In May, the renewable energy indices experienced a significant rally, particularly in the solar space. The Invesco solar ETF (TAN) surged by as much as 20% despite mixed industry news flow and the bond market suggesting that interest rates will stay higher for longer. What fueled this sudden wave of optimism?
We believe the primary driver is the expectations of soaring power demand from AI data centres. In particular, a sell-side upgrade of First Solar (FSLR) argued convincingly that the company with its domestic supply chain and large US market share is a “first derivative” play on the AI boom. As FSLR is a relatively large clean tech company with a market capitalization of around USD 30bn, it is the largest constituent in many of these indices, particularly in the TAN where it with its 16% weight is the largest by a margin. As the stock gained 54% during the month, it pulled up the indices.
It was not all about AI demand though as FSLR is also benefiting from other factors. In late April, a group of US solar manufacturers, led by FSLR, petitioned the Department of Commerce (DOC) to open a new anti-dumping and circumvention (AD/CVD) case against Chinese companies operating through Cambodia, Malaysia, Thailand, and Vietnam. These four countries account for more than 80% of imported panels into the US, and the petitioners are seeking tariffs of up to 270%. Although the final tariffs, if any, are likely to be meaningfully lower but still significant, most experts believe the petitioners are likely to prevail. Moreover, there is a significant risk that tariffs will be retroactive to at least April.
Recall that when the DOC took up a similar AD/CVD case in 2022, share prices slumped as the supply chain froze until President Biden enacted a two-year moratorium on retroactive tariffs. This moratorium ends at the close of this month and we doubt that Biden, in an election year, is willing to introduce a new moratorium that will help Chinese manufacturers and harm domestic producers.
The obvious winners from the new AD/CVD case are FSLR and other smaller domestic producers of solar panels. For everyone else in this industry, including the solar developers, the impact is negative. This is especially true for Chinese manufacturers, which typically sell less than 10% of their total volumes in the US, but due to the high US panel prices derive up to 30% of their total revenues from the country.
While the AD/CVD petition clearly helped to boost FSLR’s stock price during the month, the other solar stocks most likely went along for the AI ride. Surprisingly, even several Chinese solar manufacturers, collectively accounting for around 10% of the TAN index, saw gains of about 20% in May despite the new AD/CVD case. This could be attributed to speculations that the Chinese government might take steps to address the significant overcapacity in China’s solar manufacturing industry. In the last 12-18 months, panel prices have dropped by almost 50% and are in many instances selling below costs.
We discussed the increased power demand from AI data centers (DCs) in our March-24 report “Roadblocks on the AI highway”. As the hyperscalers (large tech companies) constructing large DCs have either 100% renewable energy targets or net zero commitments, the incremental power will have to be derived from renewable energy. Even if the hyperscalers initially should use gas or connect to grids partly run on fossil fuels, they will need to offset the additional carbon by investing in more renewable energy. As solar is carbon-free and has the lowest cost of levelized energy, this is an obvious tailwind for the solar industry.
Strengthening this view, the Electric Reliability Council of Texas (ERCOT) in its latest Capacity, Demand, and Reserves report substantially increased the expected average load growth for Texas for 2025-2029. The main reason was the many new manufacturing facilities and DCs planned in Texas. However, BNP Paribas calculates that this load requirement could boost average annual demand for new solar PV installations by more than 80% to 26GW vs their previous estimate of 14GW p.a. If the rest of the US should follow Texas’ upgrade, BNP estimates that US annual average demand for 2025-2029 would be 70GW instead of 57GW, more than 20% higher than last year’s estimate.
Surely, all signs are pointing toward a solar industry where growth will explode? Over time, we believe the outlook appear bright, however the near-term prospects are more uncertain. The challenge for the solar industry is not a lack of demand, but the bottlenecks are rather grid connections, electrical equipment, permitting and available labour. For instance, demand for grid connection has for years overwhelmed grid capacity and Berkely Labs, a national research laboratory, estimated that 2,600GW of power generation and storage assets were waiting for interconnection approval at the end of 2023. This is about twice the power of all current US power plants on a name plate capacity basis.
Significant steps are taken to clear up the grid bottlenecks though. During May, three separate but related transmission reforms were released by the US federal government. First, the US Department of Energy (DOE) proposed ten new “national interest electric transmission corridors” which allows the federal government to accelerate projects in areas where consumers are harmed by the lack of transmission. Barely a week later, the Federal Energy Regulatory Commission (FERC) released Order 1920 and 1977 which both aim to improve permitting times for construction of the transmission grid with new rules around long term planning and cost allocation between states. Together this is a significant development and has been touted by some as the most important energy legislation since the Inflation Reduction Act (IRA).
While there is no immediate impact in the growth of utility scale solar installations, the effect of AI power demand is already visible in higher power purchase agreements (PPAs) for solar installations post 2026. This is obviously good news for renewable energy developers and FSLR is again a clear winner. The company had already sold its US capacity for the next three years but will now see increased demand for its post 2027 capacity. More demand combined with less competition due to AD/CVD means increased pricing power, and FSLR has already signaled that panel prices for post 2027 capacity is 15-20% higher than today’s prices vs the expectation only a few months ago that prices would decline by 15-20%.
The outlook for the rest of the utility scale solar sector is more uncertain. Key questions remain: How many projects will be delayed due to AD/CVD uncertainty? How quickly can the many bottlenecks be resolved, and will the AI driven demand boost last? Although we expect near-term fundamentals to be soft, we acknowledge that the strong May performance might signal that investors are willing to look beyond some weak quarters into a stronger 2025 and 2026.
We continue to closely monitor developments in the solar space but remain selective in our investments. Instead, we prefer companies that are actively involved in addressing grid capacity bottlenecks, as they are likely to play a key role in the growth in both solar, AI and many other industries.
