Jump to content
Sign in to follow this  
dydo

JA Solar (JASO)

Recommended Posts

Guest Klothilde
that translates to around 18 cents on wafer level.
Now I'm not following you. What do you mean with 18 cents on wafer level? What I meant is that the production costs of a Panda module are 15 cents higher than the production costs of a multi module. These extra 15 cents will be distributed between the several production steps. Presumably the largest chunk of the 15 cents will be in the cell conversion step, which is way more complex than the multi cell conversion process.

Share this post


Link to post
Share on other sites

If you assume that the cost increase per piece is in the wafer step (to get n-type mono) and the per piece cost are constant in cell and module processing steps, then the higher conversion efficiency of the Panda wafer reduces the per watt cell and module processing cost compared to the multi wafer. So the Panda module has maybe 32 cents cost for wafer to module conversion while the multi module has 35 cents cost to do the same, thus if a Panda module costs 15 cents more, then the Panda wafer costs 18 cents more. IMHM (in my humble model). Note the assumptions. It seems they were not the same as yours (I assumed difference in wafer processing, you in cell processing). Panda is two things - mono and n-type, right? Mono as far as I know has no extra cost in cell processing (maybe opposite given the much higher price difference between mono and multi on wafers than on cells), but huge extra cost in wafer processing. The n-type is a characteristic of the doping of the wafer. Since it is a trait of the wafer I assumed this mainly affected the wafer processing, but that might not be the case..? Panda might also have some other optimizations than n-type wafers, e.g. in the cell step that increases per piece cell processing cost.

Share this post


Link to post
Share on other sites
Guest Klothilde

If you assume that the cost increase per piece is in the wafer step (to get n-type mono) and the per piece cost are constant in cell and module processing steps, then the higher conversion efficiency of the Panda wafer reduces the per watt cell and module processing cost compared to the multi wafer. So the Panda module has maybe 32 cents cost for wafer to module conversion while the multi module has 35 cents cost to do the same, thus if a Panda module costs 15 cents more, then the Panda wafer costs 18 cents more. IMHM (in my humble model). Note the assumptions. It seems they were not the same as yours (I assumed difference in wafer processing, you in cell processing). Panda is two things - mono and n-type, right? Mono as far as I know has no extra cost in cell processing (maybe opposite given the much higher price difference between mono and multi on wafers than on cells), but huge extra cost in wafer processing. The n-type is a characteristic of the doping of the wafer. Since it is a trait of the wafer I assumed this mainly affected the wafer processing, but that might not be the case..? Panda might also have some other optimizations than n-type wafers, e.g. in the cell step that increases per piece cell processing cost.

I'm afraid I made the argument unnessesarily complicated by bringing in the Panda example, shame on me. Just to close the Panda topic: assuming that cell and module conversion costs per piece are equal between n-type mono (Panda) and multi, then your 18 cents wafering cost is accurate. However in the case of Panda the cell processing step is indeed more complex and costly than with multi (a.o. includes selective emitter and more difficult passivation steps), so some of the 15 cents extra costs do come from the cell step. According to Odyd the 15 cts. cost gap can be explained as follows: 6-7 cts extra cost for p-type mono wafering relative to multi wafering. Then 4-5 cts extra for n-type mono wafering relative to p-type mono wafering. The rest of the cost gap comes from more expensive Panda cell processing relative to multi cell processing. Anyhows, if Odyd says p-type wafers cost 6-7 cts more to make than multi wafers than that would be somewhere between your and my assumtions, no? Odyd, what is your current thinking regarding p-type mono and multi wafering costs for cost leaders? Let me throw in three pieces of evidence that point to a way smaller cost difference than 10 cents between p-type mono and multi wafering 1) Check out the spot prices of mono and multi wafers at any time in Q4 (either pvinsights or solarzoom). You will find that the price difference is only around 3 cts/watt. Assuming that this represents roughly the difference in production cash costs, then it's difficult for me to imagine a further 7 cts/watt cost difference alone from depreciation (e.g. 2 cts for multi vs. 9 cts. for CZ-p-type-mono). I can imagine 1-2 cts additional cost difference arising from depreciation, bringing the total cost difference to 4-5 cents. Note: Current prices show a price differential of 7 cts/W between mono and multi wafers, however I think this includes a nice margin for the mono wafers due to current strong demand for mono products out of Japan. 2) JA said on their 12Q1 call that wafer processing cost for multi was ballpark 12-14 cents, and that for mono it was about similar: http://seekingalpha.com/article/639101-ja-solar-s-ceo-discusses-q1-2012-results-earnings-call-transcript?part=single Even if I also don't buy the "similar" I think JA would have answered differently if mono had been 10 cents above multi. 3) The Daiwa overview already mentioned, which points to a cost difference of 4 cts (scaled to today's cost it would be 3 cts.)(p19): http://seekingalpha.com/article/639101-ja-solar-s-ceo-discusses-q1-2012-results-earnings-call-transcript?part=single Not that this is the bible (as a matter of fact I distrust many analysts out there), but it's one more point of evidence. My current thinking on the distribution of JASO's 53 cts non-si costs: 16 / 16 / 21 Any thoughts and further evidence on the wafering costs is greatly appreciated. Now reverting to the COGS reconciliation: Good news is that even if we tweak the 53 cents cost distribution a little bit here or ther you will nevertheless end up with aggregated COGS that are pretty close to the COGS reported in the income statement. So thanks god that CPW are beginning to get reflected in the COGS... As a next step I'm modeling the revenue side of JASO and think that this can be indeed very interesting in Q1. Three speculative factors I think contributing to a BIG ASP jump in Q1: - ASPs in Q4 seem to have been purposely cut in order to reduce inventories and crank up cash ahead of convertible bond payment. @low inventory they may revert back to market pricing - Large fraction of high efficiency mono cells and modules in production, which are currently experiencing relatively high price and margin hikes - Strong revenue shift away from China to Japan My main concern is the lack of worldwide market diversification. Apart from Japan they don't seem to have other strong footholds. That's why I'm eager to look at the Feb export data to see how they're evolving...

Share this post


Link to post
Share on other sites
Guest larryvand

Good morning Klothilde, am I to assume the company gave you a number that separates sales & marketing expenses from their G&A?

Share this post


Link to post
Share on other sites
Guest Klothilde

No they didn't. All I asked for was the revenue and shipment split between modules and module tollings (which I got). Total opex w/out impairments is $30.7M or 6.1 cts/w. That must be the lowest among the solar 11. What is your concern?

Share this post


Link to post
Share on other sites

Check out the spot prices of mono and multi wafers at any time in Q4 (either pvinsights or solarzoom). You will find that the price difference is only around 3 cts/watt.

I'll stop you there first. On pvinsights the diff is 1.272 - 0.862 = 0.41 => up to 10 cents diff http://pvinsights.com/ The JASO comment sounds strange, sound like they meant their wafer to cell conversion cost. The most recent mono processing cost I have is from SOL Q2. 21 cents, while their multi is now at 11 cents. It is possible that mono has come down as well since then. http://phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9MTUwODIzfENoaWxkSUQ9LTF8VHlwZT0z&t=1

Share this post


Link to post
Share on other sites

Using numbers provided by Jason at one point I figured that processing cost was about $0.16 for mono wafers versus multi-wafers at $0.11. He quoted that at one point mono processing cost was about 65%, versus 50% for multi. I used 4watt multi wafer as piece and 4.5W wafer for mono. The pricing appears to be difference of 0.22 per piece using Solarzoom data, which is very near to processing costs difference for mono. As far as n-type wafer goes, making one is no different as much as making p-type, it is just different dopant. The difference is in making metalization, where cost was high due to high temperature and you have to use different more costly material to make junctions. ECN interview explains how they overcome both to make it lower now, to multi process.

Share this post


Link to post
Share on other sites

Using numbers provided by Jason at one point I figured that processing cost was about $0.16 for mono wafers versus multi-wafers at $0.11. He quoted that at one point mono processing cost was about 65%, versus 50% for multi.

Thanks, I remember this. However, for me this confirmed the 10 cents difference between mono and multi. Assumption: Multi wafer processing cost is 11 cents Jason fact 1: Multi wafer processing cost is 50% of the wafer cost => poly cost 11 cents (20 $/kg, was this the poly market price at the time of Jason's article?) Jason fact 2: Mono wafer processing cost is 65% of the wafer cost => poly cost 11 cents + mono wafer processing cost 20.5 cents

Share this post


Link to post
Share on other sites
Guest Klothilde

I'll stop you there first. On pvinsights the diff is 1.272 - 0.862 = 0.41 => up to 10 cents diff

It comes to 8 cts difference (due to the wattage difference), however as already stated I think a big chunk of this is not cost but margin in the mono wafers.

The JASO comment sounds strange, sound like they meant their wafer to cell conversion cost.

Good point, maybe. Then again 12-14 cell conversion cost @June 2012 strikes me as a bit low...

The most recent mono processing cost I have is from SOL Q2. 21 cents, while their multi is now at 11 cents. It is possible that mono has come down as well since then. http://phx.corporate-ir.net/External.Fil…TF8VHlwZT0z&t=1

THX

Share this post


Link to post
Share on other sites

It comes to 8 cts difference (due to the wattage difference), however as already stated I think a big chunk of this is not cost but margin in the mono wafers.

Yes, I have 8 cents price difference too, but think that standard mono margin is compressing below cost diff, due to lower value diff than cost diff now that it cell, module and BOS costs have come down.

Share this post


Link to post
Share on other sites

Thanks, I remember this. However, for me this confirmed the 10 cents difference between mono and multi.

Explo it is per wafer( piece) not watt. Due to conversion variable I believe my calculation is more reflective. 156mm multi 6.3, mono 8.5, avg. 4W piece , 240W module, 4.5W mono piece 270W module. Strip 17% VAT, 5.229, 7.055 mono 50% per piece =2.6145 multi 65% per piece = 4.58 mono conversion 2.61/4 = $0.11 per watt USD conversion 4.58/4.5= $0.16 per watt USD

Share this post


Link to post
Share on other sites

4W piece , 240W module, 4.5W mono piece 270W module

I'm not sure that reflects efficiency diff of standard multi and standard mono. Multi have caught up with standard mono. Multi used to be 3.7w a while ago.

Jinko has standard offering of both:

http://www.jinkosolar.com/download_131.html

As you see, on module level there's no difference. Mono on cell level needs to be at least 2% higher due to cut corners and higher CTM loss to get to the same efficiency on module level. To me this says that the difference between a standard multi and mono piece is closer to 0.25w per piece than 0.5w per piece.

156mm multi 6.3, mono 8.5

I think those are prices not costs, right? I've not said that mono and multi processing price difference might be lower than 10 cents. In fact it has been contracting and I think it will contract even more, due to contracting value diff.

Contracting price diff due to contacting value diff and less contraction in cost is precisely why I wouldn't want to be in standard mono wafer processing and why I think SOL retired half their (already small) capacity there.

Share this post


Link to post
Share on other sites
Guest Klothilde

I'm not sure that reflects efficiency diff of standard multi and standard mono. Multi have caught up with standard mono. Multi used to be 3.7w a while ago.

Jinko has standard offering of both:

http://www.jinkosolar.com/download_131.html

As you see, on module level there's no difference. Mono on cell level needs to be at least 2% higher due to cut corners and higher CTM loss to get to the same efficiency on module level. To me this says that the difference between a standard multi and mono piece is closer to 0.25w per piece than 0.5w per piece.

Let's do the crunch. Currently poly averages around 17.28% (pvinsight implied efficiency) and mono around 18.8% With total area of 243.36 cm2 for poly wafers and 238.95 cm2 for mono wafers this results in wattages of 4.21 for poly and 4.49 for mono wafers. This results in standard module wattages of 245W for multi and 261W for mono @ 3% CTM loss. So there is a noticeable difference. Please don't take Jinko as reference cuz their mono efficiency sucks. If you look at JASO you'll get a more accurate picture:

http://www.jasolar.com/uploads/files/201210/20121018110447_2GxhNR.pdf

http://www.jasolar.com/uploads/files/201210/20121018105709_ksNnGl.pdf

I think those are prices not costs, right? I've not said that mono and multi processing price difference might be lower than 10 cents. In fact it has been contracting and I think it will contract even more, due to contracting value diff.

Contracting price diff due to contacting value diff and less contraction in cost is precisely why I wouldn't want to be in standard mono wafer processing and why I think SOL retired half their (already small) capacity there.

As a matter of fact the price difference has expanded quite a bit since Q4 due to the increasing demand for high efficiency products out of Japan. Beginning of december 156mm multi wafers were priced at 6.5 and mono at 7.6 :

http://solarpvinvestor.com/spvi-news/391-weekly-solar-market-price-trends-mainland-china-report-18

Share this post


Link to post
Share on other sites

explo I am comfortable using the pricing as these are zero gross margin prices which are pretty much the cost, or being sold at the cost.

Share this post


Link to post
Share on other sites

Okey, let's use pvinsight price per piece then: Multi 0.862 & 50% processing => 0.431 poly + 0.431 processing Mono 1.272 & 65% processing => 0.445 poly + 0.827 processing Adds up again. No matter how we twist it Jason says that mono processing is roughly twice the cost of multi processing.

Share this post


Link to post
Share on other sites
Not really, per watt at 0.4 this is .10 versus 0.18 at 4.5. Taiwanese processing is 30% more expensive than in China, when it comes to mono cells. I am not sure what this argument is about, mono is more expensive to process, but it is becoming less expensive. If JASO says that they have 0.53 per watt processing, this is pretty decent, if mono dominates their production lines.

Share this post


Link to post
Share on other sites

Are mono dominating their lines? Will be tough to work down that 10 GW GCL contract then.

Share this post


Link to post
Share on other sites

why? they have 2.5GW cell capacity, and 1.8GW module capacity. If they do more than a half in mono ( I do not know) there is plenty of room. I learned they have plans out of the 1.7-1.9GW in 2013 to do 70% in module I thought their wafer is pretty much all mono is that not accurate? I have not been following JA with great detail, but I had this impression. YGE has over $1B in contracts with OCI, but has found a way to GCL. I do not think we are looking at immediate outcomes anyway.

Share this post


Link to post
Share on other sites

I haven't listened to their cc. If they say they do primarily mono then I guess they do (I would be surprised by this). They have 500 mw mono wafer capacity. They have another 500 mw multi wafer capacity, idled and heavily impaired.

Share this post


Link to post
Share on other sites

@Klothilde. Some interesting points. I may have mentioned here that Comtec supplies wafers to SunPower. They just released their FYR 2012. Sold over 310MW 125mm and 28MW of 156mm wafer , 77.7% to Sunpower with Super Mono Wafer. This is about 159% growth, 8% positive margins. Loss nevertheless. They sell what you call Super Mono Wafer, which is up to 23% conversion. They are working with a Japanese customer to sale wafers there. They are also expanding with new wafer plant in Sama Jaya-Malaysia. This could be just to side with SPWR, but they sell to many Chinese, still p-type. What is really cool they cut wafers to 155 microns, which is a lot better than 180 microns. Their whole production line .6GW is made ready for SMW. 125mm sold at 0.20 per watt?! and 0.29 per watt for 156mm

Share this post


Link to post
Share on other sites

On the same note Solargiga, exclusive mono wafer maker http://www.solargiga.com/attachment/2013032719364217_en.pdf Manufacture and ship N-type solar ingots which are used to produced high conversion efficiency solar products (22-23%) to Japanese customers, represented 80.5% of external ingot shipment volume in 2012

Share this post


Link to post
Share on other sites
Guest sony1

YFrom what I have read ASP's are the Highest in Japan.

Relatively speaking ASP is going down in Japan as well. Japan just launched a massive QE today, and Yen is down 3.5% vs Yuan.

Share this post


Link to post
Share on other sites
Guest Djovanny

Jaso is Very good speculative play with like DQ run to the upside due to low float and strong cash position Jaso is currently bottoming imho p s nice doji is forming today with higher high lower low . Good luck

Share this post


Link to post
Share on other sites
Guest JulyWebb

Relatively speaking ASP is going down in Japan as well. Japan just launched a massive QE today, and Yen is down 3.5% vs Yuan.

I saw an article yesterday on Japan recently published and it stated the Highest ASP's were the highest of all in Japan. I didn't save it. When you get a chance Read JA's Q&A from the 4'th Qtr. when asked about positive margins going forward and look at where they are showing shipments to be for 2013. A lot of China was in Q4 but it won't be for 1'st Qtr. of 2013 nor the remaining of the year from what Jian Xie said.

Share this post


Link to post
Share on other sites
Guest JulyWebb

And He did the same with FSLR and FSlR since his last call FSLR has more than doubled... If you listen to Q&A from the Conference Call JA is forecasting more of their sales for Q1 and the full year 2013 in the Asia-Pacific Regions, Japan where ASP's are higher. JA's note is due May 15'th and they are well prepared to pay it! JA has one of the best balance sheets in the industry. If you go back to August 16'th 2010 Gordon Johnson sent out a Sell Rating on JA Solar and gave JA a price target of $5.00. Two months and about a week later JA made a High of $10.14. JA made a big jump in a short amount of time. I kinda think someones need to cover their short shares. The Stock closed Green today.

Share this post


Link to post
Share on other sites
Guest
This topic is now closed to further replies.
Sign in to follow this  

  • Recently Browsing   0 members

    No registered users viewing this page.



  • Donate

    Please donate to support this community. We appreciate all donations!

    Donate Sidebar by DevFuse
  • Upcoming Events

    No upcoming events found
  • Forum Statistics

    • Total Topics
      42
    • Total Posts
      91,123
  • Who's Online (See full list)

    There are no registered users currently online

×