The only “new” aspect of this particular supply chain crisis is the number of variables causing it; the industry’s reactions and potential pitfalls remain mostly the same. - by Joseph Song
The past 20 years will go down as one of the most important periods in energy history. The maturation of the renewable energy sector coupled with escalating pressure to mitigate climate change has led to the golden age of carbon-free energy technologies. But velocity of this golden age has collided violently with the global COVID-19 pandemic, creating a “pile-up” on the freeway. As a result, all segments of the supply chain, from raw materials to construction labor, have been severely impacted. This article is an attempt to take a “deep breath” and consider the present-day situation through the lens of past experience by diving deep on the module supply chain, specifically: 1) supply-side dislocation, 2) demand-side volatility, and, 3) key considerations for project development.
It’s a mess out there right now, as anyone financing and building a solar project over the last 18 months is painfully aware. Timelines and budgets have been almost universally obliterated by supply chain problems of just about every variety. The phenomena have not discriminated by project type, geography, or stakeholder, and the problems persist for virtually every system component…even labor.
Thankfully we, the renewable energy people, are rather accustomed to getting punched in the gut by fickle, volatile supply chains.
In the earlier years of the millennium, the most notable module suppliers were companies such as BP, Shell, Sharp, Kyocera, Evergreen, and relatives of both AstroPower and Spire. Many of these companies were built with conviction and purpose on the back of promising technologies. But demand emerged slowly in those early years, and cost reductions were slow to materialize. Fortunately, progressive policy in California and New Jersey soon presented real opportunities for meaningful project development at scale, laying the foundation for a meaningful US solar market. This catalyzing market momentum led to the genesis of First Solar, SunPower, REC Solar, and, of course, Suntech. Between ~2006 – 2009, we witnessed the emergence of 1) First Solar, 2) China-based suppliers entering the US market, and 3) rich feed-in-tariff markets in Europe and Canada; the corresponding surge in demand for modules created supply chain whiplash. The 2010’s built on this momentum at a scale and velocity that practically defined the industry for a time. The cost-curve – and the various interruptions to its onward march – were perhaps the biggest variable governing industry margins and the rate of growth. We all became (understandably) reliant on downward movement in costs and the growing diversity of component suppliers. When economic, political, or technological factors conspired to disrupt the positive momentum, it shocked the industry. These handful of events served as “stress tests” of sorts, culling the industry of projects, developers, and strategies that were not built to be resilient against sub-optimal market environments. But despite the obstacles and the dead projects/companies left in its wake, the industry as a whole has not only persisted, but managed to thrive through it all.
Figure 1. Module supply/demand ebbs and flows since 2007
1. Supply-Side Challenges
These historical events all had material impacts on projects, regardless of development stage. However, the big difference between the disruptions of the past and today’s mess is that those of the past were more narrow in nature. Whether it was “CVD I & II” or “Section 201”, one could have some comfort that the impact might, say, only apply to finished products, or merely require some patience while a manufacturer picked up operations and planted themselves somewhere else in Asia. Today, all major sub-segments are challenged, creating a string of discrete bottlenecks. Moreover, the many bottlenecks are compounding each other. This is new. The ailments are many, they are simultaneous, and they exacerbate each other.
Figure 2. Chain-wide problems
Raw Materials and Upstream Components
Polysilicon prices experienced intense – albeit brief – spikes in 2020 due to two major factory accidents. The first was at Xinjiang GCL, where a very large explosion occurred, and the second at Sichuan Yongxiang where a flood obliterated a capacity of 68,000 mega-tons. Prices calmed fairly quickly, mostly due to suppressed demand caused by the pandemic. However, in late Winter 2021, a severe demand-driven price increase (discussed below) caused polysilicon to skyrocket from under $12/kg to nearly $30/kg.  Though the price volatility was driven by the demand side, it still created supply shortages as wafer/cell manufactures flinched at the extreme COGS spike (and the ongoing pandemic rendering all future demand projections useless).
Additionally, last summer a grim picture of forced labor in China’s Xinjiang region (nearly 40% of global polysilicon capacity) started coming into focus. Uighurs and Muslims are being forced (or aggressively coerced) into working in awful conditions for little/zero pay. On June 24th of this year, an official action by the Customs and Border Patrol turned last year’s US Department of Labor red flag into tangible action, creating a “Withhold Release Order” (WRO) which empowered port officials to do exactly as the title suggests. While both the product ban and the WRO are widely applauded and not a surprise, it undoubtedly further complicates the module supply chain that was already quite strained.
Fortunately, the trend has plateaued, and even reversed slightly, in recent months, creating what appears to be a crested-and-now-slowly-receding wave.
When considering how to react (or not react) to such an extreme market phenomenon, it’s important to understand the causes. Johannes Bernreuter, an influential analyst in the world of silicon, notes: “The typical driver of such a price rally arises from market psychology: Fear of shortage leads to panic buying. In the current case, speculative traders also jumped on the bandwagon and have massively hoarded material – just to sell it at a high price at a later point. According to Chinese media, at least 20% of the polysilicon market volume is now in the hands of middlemen.” 
Since the beginning of 2020, the supply of polysilicon dipped due to the pandemic, factory accidents, forced labor concerns, and low prices driven by weak demand. And then, in the late winter and early spring, as vaccines made their way into arms and Covid-19 began to recede in many parts of the world, game-theory-induced panic buying at the other end of the supply chain (finished modules) dramatically reduced cell manufacturer price sensitivity in the middle, speculators entered the fray, and polysilicon prices tripled in just a few weeks. It was a recipe for disaster, and disaster we got.
But how many of those are fundamentals-driven factors which are likely to persist? The answer is debatable, but surely some will prove to be passing phenomena.
The raw material price movement will have a follow-on effect of taking away from CAPEX budgets for manufacturers for the next 1-2 years. Numerous top manufacturers just completed construction of new manufacturing facilities outside of China, many of which were a response to Section 201, meaning that many manufacturers are still working to optimize and leverage investments from the last supply side crisis. But any further expansion efforts are likely multiple years from making a dent on the supply chain.
The positive news here is that the (many) expansions arising out of the Section 201 tariff are not fully “baked” into the supply side of the market. Though it may feel like distant history (wasn’t that like three problems ago?), manufacturing expansions and upgrades – especially those crossing borders within multi-national corporations - take time to ramp to full efficiency and production. Even for those facilities nominally completed a couple years ago, additional lines (e.g. East Hope and GCL facilities) are often added and/or efficiency improves rapidly during the first few years of operation.
To put that in context, in 2020, Trina reported shipments of 16GW, a genuinely impressive 80% increase above 2019. Similarly, another “Solar Module Super League” (SMSL) member, JA Solar, reported shipments of 16GW in 2020, up from 10GW in 2019. And finally, the current king of the hill, LONGi, reported approximately 25GW of shipments in 2020 and expects to be at nearly 65GW(!) of module capacity by 2021. Overlaying that end-product data on top of the polysilicon dislocation might prompt mixed reactions. On one hand, one could conclude at least some level of polysilicon price pressure is likely to persist given the intense (and surely sustained) downstream demand. On the other, that data makes it pretty clear “the show will go on”, regardless of any psychology-driven panic-spikes in raw materials pricing.
One of the reasons that downstream market momentum can seem to overpower upstream phenomena is that the breadth, depth, and ambitiousness of module manufacturers continues to expand. Compared to just 10 years ago, the diversity of providers and their factory locations is jaw-dropping. This is important in the context of today’s problems because it helps the supply chain react to global demand volatility without dependence on one region’s economic, labor, or pandemic-recovery status. And, as has been the case for ~15 years, that market’s prioritization of scale, growth, and job creation continues to absorb upstream problems. Said more technically: end-product price elasticity of COGS is well below 1. This is unsustainable over the long-term of course – an industry mostly disinterested in profits is not in a “steady-state”. Yet you’d struggle to find many folks out there who think the module manufacturing industry is about to turn on a dime and start focusing on net income in the near term. The top-line focus is likely to persist and absorb upstream supply chain “shocks” for the foreseeable future.
Figure 3. Geographical Distribution of Manufacturing Capacity
The current pace and magnitude of solar PV innovation (and adoption thereof) is surprisingly underappreciated. There has long been a misconception among industry outsiders that solar is a “tech” industry. Company growth expectations (consistently too high and too parabolic) and project risk assessment (systemically overestimated by debt and equity until only recently) have demonstrated the consequences of this misconception time and time again. But in the last 3-5 years a few technological leaps have had an outsized positive impact on the industry.
From about 2006 – 2016, a developer could safely assume they would eventually procure 60 – 72 cell, monofacial polycrystalline PV modules for which form factor, performance traits, and racking system compatibility were highly predictable across providers. But in the past few years, we have seen hugely impactful changes, including:
- Monocrystalline PV cells – generally more efficient because they are cut from a single source of silicon.
- PERC/Bi-facial modules - First PERC, and then full-blown bi-facial modules have led to energy yield increases significant enough to safely exceed the materiality of the marginal cost.
- Module Capacity and Efficiency - The predictable annual 25-watt power class increases that brought us into the upper-300-watt range by 2018 recently gave way dramatically more significant leaps forward. Suddenly, 600-watt modules are fairly common. And a couple additional leaps forward seem imminent in the form of TOPCon N-type doped solar cells or the increased adoption of heterojunction technologies.
Those leaps came (and will continue to come) attached to supply-chain dislocation. Manufacturing lines need to be taken offline and updated, supply contracts need to be renegotiated, logistical kinks need to be smoothed out, etc. It just so happened that the “digestion” phase of these technological leaps landed on top of other major sources of supply-chain dislocation. Though this has contributed to a great deal of pain in 2021, I doubt anyone wishes the leaps never came.
2. Demand-side Volatility
Industry participants and even casual industry followers already know the demand-side story: Robust, secular demand growth collided with the “wait….wait…wait…..GO!!!!” cycle imposed by the pandemic.
Renewable non-carbon-based energy has quickly evolved from a “cute” dream-for-the-future to a foundational component of the long-term climate change solution. By 2026 another 160GW of new capacity is expected in the US alone; just a couple months ago we hit the cumulative 100MW installed capacity mark. The International Energy Agency recommends that, by the year 2030, ~630GW of new solar be installed globally each year, equivalent to 4x current levels. This means that the current ~800GW of existing capacity would soar to nearly 5TW (yes, it feels good to put that T in front of that W) by the end of 2030, and 15TW by 2050, representing as much as one-third of global power generation (compared to 3% as of 2020).
Energy storage is set to achieve 12,000 MWh of new installations in 2021, versus 3,500 MWh in 2020, and is on a path to build towards an annual 33 GWh market by 2026. With so many tailwinds, it’s tempting to take “the over”. As many others have written about at length, the emerging “hockey stick” in the storage sector will further bolster solar penetration in the energy mix.
Those secular trends are best appreciated at the macro level, but if one really wants to get their finger on the pulse of module demand, they need to analyze the pipeline of projects set to go under construction. Generally, and on an absolute basis, the renewable energy industry (both domestically and globally) can objectively claim 2020 a success…no pandemic-related qualifiers necessary. The industry broke installation records while employment and growth remained high. But look a little closer and things aren’t so rosy. The pandemic caused project setbacks across the board, whether it be due to broken links in the supply chain, installation labor sheltering at home, “non-essential” characterization of key development activities, permitting delays, debt/equity “pauses”, or just general pandemic-induced paralysis. In many instances, hiccups were navigated, and projects recovered reasonably well; in other cases these setbacks pushed a 2020 project into 2021 or beyond. Not all delays are “day-for-day”. And as projects go, so does module procurement. Suppliers faced commitments that were pushed back or cancelled, triggering the aforementioned domino effect throughout supply chain.
As the recovery and stabilization took place, the demand came rushing back with a hunger and velocity that outpaced that of the supply chain’s recovery, making for a terribly imbalanced market in the short term. Perhaps nothing captures the imbalance better than this simple reality: most proven EPC providers in the US market aren’t taking new business right now. That’s a new experience for every single person in the industry. Last year’s delayed projects got back on track and overlap with all those projects right on schedule to construct in 2021. Add on the motivation to deploy safe-harbor equipment, very hungry capital markets, some very good state/regional programs coming into their implementation phase… yeah, a lot of folks are building projects these days.
3. Key Considerations and Impact on Project Development
So, it’s the perfect storm of low-supply and high-demand. It happens. But how should a developer react? The answer to that question depends heavily on how long one expects the phenomena described in the previous sections to persist. Naturally, there is no knowable answer there because there are simply too many variables. However, renewable energy history suggests a likely behavioral pattern:
Anecdotal outliers lead to an overestimation of the severity of the problem.
Panic, overreaction follow.
The pendulum usually swings back quicker than initially expected.
Surely anyone working on projects placing in service in 2021 or 2022 are already thoroughly battered and bruised (or will be soon enough), and are probably not up for a lecture on playing it cool. For projects at, or nearing, the construction phase, there’s little choice but to brace yourself for the haymaker to the face. Many difficult decisions are in process or were painfully made with insufficient information. It has been, and will continue to be, brutal.
Alas [sigh]…. that’s development in a nutshell. By now many of us are getting pretty good at getting up off that proverbial mat and coping with the next “round” as best we can. Perhaps more importantly, unlike the firms constructing projects in the immediate/near term (who are icing their black eyes from that unavoidable haymaker), those focused on projects a little further off on the horizon have some choices to make.
The primary decision is whether a developer i) “locks-in” major equipment and/or contractors well in advance of construction start, ii) employs “just-in-time” procurement strategies, or iii) does something in between. The table below summarizes just a few of the common development phase procurement strategy considerations and which end of the hedge vs. “just-in-time” spectrum is favorable for that consideration. The commentary corresponds to today’s scenario: we know for sure it’s very bad right now…it’s nearly certain that things will improve with time…but improvement is not assured….and, in any event, no one knows for sure how much time is required for the situation to improve.
Figure 4. Tactics for dealing with supply chain uncertainty