Reality Check: First Solar and Tellurium

Is there enough Tellurium to meet the future needs of First Solar (FSLR) and other thin-film CdTe solar sell producers? The answer is crucial to long-term investors in FSLR who are looking at a P/E over 100. With a market capitalization of $16 billion and a likely future profit of $1 per peak watt ($1/Wp), FSLR needs to produce 1 GWp per quarter to have a P/E of 18. Old research from 2000 (and still being referenced as a 2006 Hoffman article) claims there's enough Te for 20 GWp/yr. Newer 2005 research says otherwise.

FSLR expects to produce 0.380 GWp in 2008 which will use only 10% or 20% of the world's 200 to 400 tonnes/yr Tellurium production ( 100 tonnes Te per 1 GWp is required for a 3 micron thickness). But FSLR grew from 0.002 GWp after its 1st year of sales (its end of 2002 capacity) to the recent conference call's "announced" production capacity of 0.9 GWp at end of 2009. No one thinks they'll be able to increase capacity another 45,000% in 7 years (120% growth per year), but lack of Tellurium is the only physical limiting factor. FSLR expects to go from 0.2 GWp in 2007 to apparently at least 0.9 GWp in 2010, which is at least 65% growth per year. More realistically, we can look at the 0.2 GWp produced by FSLR in 2007 and 0.38 GWp expected in 2008. That gives a 90% increase per year, still below the 5-year history of 120% growth per year. This gives us 0.2 x 1.9^5 = 5 GWp in 2012 which gives a PE of 10 in 2012 if they can keep the profit margin as expected to $1.3/Wp. But this requires 500 tonnes of Tellurium with a 3 micron process, nearly double the current world production of Te.

I have no doubt that FSLR can predict and is fully aware of its Te needs over the next 3 years. Hopefully, they have a good idea of what they will need and will be able to obtain for the next 5 years. Maybe the run-up in Te price from $10 to $200 per kg in the past 5 years is from CdTe players buying more than today's needs.

Obviously, anyone interested in FSLR needs to be concerned about the availability of Te rather than simply relying on 2000 research that claims 20 GWp/yr is possible. There are 4 questions that need to be asked:

1) How much Te can the world produce?

2) What percentage the world's Te can FSLR acquire?

3) How much Te will FSLR need per Wp if it improves efficiency and reduces thickness?

First question: According to the most optimistic and newest research, 340 tonnes were produced in 2006 from all sources, 60% more than observed and reported by the USGS. Te is locked up in copper at an average of 100 ppm and only 33% can be recovered at the coppoer producers that use an electrolytic process. This recovery rate might be doubled with research and investment. 90% of Te comes from copper, so other sources do not provide much hope. Te makes up about 1% of a copper producer's profit, so there is not much hope that even another 20x increase in Te price will substantially increase copper production. If copper production continues to increase at 5% per yer until 2012, we can estimate the world's available Te will be 340 x 2 x 1.05^5 = 867 tonnes. This optimistically assumes we can double the amount of Te being recovered from all sources. The factor 1.05^5 is a 27% increase in world copper production in 5 years.

Second question: FSLR pays about $0.023/Wp for standard grade Te ($200/kg) verses a profit of $1.3/Wp. With a 50% thinner CdTe layer of 1.5 micron, Te will be only 1% of their expected profit. So FSLR can support another 10x increase in the cost of Te. However, as explained above, this cannot cause much more Te to be produced. So the question becomes, do other industries that require Te currently pay more or less than 1% of their profit towards Te? This is a question that requires a great deal of knowledge about the Te world, but the Te world is hard to decipher. The answer could mean the difference between CdTe manufacturers being able to obtain 30% or 80% of the world's Te. Let's optimistically assume FSLR by itself can obtain 50% of the hoped-for 867 tonnes available in 2012.

Third question: How much Te per Wp is FSLR going to need? FSLR needs about 10 tonnes of Te to make 100 GWp of solar panels. In the 2002 NREL article linked to above, FSLR concluded that they would go from 8% to 13% solar efficiency in 5 years. In the 3Q, they reported only 10.5%, only half of their expected improvement. I have seen one article where a lab was not able to get better than 12% with thin layers of CdTe. I believe mass production CdTe is still 5 years from being better than 12%, a 20% improvement. If Te is the limiting factor, 20% is small compared to the benefits of making CdTe thinner. So the question becomes "How thin can CdTe be?" It's reasonable to hope that FSLR can reach 1.5 micron in 5 years while increasing to 12% solar efficiency. This would optimistically require only 40 tonnes Te/GWp in 2012.

So, we have 867 tonnes Te x 0.50 Te acquired x 1 GWp/40 tonnes = 10 GW possible for FSLR in 2012. This is twice as much GW as they will be able to produce if the 2007-2008 trend continues. So they may have plenty of Te for a PE of 10 in 2012 and enough for a PE of 5 in 2013 (using $1.3/Wp). But if they acquire only 25% of the world's hoped-for 867 tonnes of Te in 2012 and they have not stocked-up on Te, then the fantastic growth has to stop in 2012 with a PE of 20 at today's market capitalization.

0.9 microns CdTe with 12% efficiency was demonstrated in the lab (April 2007) which would allow 15 GWp/yr with the above 430 tonnes Te, but even that level of technology can not extend FSLR's current trend for more than 7 years. Theoretically, 0.4 microns with a reflective backing is possible, allowing up to 35 GWp/yr at 430 tonnes Te. With the USGS reported world available reserves of 21,000 tonnes Te, 2 TWp (40% of U.S. electrical grid needs) are possible with this hoped-for 0.4 micron technology. It will take at least 30 years to mine the copper.

The moral of the story is that there might be enough Tellurium for FSLR to have a fantastic run for 5 or 6 years and suddenly be stopped in its growth. This would justify current share price. Under less hopeful tellurium assumptions, the share price is not justified. Only under extremely dreamy assumptions could FSLR continue the great growth trend for 7 years.

Disclosure: At the time of writing I do not hold a position in solar cells, tellurium, or indium.

Comments

[Mark Anthony:]

Enter your comment hereNice article. But first you need to double check your math. At the expected 0.38GW in 2008, FSLR will be using 38 tons of tellurium, that's 20% of global production, NOT 2%.

With other industry demand growing rapidly, I do not even know how FSLR can squeeze the rest of tellurium users to spare 20% of the tellurium for its own use, because in 2004-2006, without FSLR demand, the other industry demands already exceeded 100% of the supply, causing dramatic price increase. Now for FSLR to get its share of 20%, the other industry users must reduce usage from 100% to 80%.

What kind of price raise is needed to force other industry users to reduce usage? I would say the price raise will kill FSLR's profit margin before it cause any other industry users to cut back. Because no one uses as much tellurium as FSLR does!!

Now regarding CdTe layer thickness, I don't think they can really push the thickness down to 1.5 micron. They have been trying to reduce the thickness for all the years, if they could do with 1.5 micron they would have long been doing with 1.5 micron. They might get something working in the lab, but rapid mass production is a different story. Electric current leakage is an obstacle they can not overcome when reducing thickness.

Don't believe what Ken Zweibel said. He is an expert. But he also has a huge vested interest in defending the tellurium availability, because he is running a CdTe solar panel startup himself, the PrimeStar Inc. He would never even publically admit tellurium is a problem. He could throw out all kind of numbers but you can't take his number too seriously when he was cornered to throw out the extreme argument that the universe is full of tellurium!!!

I am writing up another article to discuss other tellurium usages and their impact on FSLR. But before then, please read this some what related article on ENER, and phase change memory:

seekingalpha.com/artic...

Be sure to read my comments below that article.

2007 Nov 30 04:10 AM

[dslnbull:]

mr. Mark, It does not matter what do you say. Please live in the real world..Our stock market is worst in all over the world. We have more croock then good people at wall street. Our sec. chairman is crook.

Insiders can exercise the option but should not allow to sell the same day. There should be some kind of window , a week or month. This is a big issue but our COCX is a sucker and he will not take any action unless some body talk on CNBC.
I am sorry this is off topic but very important for small investors.

2007 Nov 30 08:26 AM

[GH:]

Your writing is better than Marc's, but not your analytical abilities. You write, "90% of Te comes from copper, so other sources do not provide much hope."

Your entire premise relies on this statement... which is utterly made up. You have no idea what the productivity or costs would be of Te produced by a direct Te mining effort rather than a copper side product.

2007 Nov 30 09:53 AM

[zawy:]

If you find a significant source of Te that is better than a few hundred 100 ppm, please let us know. I do not invest in faith of the invisible. All sources I have seen do not indicate there is any other substantial source, at any price. Gold is the only metal of the USGS's 9 rarest metals group (that includes Te) that is mined in the 2,000 tonnes/yr range you're hoping for. Gold costs 120 times more and has had centuries of research and investment. The least expensive of the others is palladium, 50 times more than Te and has 1/2 the Te production.

2007 Nov 30 11:59 AM

[zawy:]

As Mark said, FSLR will use 10% to 20% of global Te production next year, not 1% or 2%.

Other number/typo errors:
1.7%=> 1% (Te cost as % of current profit at 1.5 micron)
0.3% => 1% (Te profit as % of copper profit)
"10 tonnes to make 100 GWp" => "10 tonnes to make 100 MWp" (typo)
PE of 20 in last sentence should be PE of 10 (correctly stated elsewhere)

With that many numbers, 2 or 3 errors could be expected.

2007 Nov 30 12:22 PM

[zawy:]

Correction to "Newer research" link that has the most-accurate information on Te availability:

doi.wiley.com/10.1002/...

The full article costs $25.

2007 Nov 30 01:42 PM

[Mark Anthony:]

If you really want to see that paper, email the author of that research paper. He might be kind enough to mail you a copy without $25 fee.

Otherwise I guess it is fair use to quote some basic numbers. It is extimated that globally about 12.4 million tons of copper is processed using electrolytic method. Tellurium content varies but an average of 100 grams per ton of copper is a reasonable estimate. The exisiting process ultimately produces about 33% of the original tellurium in copper. So that's 33 grams per ton of copper. Theoretically if every copper producer is using the same method to extract tellurium, that leads to a global production of 400 tons a year.

Actual annual tellurium production number may be way much smaller, because some smaller smelter, or copper mines with poor tellurium content probably won't process tellurium at all.

2007 Nov 30 03:49 PM

[Andrew Ling:]

"Scott Roberts" is likely just another account that "Mark Anthony" has. In the Yahoo message boards he uses his numerous accounts to reply to his own messages. The fact each post under different names use the same false facts proves that any time you hear mention of "Tellurium shortage" it's probably the same guy.

2007 Dec 02 05:09 AM

[zawy:]

If you'll notice, "Scott Roberts" is predicting FSLR could have enough Te for a PE of 5 in 2013 and that there is no Te shortage for at least 3 years. "Mark Anthony" would never say such things. Andrew Ling has just lied twice when he says Mark and Scott post the same "false facts". Their numbers are different by nearly 100%, and both have provided references to back their numbers up. Mark relies on USGS and NREL data whereas Scott uses peer-reviewed data. If Andrew Ling cannot respond to this post with facts to contradict Mark or Scott, then we can accept Andrew Ling as a known lier and/or clueless. However, I'll agree that Mark uses multpile identites and responds to his own posts in order to bash FSLR.

2007 Dec 02 09:28 AM

[FSLR Analysts:]

It looks as if First Solar is also concerned about the supply. They are vertically integratrating when demand for their products is well above their ability to supply. The Acquisition of DT was a bit odd for them. They are above 20% OM, which gives them cash to do two things - lower the solar price further and expand the business. Instead of preparing for further GW capacity increases, they are moving verticle. To me, this signals a hint of risk. They are moving into the US market with the acquisition but they are also moving away from PV manufacturing as the sole source of revenue. It looks like Te supply is a concern for FSLR afterall.

2007 Dec 05 12:33 PM