Compared to cell supply for branded shipped modules, module production is more in-house, but still has a relatively high component (approximately 25-30%) coming from third-parties. Image: Astronergy

Compared to cell supply for branded shipped modules, module production is more in-house, but still has a relatively high component (approximately 25-30%) coming from third-parties. Image: Astronergy

The use of third-party cell and module sourcing has seen a massive uptick in the solar industry, during the past 2-3 years.

The appetite of leading global module suppliers to rely on outsourced cells and modules has been clear to see in the past 12-18 months, but tends to go through almost quarterly tactical adjustments, often driven by factors that are potentially creating risk for long-term downstream asset holders of utility-scale sites.

This article explains why OEM supply has become key to some of the Asian (and in particular Chinese) based module suppliers that have prioritised rapid shipment growth targets, while seeing non-domestic overseas module supply market-share gains as a dominant operating strategy.

Why has OEM supply increased in recent years?

While on a smaller scale, Chinese module suppliers used to rely sporadically on third-party cell and module supply from brand-limited minor industry producers in China, while seeing Taiwan cell makers as the quality outsource route when higher performance modules were needed for module shipments.

As soon as trade import tariffs from the US and Europe were imposed however, Southeast Asia become the de-facto region to have cells and modules produced. While several of the leading Chinese module suppliers, that had ready access to cash, set up dedicated cell and/or module facilities in countries such as Malaysia and Thailand, this was effectively the start of the multi-GW low-cost multi-GW OEM factory building period for the industry in Southeast Asia.

To begin with, most of the Chinese module suppliers had no option other than to find cells made in the region (or from Korea or Taiwan) that would ship to one of the OEM fabs to be assembled into modules, and then given a label of the company that effectively ended up being no more than a glorified distributor.

While some of these factories in Southeast Asia may appear like RMB-financed pseudo-sweatshop operations, simply outsourcing modules has been adding 2-3 pennies onto blended module COGS. When global ASPs (EXW) were trending at the 60c/W mark a few years ago, a few pennies onto COGS was a small price to pay for retaining access to markets in Europe and the US.

From 2016 onwards, this broadly created an environment that only added to the premise that such OEM supply was acceptable from a business-model perspective. And in keeping with the rather knee-jerk capacity-expansion excitement that tends to underpin anything related to the terms solar and growth, new OEM sites continued to sprout in Vietnam, Thailand and Indonesia, to name just three.

How much global module supply is coming from OEM-made cells and modules?

The simple answer to this question is that no-one on the plant knows this to within about 10% at best, such is the level to which is seems to have become an embedded form of practice, with seemingly very few questions ever asked.

For many of the installations done this year in the solar industry, it may be years until someone looks into potential underperformance on site installations, but there are a few ways in which we can at least get a handle on how much in-house cell and module production is being done across some of the leading c-Si global brands today.

As we have been discussing for a past couple of years on PV-Tech, there are only a handful of global PV module suppliers today that have meaningful market-share globally (including China). Most of these companies are included in our silicon module super league (SMSL) grouping of Canadian Solar, GCL-SI, Hanwha Q-CELLS, JA Solar, JinkoSolar, LONGi Solar and Trina Solar.

Therefore, this is perhaps the best place to start, in terms of getting a feel for OEM cell and module outsourcing.

For the purposes of our new analysis here, we have removed GCL-SI and LONGi Solar from our dataset, owing to the companies being relative newcomers to overseas module shipments. We have also taken out Hanwha Q-CELLS due to the company largely having a strategy from its Korean-driven headquarters to use 100% of in-house cells and modules for shipments, and for matched cell/module capacity levels in each of its three regional and technology-specific manufacturing locations.

Therefore, we focus on the in-house cell and module production levels from Canadian Solar (CSIQ), JA Solar (JASO), JinkoSolar (JKS) and Trina Solar (TSL). In many respects, these four companies are the leading Chinese global module suppliers in almost every end-market of relevance in the industry today. From an operational standpoint, there are many similarities, including:

Each of the companies created a niche and value-added offering in the industry, either from being a cell-production specialist (JASO) or a module supplier that moved into cell production at the GW level some years back (CSIQ, JKS, TSL).

Each has had on/off investment phases in downstream projects activities, with CSIQ being the standout player today that could easily stand-alone based on upstream or downstream operations.

All of the companies become globally recognized based on its multi-GW cell/module capacity located in China, but each now has varying levels of in-house cell and/or module operations located in Southeast Asia. Other module facilities outside Asia do exist, but very much from an arms-length perspective, and not as core manufacturing sites.
Crucially however, each of the companies has been actively trying to grow its manufacturing business based on rapid module shipment growth, and simply relying on in-house supply and third-party sourcing to fulfil sales pipelines in the near-term.

This is blatantly obvious for anyone looking at nominal nameplate cell and module capacity levels and cross-checking these to either quarterly or module shipment volumes. 
However, while this is a far as many market observers go with this analysis, it is only by backing out precise manufacturing line utilization rates for each company (by quarter, by technology, by manufacturing location) at the cell and module stages, that we can really get to the bottom of the amount of third-party supply going on.

With JinkoSolar reporting its quarterly results for Q2’17 yesterday, we spent a couple of hours earlier today updating our models for the above four companies, and adding up all the data.

The exercise was in fact prompted by JinkoSolar stating that it was going to actively seek to reduce OEM module supply during 2H’17, compared in particular to the higher-than-usual levels the company was required to use in Q2’17, as it set a new world record for quarterly module shipments by a company in the industry (almost 2.9GW of module shipments in just 3 months).


Analysis and discussion

We focused on two issues. How many of the shipped modules were using cells produced by the company that was selling the module? And similarly, how many modules were assembled in-house?

The graphic below shows the annual consolidated output, looking at the cell production question.

The use of third-party sourced cells has increased significantly between 2013 and 2017, in part reflected by the new cell technologies and processes that have replaced the legacy standard p-type multi cell lines that could be constructed and ramped quickly.

The use of third-party sourced cells has increased significantly between 2013 and 2017, in part reflected by the new cell technologies and processes that have replaced the legacy standard p-type multi cell lines that could be constructed and ramped quickly.

The dependence on third-party cells is often overlooked, with many of the companies in question happy to present data on third-party OEM module supply only. So, let’s try to explain why third-party cell supply has been increasing so much in the past few years, despite many companies adding new cell capacity and being highly vocal on new roadmaps based on mono-PERC and diamond-wire-cut multi wafers.

It is true to say, there is no single reason; rather a host of factors. One of the most simple explanations is that adding module capacity is easy, is costing pennies in often state-financed buildings, and is much more low-tech compared for example to making a mono-PERC cell. Also, module lines can be tweaked to use new technologies at the cell side.

Another key issue has been the change in cell technologies that has rendered 3-busbar p-type multi standard (Al-BSF) cells as almost obsolete today. When most of the Chinese companies came into cell manufacturing, they did so by choosing the established entry-route at the lowest technology node. Adding multi-GW of this capacity was easy, many of the companies could even use Chinese equipment suppliers, and the know-how was there to ramp up quickly.

New capacity at the cell level today has to be 5-busbar p-mono-PERC or a multi variant that can be slotted into a 60-cell module and be specified above 265W (p-dc, STC). Adding multi-GW of factories at this level is a different proposition, and far removed from the ease of module expansion.

Another factor is new technology cell-line ramping and optimization. While many observers have almost accepted at face-value that PERC is established, even on mono, it is taking even the major Chinese companies at least two years to work out the recipe. The gap between these companies and Hanwha Q-CELLS and REC Solar is significant, and confirms that there is still much learning to do before we can say that PERC is a mainstream module offering across the industry, even on mono.

Somewhat related to the difference between cell and module stages discussed above, module outsourcing is not anywhere near as high as third-party cell use.

This is shown in the figure below, where we have started the y-axis at 60%, to show the annual trending in a more visual way.

In-house module assembly remains above 70% from the leading companies shown, but third-party assembly has been increasing compared to 2013/2014 when most modules were produced in-house.

In-house module assembly remains above 70% from the leading companies shown, but third-party assembly has been increasing compared to 2013/2014 when most modules were produced in-house.

Compared to cell supply for branded shipped modules, module production is more in-house, but still has a relatively high component (approximately 25-30%) coming from third-parties. The trend has certainly been increasing in the past few years, and can be explained almost entirely from the companies each having highly aggressive market-share targets, well above the expected module production levels in-house.

Another factor keeping OEM module supply at high levels comes from the volatility of the industry, and in terms of sales pipelines. Having 6 months good sales visibility today in the PV industry is a luxury: planning manufacturing lines based on this timeline is unrealistic.

Therefore, companies have to make the option of running lines at high utilization rates constantly, and adjusting upstream and downstream inventory levels, or simply pushing out deliveries when sales volumes exceed factory output numbers.

Otherwise, the option of ramping on and off GW-factories is completely unsustainable. As such, this simply means that short-term shipment changes have to be dealt with through third-party OEM suppliers. When many of these OEM suppliers exist largely for this situation, managing delivery contractual obligations is hardly a worry.

Why is no-one talking about quality and reliability?

This has remained a mystery for years. Almost always, the justification for turning on and off OEM supply of cells and modules has been based upon how quickly a few percentage points can be added to next quarter’s gross margin figures; or whether or not shipment guidance can be met.

The latest driver (at least this quarter) is coming from Q3 seasonality resets following a blistering Q2 for many, and also to counter poly pricing going up for a couple of months.
In fact, I cannot recall a time when anyone ever spoke about the need to limit OEM supply, in order to improve module quality and reliability by having 100% control of in-house production, tools and bill-of-materials.

If you are a multi-GW asset owner, whose investments hinge on maximizing 20-30 year investment returns, what matters to you? A penny on a quarterly blended cost for the manufacturer: or minimizing risk of module failure or underperformance over the next 20 years?

Solar panel supply is a 20-30 year proposition, not a consumer electronics product with a lifetime of 1-2 years, and whose manufacturing route can be readily outsourced to the next low-cost labour market in Southeast Asia.

Until now, it is fair to say that First Solar and SunPower have been the only major module suppliers that have linked manufacturing processes with reliability tests, field data and long-term project returns. These companies certainly have a strong case to be at the forefront, given that each is driven by maximizing profits at both module and project sides of the equation, and one part subsidizing the other is simply not an option.

While many of the Asian companies have moved downstream with project investments (many of which still hold relatively high asset levels), the fact that the downstream operations are typically left to buy any module type instantly negates these companies from having the credibility to talk in-house product quality in relation to downstream project returns using 100% module supply (whether the product is even made in-house or third-party).

Just recently, Hanwha Q-CELLS started to make some noises about 100% in-house cell supply and what multi-GW of PERC production quality means for site returns. This is something that you can only do if you can stand behind the intel-inside argument, and for now, very few of the major c-Si module suppliers can legitimately fly this flag.

PV ModuleTech 2017 to address this issue head-on

The whole issue of what modules are being supplied to utility-scale solar globally today, who makes the key parts (cells and modules), who supplies the key materials and module manufacturing equipment, and how can we minimize the risk as the industry considers another shift to bifacial or glass/glass modules for example, are some of the themes that underpin PV-Tech’s first staging of the PV ModuleTech 2017 event in Malaysia, on 7-8 November 2017.

How do we really connect module supply traceability, material selection, testing and certification, insurance, reliability and risk-mitigation for asset owners? And how should this be explained to them in a way that does not mean they default to a nonsensical academic Tier-list publications, or certification that is not applicable to the products being used on-site?

PV ModuleTech 2017 is expected to see all the major global module suppliers speaking about why their module supply is best suited for project developers, EPCs and site owners in 2018, backed up with data that matters and communicated in a way that is understood by downstream players. At the other side, the event will include the major EPCs, developers and site owners explaining what matters to them from potential GW-level module supply partners.

To register to attend the PV ModuleTech 2017 event or to get involved more, please follow this link.

Tags: c-si manufacturing, pv modules, smsl, solar cell, pv moduletech, jinkosolar, ja solar, canadian solar, hanwha q cells, longi solar, gcl system integration technology, trina solar, first solar inc, sunpower corporation