PJM Interconnection’s capacity auction for the 2028/29 delivery year, published July 14, 2026, procured 138,318 MW at the $325/MW-day price cap — the third consecutive auction to clear at the ceiling — and still fell 6,831 MW short of the grid’s reliability requirement. Only 525 MW of the capacity it procured was new generation.

On July 31, 2026 PJM filed a Reliability Backstop Procurement at FERC (docket ER26-3380) to fill the gap. It offers 15-year capacity contracts covering delivery years 2028/29 through 2042/43, open only to new resources, uprates and repowerings that can show commercial operation by June 1, 2032. Bidding runs September 30 to October 21, 2026, the price cap is $555/MW-day against $325/MW-day in the regular auction, and results are due by early December. FERC’s order is expected September 29. Bilateral contracts signed between large loads and developers before the auction reduce the amount PJM buys centrally.

In its results of August 12, 2026, OPC Energy, the Tel Aviv-listed parent of US developer CPV Group, said CPV holds 70% of two PJM gas projects: Shay (2.1 GW), which has a 10-year gas supply agreement with EQT and is expected to take part in PJM’s long-term capacity auction, and Walker (1.5 GW), which has a turbine supply agreement and is negotiating a long-term power purchase agreement with a hyperscaler. OPC reported consolidated Q2 2026 EBITDA of $131 million.

Opportunity

The obvious reading is that PJM’s shortage is old news and already priced into the large independent power producers that own existing plants. That misses what is new about this auction: existing plants cannot take part. It is a contract only for capacity that does not exist yet, and a 15-year fixed capacity payment changes what a new plant is worth to a lender far more than one more year of high merchant prices does.

The binding constraint on new build is not demand, it is equipment and paperwork: gas turbines are sold out for years and a PJM interconnection position takes years to obtain. So the auction does not reward the industry evenly. It rewards the handful of developers who already hold turbines, fuel, sites and interconnection rights and can pass PJM’s feasibility review for a June 2032 in-service date.

Hypothesis: at the $555/MW-day cap, 2.1 GW of accredited capacity would be worth roughly $425 million a year for 15 years before accreditation haircuts — a large number next to a company whose whole consolidated EBITDA ran at $131 million a quarter in Q2 2026. If Shay clears anywhere near the cap, OPC Energy is being valued as an Israeli utility with a US development option while holding a contract that looks more like a regulated asset.

A separate inference: a 6.8 GW target with a three-week bidding window may clear with little competition, because so few projects can credibly meet the 2032 date, which argues for clearing prices closer to the cap than to the regular auction’s $325.

How it could play out

FERC approves the design on or near September 29, 2026 → developers with turbines and interconnection bid between September 30 and October 21 → few qualifying projects means clearing near the $555/MW-day cap → winners announce 15-year contracts in early December → projects become financeable with contracted revenue, lowering their cost of capital → listed owners of winning projects are re-rated from development-stage to contracted-infrastructure valuations, and turbine, engine and storage suppliers book the orders.

The failure path: FERC sends the filing back or orders changes, delaying the auction, or bilateral deals with data-center buyers absorb most of the 6.8 GW before the central auction runs.

Questions worth asking

• Does CPV’s Shay project clear, and at what price? One number decides whether OPC Energy’s US business goes from a development pipeline to a 15-year contracted cash flow worth hundreds of millions a year.
• How many megawatts of projects in PJM’s queue actually hold turbine slots and gas supply today — is 6.8 GW more than the credible supply, which would push clearing to the cap?
• Does the auction favour batteries, which can be built quickly and sited near load, and if so which storage integrators and cell suppliers capture the orders?
• Who pays: the costs fall on load-serving entities by zone, so which state regulators push back first, and could that political risk reach FERC?
• Which listed companies are the other side of the bilateral route — data-center owners signing 5-year-plus contracts to avoid PJM’s planned curtailment of unsupplied large loads after June 1, 2027?

Where to look

• OPC Energy (OPCE.TA) — controls CPV, whose 2.1 GW Shay project is expected to bid and whose 1.5 GW Walker project has turbines secured
• Kenon Holdings (KEN) — NYSE-listed largest shareholder of OPC Energy, the US-listed route to the same exposure
• EQT (EQT) — 10-year gas supplier to Shay, a small read-through if more gas plants are contracted
• Fluence Energy (FLNC) — battery-storage integrator, if storage takes a large share of a speed-constrained auction
• Wärtsilä (WRT1V.HE) — reciprocating-engine plants that are faster to deliver than large turbines
• GE Vernova (GEV) and Siemens Energy (ENR.DE) — turbine suppliers whose sold-out order books are the constraint that limits who can bid

Thesis check

The mechanism is clean and dated: a known shortfall, a fixed bidding window, a published price cap, and at least one listed developer on record saying its project will bid. The contract length is what matters, because it turns merchant power plants into long-duration contracted assets.

The weak link is that nothing is certain until FERC rules and the auction clears: FERC may modify the design, bilateral deals may shrink the central auction, and the $555/MW-day figure is a cap, not a price. The OPC Energy upside also depends on Shay passing PJM’s feasibility review and on OPC’s share of the project after partners.

Timing

FERC order expected September 29, 2026; bidding September 30 to October 21, 2026; results by early December 2026

Sources

Utility Dive, Aug 2026 · Syso Technologies, Aug 21 2026 · mGrid, Sep 9 2026 · PJM FERC filing ER26-3380, Jul 31 2026 · OPC Energy Q2 2026 results (PR Newswire), Aug 12 2026 · CPV, Aug 14 2026 · Morgan Downey's Commodity News, Sep 24 2026

On September 22, 2026 a spokesperson for PT Indonesia Morowali Industrial Park (IMIP) on Sulawesi told Reuters that a water shortage linked to El Niño had reduced nickel pig iron (NPI) production at some of its smelters, that tenants had been told over the preceding weekend, and that operations had not stopped and no workers had been cut so far. IMIP, majority-owned by China’s Tsingshan, houses more than 50 tenants making nickel products for stainless steel and battery materials, with about 4.2 million tonnes a year of NPI capacity according to Bloomberg. Water is used to cool and protect smelting equipment.

Shanghai Metals Market (SMM) reported that IMIP’s RKEF smelters cut operating rates by around 30–40% from September 22. It estimated a two-week disruption removes 50,000–70,000 tonnes of NPI (5,500–7,700 tonnes of contained nickel) and a longer one more than 100,000 tonnes. Bloomberg reported on September 15 that another nickel producer had halted a plant ramp-up because of the same water shortage.

The cut lands on a supply base already being squeezed by policy: Indonesia set its 2026 nickel ore mining quota at about 260–270 million wet tonnes, against about 379 million in 2025, and imported more than 11 million tonnes of Philippine ore in the first seven months of 2026 to fill the gap. NOAA’s Climate Prediction Center has given better than 90% odds of a very strong El Niño through the 2026–27 northern winter.

Opportunity

The obvious reading is that this is a two-week weather blip in an oversupplied market. SMM makes that case itself: Chinese port NPI inventories rose 52.5% between September 3 and 17 to 53,800 tonnes of contained nickel, and China’s October 300-series stainless output is expected to fall by more than 100,000 tonnes from September, so the demand cut roughly matches the supply cut for now.

What that reading leaves out is duration. The cause is not a breakdown that gets fixed but a drought produced by what forecasters expect to be one of the strongest El Niño events on record, and Indonesia’s dry season lengthens in El Niño years. Morowali is also not one smelter among many: Indonesia supplies more than half the world’s nickel, much of it from a few industrial parks, so a water constraint there is a constraint on the marginal tonne for the whole market. The newer high-pressure acid leach plants that make battery-grade nickel use far more water than RKEF lines — 200–400 cubic metres per tonne of nickel, according to Skillings.

Hypothesis: the nickel market has spent two years pricing Indonesian supply as elastic and effectively unlimited; the combination of a roughly 30% ore-quota cut and a physical water limit at the largest hub during a multi-month El Niño makes Indonesian supply less elastic than any time since the ore export ban, and producers outside Indonesia — whose costs are unaffected — are the ones that gain if the cut lasts beyond the inventory buffer.

How it could play out

El Niño keeps Sulawesi dry into late 2026 → IMIP cuts last beyond two weeks and spread to water-hungry HPAL plants → the port-inventory buffer SMM identified is drawn down → NPI and then LME nickel prices lift off their lows while Indonesia is still enforcing lower ore quotas → margins widen for ferronickel and nickel producers outside Indonesia and for stainless makers competing with Indonesian stainless output → equities priced for the 2024–2026 glut are re-rated.

The failure path: rains arrive on schedule, IMIP secures alternative water, and weak Chinese stainless demand absorbs the lost tonnes, so the episode passes without a price response.

Questions worth asking

• How long do the cuts last? SMM says anything beyond two weeks exhausts the buffer, so the date IMIP restores full operating rates decides whether this is a blip or a supply shock.
• Which Indonesian nickel operations depend on river water, lakes or hydropower rather than desalination, and are any of the HPAL battery-nickel plants already restricted?
• Does Indonesia respond by loosening ore quotas to protect smelter jobs, or by tightening them further to support price?
• Which producers outside Indonesia have idle or curtailed capacity that restarts at a modestly higher nickel price?
• Who loses: which battery-material and stainless producers are contracted to buy Indonesian intermediate products with no alternative source?

Where to look

• Nickel Industries (NIC.AX) — ASX-listed operator of RKEF lines inside IMIP, directly exposed on volume but levered to any price recovery
• PT Vale Indonesia (INCO.JK) — Sulawesi nickel producer whose smelter runs on hydropower, so it is exposed to the same drought from a different direction
• Vale (VALE) — parent of PT Vale Indonesia and owner of nickel operations in Canada and Brazil that do not face Indonesian water limits
• South32 (S32.AX) — Cerro Matoso ferronickel in Colombia, a non-Indonesian producer that gains from higher prices
• Eramet (ERA.PA) — New Caledonia ferronickel plus ore sales from Weda Bay in Indonesia
• Sumitomo Metal Mining (5713.T) — refined nickel producer with its own feed sources
• Outokumpu (OUT1V.HE) — European stainless maker that competes with low-cost Indonesian stainless

Thesis check

The chain is short and physical: no water, no smelting, and the place without water is where the marginal tonne of the world’s nickel is made, during an El Niño forecast to run into 2027 and on top of an ore-quota cut of about 30%.

The weak link is the size of the buffer. SMM’s own figures show port inventories up by half in two weeks and Chinese stainless demand falling by roughly as much as the supply cut, so a short disruption can pass with no lasting effect on price. The thesis needs the cuts at IMIP to last weeks, not days.

Timing

SMM says cuts lasting beyond two weeks from September 22, 2026 exhaust the inventory buffer; NOAA's next ENSO update is October 8, 2026

Sources

Reuters via Kitco, Sep 22 2026 · Mining Weekly (Bloomberg), Sep 22 2026 · SMM, Sep 22 2026 · SMM analysis, Sep 2026 · Bloomberg, Sep 15 2026 · Skillings, Sep 2026 · NOAA CPC ENSO diagnostic discussion, Sep 2026

A presidential proclamation signed August 6, 2026 under Section 232 imposes, from December 4, 2026, a 15% tariff on imported polysilicon and its derivatives — ingots, wafers, solar cells and modules — plus minimum import prices of $21/kg for polysilicon, $100/kg for ingots and wafers, $0.22 per watt for cells and $0.38 per watt for modules, with a specific duty covering any shortfall below the floor. Rates are 10% for the UK and capped at 15% combined with existing duties for Japan, South Korea, Taiwan, Switzerland and the EU. Companies with approved US onshoring plans can import duty-free.

For scale: SMM put Chinese n-type polysilicon at about $4.76–4.88/kg and Chinese TOPCon modules at about $0.105 per watt on August 6. Solar Power World reported in August that US-made panels averaged 31¢ per watt, Southeast Asian panels 27¢ and Indian-assembled panels 14¢. Only Hemlock Semiconductor and Wacker Chemie make polysilicon in the US; China holds 93.5% of global capacity. A Commerce temporary final rule issued September 22 and published September 24 caps weekly imports by importers registered after August 6 (for example, 55 modules a week) and lets Commerce restrict existing importers whose volumes run well above historical levels, to stop stockpiling before December 4.

First Solar closed at $172.16 on September 24, 2026, down 10.3% on the day, its lowest close in a year, 31% below its August 7 close of $250.05 and 46% below its June 3 close of $318.25. The drop followed First Solar’s withdrawal of a patent complaint at the US International Trade Commission and came as the 10-year Treasury yield rose above 5%. T1 Energy closed at $3.81 (35% below August 7) and Canadian Solar at $10.95 (31% below). Wacker Chemie closed at €87.10, 9% below its August 6 close.

Opportunity

The obvious reading is that US solar equities are falling on rates and demand, and that tariffs only raise the cost of a product whose buyers are already under pressure. That reading treats the domestic industry as one trade, which the rule does not.

The floor hurts assemblers that import cells: a $0.22 per watt cell floor raises the input cost of every US module plant running on foreign cells, which is a reasonable explanation for T1 Energy’s and Canadian Solar’s falls. It does the opposite for companies that need no imported silicon at all. First Solar’s cadmium-telluride panels use no polysilicon, and Hemlock and Wacker’s Tennessee plant are the only US polysilicon sources while the import floor sits at more than four times the Chinese price.

Hypothesis: once the $0.38 floor applies, the cheapest imported panel in the US costs more than the average American-made panel did in August, so domestic makers with domestic inputs gain room to raise prices on new contracts rather than merely defend share; the market is selling First Solar with the import-dependent assemblers and pricing the whole group on rates, rather than separating the companies the floor protects from those it taxes.

A second inference: the $21/kg polysilicon floor is the most extreme ratio in the whole schedule, so the most direct beneficiaries may be the two US polysilicon plants rather than any panel maker.

How it could play out

Imports are capped until December 4 by the anti-stockpiling rule → from December 4 imported modules cost at least $0.38 per watt plus tariff → developers sign 2027–2028 supply contracts at prices set against that floor → makers with US cells and silicon, and US polysilicon producers, lift contract prices while import-dependent assemblers are squeezed until their own cell plants run → earnings estimates for the protected group rise and the group separates from the rest of US solar.

The failure path: higher prices and 5% Treasury yields push developers to delay projects, total US installations fall, and a higher price per watt on fewer watts leaves domestic makers no better off.

Questions worth asking

• How much of First Solar’s 2027–2028 output is already sold at fixed prices, and how much is left to be priced against a $0.38 per watt import floor? That share decides whether the floor reaches earnings before 2028.
• What does Hemlock charge US buyers today, and does a $21/kg floor let it and Wacker’s Tennessee plant reprice solar-grade polysilicon for US customers?
• Which module assemblers will have approved onshoring plans before December 4, letting them import duty-free, and which will not?
• How much did developers import before August 6, and how long does that inventory delay the floor’s effect on contract prices?
• Who loses most: which US developers and residential installers carry fixed-price contracts signed on imported-panel costs?

Where to look

• First Solar (FSLR) — US maker using no polysilicon, at its lowest close in a year on September 24, 2026
• Wacker Chemie (WCH.DE) — one of two US polysilicon producers through its Tennessee plant; also a semiconductor-grade supplier
• Corning (GLW) — majority owner of Hemlock Semiconductor, the other US polysilicon maker, though solar is a small part of Corning
• Hanwha Solutions (009830.KS) — parent of Qcells, which is building integrated ingot-to-module production in Georgia
• T1 Energy (TE) and Canadian Solar (CSIQ) — module makers that depend on imported cells until their own cell plants run, the side the cell floor hurts

Thesis check

The price arithmetic is published and dated: the module floor is above the reported average US panel price, the polysilicon floor is more than four times the Chinese price, and the rule starts December 4, 2026. The sell-off since August has hit protected and exposed companies almost equally.

The weak link is demand and contract timing. First Solar sells much of its output years ahead at fixed prices, so a higher floor may not show up for a long time, and 5% Treasury yields raise the financing cost of every solar project, which could shrink the market faster than the floor raises prices.

Timing

Section 232 tariffs and minimum import prices take effect December 4, 2026; Commerce's anti-stockpiling limits apply until then

Sources

White & Case, Aug 2026 · pv magazine USA, Aug 7 2026 · Solar Power World, Aug 2026 · SMM analysis, Aug 2026 · Federal Register (Justia), Sep 24 2026 · Mondaq (Baker Botts), Sep 2026 · Benzinga, Sep 24 2026