted a pickup in German demand in Q3 (it can be seen in the Taiwanese monthly reported solar sales also).  There is anecdotal evidence of installations speeding up concurrent with the Sept talk of larger subsidy cuts.

Module purchases would likely have to be made by Oct to complete installation for the 2009 FiT.  As mentioned, during the Sept 2009 elections, talk of an additional FiT cut spread.  FSLR has said the incremental cut might be 5% (implemented mid-year) for a total 2010 cut by June of ~15%.  Conservative politicians have mentioned that FiTs should be reduced ~30% in the near term.  It's likely that the total reduction could approach 20%.  FSLR has a poor record in forecasting this - they said on July 30^th that they expected no additional FiT reductions.  Either way, the pressure on "free field" investors is clear, with little in additional cuts beyond the planned Jan 1 cut needed to essentially wipe out their IRR.  There is already another in place cut planned for Jan 2011.

FSLR is recently trying to argue a way to preserve "free field" speculator IRRs.  FSLR IR argues that project developer margins have been bloated recently and that they have 20 cents / watt of margin to give up to get back to their normalized 10 cent margin.   If all this return was surrendered to the equity investor, they might be able to keep a positive single digit return, but this is unlikely.  EPC margins are already close to 0 so if anything, more margin would have to be surrendered by the module makers like FSLR (modules are still ½ the cost of a project).   (For background on "free fields" economics:  the economics of "free fields" are usually split 4 ways:  1) project developer (10 E cents / watt),  2) EPC (cost plus of 2 E / watt),  3) module maker 1.60 - 1.80 E / watt depending on what market prices you believe,  4) equity investor.  All in costs can be close to 4 E / watt.)

There's another reason FSLR's German exposure is so risky:  FSLR's "thin-film" module is better suited to ground-mounted "free field" installations than residential rooftops.  IRRs are higher on residential rooftops than "fields" because of higher FiTs.  It is likely that further FiT cuts will be more punitive to the "free field" speculator than the residential rooftop customer.

For the rooftop business that FSLR has, it has been more oriented towards industrial and commercial rooftops.  This is because FSLR's Cadmium-based thin-film technology has a watt / lb ratio that is 30-60% lower than polysilicon-based peers.  It has often been reported that by not using silicon as an input, FSLR has had the lowest cost modules.  However, this is at a price.  FSLR's modules have 10-11% panel efficiency vs 18-19% for PS-based comps like Sunpower.  On smaller residential applications, area and weight efficiency often moves the winning sale to PS-based modules.

The argument is still out also on whether FSLR's Cadmium Telluride modules are safe for residential use or whether there will be an "asbestos"-type risk tail.  The Cadmium is apparently trapped in the chips and FSLR has offered to do all recycling.  Still, there is some belief in the industry that the CadTel module is better suited away from residences towards "free fields" and industrial rooftops.  The Cadmium argument is still developing and there is a non-negligible risk of EU regulations against it.  All these arguments may indicate that when residential demand increases as a % of the German demand mix, FSLR may have a harder time preserving total German market share.

3)  By definition, FSLR's entrance into EPC / contracting will substantially lower EBIT margins (not fully reflected in street estimates).

Although FSLR doesn't disclose module ASPs, it is possible to back into them and derive a gross margin $'s model.  The model below uses estimated ASPs and reported costs per watt to generate implied revenue and implied gross margin $.  The estimates in green show Barclays' recent revenue and gross margin $ estimates for FSLR.  As can be seen, the revenue and gross margin $ match up historically with reported numbers but future estimates diverge from the "implied" numbers (which are based on Barclays' ASP driver assumptions).  This divergence is created by the startup of FSLR's "Systems" business - essentially the earnings "plug" to make up for the German business. 

The "Systems" segment will have its first contribution to revenue in Q409 (guidance of ~20% of rev).  Systems is essentially the non-German business (mostly North American) where revenue will be made up of module sales + EPC rev (~0% margin).  FSLR management has said to think of EPC revenue as "module enabling" revenue, essentially zero margin installer revenue helping to drive module sales - the business should be roughly ½ 0% margin and ½ full module margin.  FSLR has been unable to take on the EPC / installation biz in Germany but it is using it as a revenue driver in North America.

The net effect of all this is that by definition, consolidated margins should go down and stay down.  However, this is not reflected in street estimates.  As can be implied from the model above, two things have to happen for FSLR to hit Barclays' Gross Margin $ targets:  1)  systems revenue has to take off from being 20% of revenue in Q409 to 45% of revenue in Q410  2)  systems revenue has to be at a similar margin to module revenue.  Barclays' segment-driven estimates really only have gross margin % declining from 41% to 38%.

In reality, in the very optimistic scenario that Systems revenue goes from 0 to $300M per quarter in the 6 quarters up to Q410, gross margin dollars could decline 28% on a quarterly basis and EBIT could decline 35% (even with significant reductions in Op Ex and R&D).  The implication of this is that EPS estimates would have to be closer to $4 in 2010 vs the current ~$6.60.

A further implication is that there is not going to be a snapback in 2011 EPS.  If you take a business that was doing 35-40% EBIT margins, and restructure its revenue base so that 20% - 40% is at half of the old margins, you significantly dilute the operating margin structure.  Even if FSLR is able to double revenue from the current base (which is unlikely because new capacity wouldn't come on for 6 quarters at least), it may not offset the margin pressure from the mix shift to Systems and the ASP pressures industry-wide.

FSLR's IR seems to understand that it may have an increasing negative perception of operating margins on its hands.  It says they are considering offering a "module" and "systems" breakout in their future reporting.  To me, this just signals that "systems" will be a greater, margin-dilutive part of their revenue.

4)  FSLR's efforts to acquire customers (as a "captive business" strategy) is a low-margin, low-quality "revenue grab" strategy.

The good work of Gradient and other accounting based shorts on FSLR's financing of customers doesn't need to be repeated here.  FSLR has played accounting games with its investment in the Lieberose solar plant, a project of a major customer juwi GmbH.  After a reclassification of their investment in the project earlier in the year, FSLR started recognizing revenues from the project (a reversal of earlier guidance that it wouldn't recognize module sales but instead investment returns on the solar field itself).  This may have boosted Q2 revs by 15%+.

FSLR's IR department described to me management's efforts to create "captive" business to replace "transitional" markets such as Germany.  It was even implied that they have enough control over these projects to control the timing of revenue recognition.

The "captive" revenue strategy's figurehead is the $400M acquisition of OptiSolar in early 2009.  This was bought after module sales fell off a cliff in the months after Lehman and the beginning of Spain's drastic subsidy reduction.  OptiSolar was essentially bankrupt, with very little book value and had recently laid off half its staff after being unable to continue with financing for its North American solar developments.  OptiSolar has a large project in California and other projects in Canada and the U.S.  This echoes other deals done in the space recently (see WFR's recent acquisition).  It is essentially a trend of manufacturers buying zero-value added projects developers (OptiSolar) or distributors in an attempt to find off-take for their increasing oversupplied modules.

Although GAAP prevents FSLR from recognizing sales to itself for projects in development, there are levers that FSLR can pull to generate revenues from these captive projects.  Blythe is a U.S. project that FSLR invested in before OptiSolar and it recently announced the sale of the project to NRG (Q409 effect on revenues).  The stock rallied on this announcement, but details of the sale were not disclosed and it is still unclear what margins will come through the income statement on that project.

5)  FSLR's recent low quality earnings and its stretch to "make the numbers" could be evidence of the earnings danger ahead.  Its Mongolian solar farm development also appears to over-promise.

Gradient and others have mentioned FSLR's and the industry's use of the balance sheet to maintain the appearance of strong net income after the module downturn in Q109.  If DSO's had not grown massively in Q209, revenues would have been -25% sequentially for the industry.  Since Q209, the "cash for clunkers" race to beat the German subsidy sunset has buoyed H209 from being a complete industry blowup.

FSLR's own numbers show an increase in DSOs and DSIs (on top of an oversubscribed pricing rebate program they starting offering recently in Germany).  FSLR's Q3 10-q showed that finished goods inventory grew from 11% of inventories to 31% over the last 3 quarters.

Besides the deterioration in earnings quality metrics, FSLR showed a drastic and unexpected Op Ex cut in Q309 to "make the numbers" after sales decline 10% q/q and gross margins unexpectedly fell 6 percentage points q/q to 51%.  Without a $20M reduction q/q in SG&A FSLR would have strongly missed EPS estimates.  Further SG&A reductions are certainly possible to stretch to make the numbers, but starving the R&D line (~ $65M / year) to maintain EPS would probably produce a lower multiple on FSLR's reported EPS.  It is important to note that operating margin has fallen from 40% to 34% in the last 3 quarters even with Q3's massive SGA reduction.  Q4 guidance was for further margin weakness to the 24% level.  Management was hopeful and implied this might be a trough quarter for EBIT margins (thus the street's optimistic projections).  However, with the increasing share of systems revenues, it's likely there is further downside.  Compare FSLR's margins with the rest of the module space:  no other player even comes close.  FSLR is still afforded a sector-premium multiple on EBIT margins that have the most downside.

Expectations for the Ordos project in China may be too high.  FSLR management has been light on details, but heavy on potential in describing the "2 GW" project they landed in China in Inner Mongolia.

A news piece on Ordos can been seen here:  http://www.youtube.com/watch?v=0h7V3Twb-Qk   Ordos is an experimental Chinese city where the government essentially built a new city without residents.  The video describes how a new city was built 20 km from "old Ordos" and the pictures of brand-new, empty roads and unaffordable apartment buildings is clear.  Ordos is being used as the poster child example of overdevelopment in China.  The video also notes that Ordos is China's "capital of coal", holding 1/6^th of the reserves.  So FSLR's targeted area in China for a "2 GW" solar plant is a city that has already had a massive infrastructure overbuild and is rich in coal.  (http://en.wikipedia.org/wiki/Ordos_City)  Ordos is near the Mongolian border to the north and 500 miles from Beijing - a bit stranded for a massive solar project.

The latest talk on a possible Chinese FiT has been 16-20 cents (US equivalent).  This is 35-40% of the level of the German subsidies - and we know that FSLR has implied that "free field" IRRs would be wiped out with a 10% German FiT cut (without margin surrender by the EPCs). 

The Ordos announcement itself is a bit odd, with a "2 GW" demand headline (2 years of FSLR's capacity) but a notice of phase I demand of 30 MW (1.5% of 2 GW).  The additional phases are in 2014 and 2019.  FSLR's press release activity on new projects has picked up recently, but they all take a similar form:  large headline reported numbers of MW's of demand but light on deal specifics, timing, and in some cases they imply co-investment or project financing by FSLR.

6)  The polysilicon market is finally cracking, which is eroding FSLR's cost advantage as a non PS-based module manufacturer.  There is strong evidence that the traditional boom / bust model in the semiconductor business may finally be applicable to the solar components business in 2010.

This is been a perennial and painful prediction for FSLR shorts:  that all the capacity expansion would finally matter and stocks like FSLR would start trading on replacement cost after their earnings experienced strong cyclicality similar to the semiconductor business.  These predictions have been early but evidence is finally mounting of a coming "bust" in polysilicon prices and modules.  Thin-film modules such as FSLR's are ~15% of the solar business (the rest of the market is polysilicon-based modules like those of SPWRA).

FSLR's premium valuation and high margin structure has existed for the past few years because silicon prices made FSLR's Cadmium-based modules extremely cheap to produce in comparison to PS modules.

Recent FSLR cost per watt of 85 cents still beats many polysilicon based producers' cost of $1.40- $1.60, but the gap has been closing.  When PS prices were $400 / kg, FSLR's cost advantage was much wider because the PS modules were forced to add $3+ / watt in silicon alone to the cost of a module.  As shown in the table below, PS module producers aren't going to be able to reach FSLR's cost if PS falls to $30 / kg, but at 99 cents, they will be fairly close (and this assumes that PS prices with a 70% increase in new capacity for 2010 stay above the cash cost of production of $25).

What is clear is that FSLR's cost advantage has been eroding - it is certainly not the $3+ it used to be and FSLR's 55% gross margin quarters are likely over.  The last 7 quarters have seen a cost per watt reduction of ~30 cents to 85 cents for FSLR:  a good deal of this has come from the ramping of the low cost Malaysian facility and to a lesser extent, levering overhead and stock expense.  FSLR's production has gone from about 25% Malaysia a year ago to 73% recently.  FSLR is promising costs per watt in the 52 cent range in 2014 based on further efficiency improvements.  However, the easy cost gains have been made.

Polysilicon is ~$60 / kg now with analysts estimating a large "cash for clunker" pull forward effect benefitting demand from module makers to feed the German market before the FiT sunset.  UBS estimates that the lower grade polysilicon available to the solar industry will be 109k mt in 2010, up 73% and equivalent to 14.6 GW of solar module demand (neglecting the 1.5 GW that comes from non-PS based modules like FSLR).  Compare this 14.5 GW of PS equivalent supply to the likely 5.5 GW of demand for PS-based modules next year. 

There have been widespread reports of renegotiation of supply contracts for polysilicon and also wafers (see the Q-cells vs LDK headlines recently).  It is starting to appear that any supply contract in the industry signed a few quarters ago is at risk and considered toxic by the buyer.

Moving past raw polysilicon into modules, there is a similar ~70% increase in module capacity predicted for 2010 primarily from Asian PS-based module production.  See the table below:

There is a very strong argument to be made that demand could be flat in 2010 (around the current ~6.5GW level which includes the German demand pull-forward).  Most street analysts are still predicting a strong demand increase in 2010 but their estimates have been moving in the wrong direction.  Most analysts are moving close to 9-10 GW of demand after being in the mid teens on their estimates only a few quarters ago.  It is important to note that there are 7-8 quarters of lead time on the new module and PS capacity installations so that new capacity builders were basing their capex plans on market projections of 2010 demand that are now 2 years outdated.  Before Spain blew up, the street had much higher predictions for 2009 demand. 

The market is already in already in an oversupply situation (notice the bloated inventory levels at major FSLR distributor customers in Germany (Phoenix Solar, juwi, Conergy).  The short term German demand pull-forward has helped relieve some of this overcapacity but channel reports are still for bloated inventory levels at all parts of the supply chain.

Even recent indications of demand where there is monthly data shows a sharp drop off in recent months.  Italian demand (traditionally 1/10^th the size of Germany) has fallen off a cliff to 6 MW / month in the last two months (vs averaging 30-40 MW / month over the summer).  California has gone from installing 14-18 MW / month since late 2008 to 2.4 MW in November.  These are small markets compared to Germany, but these are also some of the markets that street estimates depend on to make up for some of the slack in a weakening German market scenario.  They are moving in the wrong direction.

Let alone the fact that it doesn't make great sense that Germany should be 40% of world demand when its sun cover isn't exceptional.  Street analysts estimate doubling of U.S. and Chinese solar demand in 2010 to keep the market going and there is no evidence of that in the recent numbers or realistic, bottom-up estimates of country specific demand.

It is clear that some time in the intermediate future, solar demand will grow from the current 6.5 GW level to 10 GW +.  That could be in 2011 / 2012.  Then could it grow much further as residential rooftop demand takes hold.  But rationally you have to ask:  at what price points and at what margins?  It is difficult to push residential solar at the current costs of $30k / home.  Prices (and therefore margins) have to come down significantly and as mentioned earlier, the residential market may not be friendly to Cadmium products.  The near term best hope is that U.S. and Chinese utilities but their priorities lie more with preserving their own businesses (namely the grid).  Does a U.S. utility really want to see mass adoption of solar at the residential level in the U.S.?  That could make the grid obsolete.  U.S. utilities may prefer wind or solar-thermal projects that will necessitate preservation of the grid.

7)  Recent rally from customer Phoenix Solar's PR last week creates a shorting opportunity (+ discussion of other catalysts)

The stock & sector have been rallying on the Copenhagen conference and FSLR customer Phoenix Solar's announcement last week that it had a record month in November.  Phoenix is a German distributor and its strength could be more indicative of the subsidy sunset effect than a sustainable rally in German demand.  The Jocksdorf project development work and installation has particularly boosted Phoenix Solar's quarter and that installation was using FSLR modules.  However, those modules were likely sold in to Phoenix in Q3.

FSLR's recent rally could prove to be short-lived with the Dec 16^th guidance call coming up.  If you look at the recent RBC notes and other analyst notes from the Nov roadshow and Dec conference presentations, street analysts are attempting to take a bullish tone while "burying the lead" that 2010 street expectations have to come lower still.  FSLR IR may be trying to nudge the street lower for 2010, but it hasn't fully sunk in yet.  The street is bullish about the possibility of a new plant announcement (FSLR was supposed to announce a new plant back at the spring analyst day).

FSLR's comments at the CSFB conference last week indicate a conservative view of global demand next year at least relative to street expectations (they mentioned 7-8 GW).  The CFO also declined to comment on whether there was an "unnatural" (unsustainable) component to Q409 German demand ahead of the Jan 1 FiT reduction.

The wild card is whether Germany announces another round of FiT cuts before the end of December (as has been rumored by some analysts recently).  The stocks are acting a bit like no incremental reduction will be announced or that it will only be 5%.  Either way, the already planned FiT reduction on Jan 1 has likely created the conditions for a sharp fall off in German ground-mount / "free field" demand in 2010.

Catalyst

Catalyst:  Dec 16^th introduction of 2010 guidance could be below street estimate of $6.69 (could be $6 or below).  There is a good chance an additional German feed in tariff (subsidy) reduction could be announced by the end of December.  Full negative revisions of EPS on the out quarters will have to wait until Q409 and perhaps Q1110 is reported.