First Solar's $1/Watt Capex: Half the Industry Average

Last week I wrote an article entitled First Solar's Rapidly Expanding Economies of Scale in which I detailed the various advantages which have led to FSLR astounding returns this year.

This week I will delve deeper into capex costs. On Monday, Sharp (SHCAY.PK) announced a thin film plant costing 380 billion yen which will start producing in March 2010.

At the current conversion rate of 110 yen to the dollar this amounts to a huge investment of USD $3.45 Billion. Compare this to First Solar's (FSLR) recent Malaysian plants which will produce 168 MW annually with construction costs estimated at $150-200 million. Sharp's capex costs/watt are approximately triple that of FSLR. Also, construction times are nearly double that of First Solar's. Formerly one of the world's leading PV manufacturers, no wonder they've been stuck in the mud the last few years.

Stuart Wenham from SunTech Power (STP), themselves working on thin-film technology, was quoted as saying "I think with regard to a lot of the thin-film technologies around at the moment, the CapEx costs tend to be very high and it’s certainly a key issue that we are addressing in terms of the development of our new thin-film technology." This is why many so-called thin-film competitors to FSLR are dead in the water. In order to compete with FSLR's advantages a startup would require a far superior technology. In the case of the later they will probably still need billions and several years of tweaking manufacturing. CIGS, the so called "third generation" of photovoltaics may be the answer but to date it has failed to yield greater than 6% outside a lab. Miasole infamously proclaimed back in 2004 that they would bring a product to market by 2005. They have yet to commercialize a product. Neither has Nanosolar.

Contributing to FSLR's expansion in it's economies of scale are these unheard of $1/watt capex costs. With profit margins of roughly $1/watt, they have a one year return of capital investment. If you could get $1 back each year for each $1 invested you'd be expanding at breakneck speeds also. Of course at the current pace of expansion there is a question of how much FSLR product that market can absorb at current prices. Monday's investor's presentation stated that typical solar farm returns for systems 100KW-20MW range from 8-12% annually. Since these return are guaranteed by the government through their 39 euro cents/kwh feed in tariffs, combined with the predictability and reliability of solar, these returns are almost as safe as those from government bonds. Imagine purchasing a government backed bond yielding 8-12%. It's no surprise that demand has been nearly limitless in Germany, Spain, and Italy where generous feed in tariffs exist.

FSLR's economy of scale advantages are indeed multi-faceted. For example, the following conditions combine to produce astronomical growth:

1) highest profit margins in the industry
2) lowest capex/watt in the industry
3) skyrocketing stock price

This allows FSLR to issue secondary offerings much like they did back in July. The funds raised through these financing round can be invested almost exclusively back into new manufacturing plants. As more plants come online the pace of cost reduction increases as mentioned in the previous article. Upper management has been quoted stating they have access to vast amounts of financing but will not utilize it until necessary.

If FSLR announces another secondary offering raising $1 billion for example, rest assured that within a few months there will be an announcement of 4 new plants. This will probably occur in the first half of 2008. I for one will not be surprise if in addition to the 900 MW 09 production estimate we have a 2 GW 2010 estimate in 6 months.

Economics involve tradeoffs. If you want a car with a powerful engine you generally have to sacrifice mileage and pay a higher sticker price. Imagine if the more powerful the car became, the better mileage it got, the more reliable it became, the better it handled, and the cheaper it became.

This is what FSLR is trying to achieve. Revolutionary technologies overwhelm the traditional marketplace by finding ways to bypass the tradeoffs that entrenched business have to balance. You would expect to pay a premium for clean energy much like the premium you pay for a Ferrari. But when FSLR reaches grid parity, it will be like Ferraris selling for the price of Buicks. The first to reach grid parity will no doubt become a $100 billion market cap company. In the solar sector past earnings are relevant only in proving the future viability of a product and serving as a baseline from which progress is made. After all, if they could turn a profit earlier in the year with a mere 75MWs production, something no other thin-film company has yet to achieve, imagine what they can accomplish at 1gw, 2gw, and 10gw. The market is handicapping FSLR as the overwhelming favorite, with STP and SPWR now running a distant 2nd and 3rd.

Disclosure: Author is long FSLR, STP and SPWR with largest position being FSLR.

Comments

[Mark Anthony:]

All of FSLR's cheap and crappy factories are invain and useless, without an abundant and cheap tellurium supply.

The FSLR stock was hyped based on the WRONG perception that it can literally grow unlimited. If that is the case I would have bought it at current price level. But every one ignored on critical and fatal issue, that is the critical FSLR's tellurium supply. FSLR can produce nothing without tellurium.

I believe that not only FSLR's growth is unsustainable, it could actually be forced to go OUT OF BUSINESS, within two years, or it could happen within 6 months if a tellurium rush results. Read here:

seekingalpha.com/artic...

The price target of FSLR is below $20 within 18 to 24 months. No joking!

2007 Dec 04 07:07 PM

[solar jim:]

Hey, Mark Anthony, you might actually want to read this article as it has something your posts do not. Relevant theories built off of actual facts. Your Te item is important how? What purity of Te is the price you are quoting? What is the purity of Te that FSLR uses? You don't actually know, do you? Go away, please. Really. I get that you are short FSLR. Really, I do.

2007 Dec 04 10:27 PM

[Andrew Ling:]

Yes, any time your comment concludes with the statement "no joking" you know you have problems. Of course he could should be ending every sentence with that.

2007 Dec 05 01:53 AM

[FSLR Analysts:]

Andrew,
Your Sharp numbers aren't quite right. The 1GW fab is also an LCD plant at Gen 10. The CapEx for Sharp is more like $1.38 (see article about 160MW facility: 145MW add for $22B Yen.)
In addition, when you look at cost depreciation over 7 years for these systems, the CapEx comes out to around 25%-30% of the cost of sale. So a $1 vs. $1.38 is not too different on a Cost/Watt scale.
The big differentiator is in the conversion efficiency. Sharp's triple junction is comparable to FSLRs CdTe (10% module for Sharp, %10.5 stated for FSLR) so the two are price comparable. Sharp also has the advantage of no Cd and no Te to worry about.
Something to think about - and a bit of perspective before discounting Sharp.

2007 Dec 05 12:49 PM

[Andrew Ling:]

Thanks for pointing out the error I made with Sharp's capex. Is efficiency estimated to be greater than 10% by the time the plant is completed? I don't know how well Sharp moves technology from the lab to mass production. FSLR's 10.5% efficiency will most likely be around 12% by that time.

2007 Dec 06 04:45 PM

[JoeJoe:]

I think you also got the CIGS conversion efficiency wrong. Global Solar seems to be the leader with an efficiency of about 10%. - news.cnet.com/CIGS-cel...

There's a discount between the cell efficiency and the module efficiency but it wouldn't go all the way down to 6%. Currently, Global seems to be getting the bugs out of their process and if you look back at the history of First Solar you'll see they too had to get the bugs out of their process over many years. From my perspective it seems like an interesting technology to watch because CIGS should have more potential for improving efficiency than CdTe based upon the highest efficiencies achieved in laboratory settings.

2008 Sep 18 05:42 PM