top of page
Search

Global Solar Market Slowdown 2026–2027: Investment Risks, Manufacturing Overcapacity and the Next Growth Markets

Writer: Green Fuel Journal
Green Fuel Journal
1 day ago
49 min read

Updated: 3 minutes ago

By the Green Fuel Journal Research & Intelligence Team

Published 10 November 2026 under the ISSN 2979-3777 | Written for energy investors

Last reviewed: November 2026

Next review trigger: 4 December 2026, when the US Section 232 tariffs and import price floors take effect


Legal Disclaimer

This report is published for information and research purposes only. It is not investment, financial, legal or tax advice, and nothing in it is an offer or a recommendation to buy, sell or hold any security or asset. Forecasts and scenarios can change, and several figures come from third parties, including trade press, whose data we could not always check against primary documents. Readers should carry out their own due diligence and consult a licensed adviser.


Journal GreenFuel cover showing 8% solar installations in 2026, a red down arrow, and headline on global solar market slowdown.

Executive Summary

The global solar market is heading for its first annual decline in more than two decades, with SolarPower Europe (SPE)'s medium scenario putting 2026 additions at about 612 GW, down 8% from a record 664 GW in 2025. China explains most of the fall: additions for January to September 2026 were 110.83 GW (AC), 54% below the same period of 2025, which closed at 315.07 GW. The financial damage sits in manufacturing: the six largest Chinese module and wafer makers lost about CNY 25 billion in the same nine months. BloombergNEF's October forecast is lower than SPE's at 584 GW (DC), down 10.8%. Outside China the picture splits. Wood Mackenzie expects India to be the only top market with higher deployment in 2026, and Ember estimates African solar installs rising 45% to 17 GW [SINGLE-SOURCE]. Deployment is shifting between markets rather than disappearing.


Those losses, roughly US$3.7 billion in total according to Yicai Global, came as polysilicon inventory reached a record 600,000 tonnes in August. Daqo reported polysilicon prices recovering to RMB 40–43/kg by the end of Q3, but Caixin counted under 10% of capacity cut by March against a one-third target [SINGLE-SOURCE]. The evidence leans toward price support rather than capacity removal; that reading is Green Fuel Journal's inference, not a sourced finding [GFJ INFERENCE]. India shows the same pattern at smaller scale: 233 GW of module capacity running at 35–40% utilisation, and combined US duties on Indian cells of 249.13% in Commerce's final determinations.


For investors, revenue quality now matters more than irradiance. In April 2026, solar's capture factor fell to about 0.10 in France and 0.26 in Germany, according to Pexapark [SINGLE-SOURCE]. Lazard's July 2026 analysis puts unsubsidised US solar-plus-storage at $61–156/MWh against $40–98/MWh for utility solar alone [SINGLE-SOURCE]. In growth markets the limits are grids and finance: only 18% of ASEAN solar and wind projects had multilateral or concessional funding [SINGLE-SOURCE], and Brazil curtailed 17.8% of potential wind and solar output in 2026 up to 10 May. Bankable offtake and grid-ready sites matter more to returns than falling module prices.


1. Executive Intelligence Synthesis

Direct answer: The 2026 slowdown is a China slowdown, and five facts carry the evidence. SolarPower Europe's medium case has 2026 global additions at 612 GW, 8% below 2025 (DC basis). China drives the fall: 110.83 GW (AC) in January–September 2026, 54% lower. Six Chinese manufacturers lost about CNY 25 billion in the same nine months. Revenue quality is weakening, with April 2026 capture factors at 0.10 in France. Growth is shifting to India and Africa, where grid access and finance set the pace.


1.1 Five Executive Signals


Signal 1. The 2026 decline is a China decline.

  • FINDING: China added 110.83 GW (AC) in January–September 2026 against 240.27 GW a year earlier, and the China Photovoltaic Industry Association (CPIA) forecast in February 2026 a full-year 180–240 GW (AC) after 315.07 GW in 2025 (NEA via TaiyangNews, 27 October 2026; PV Tech, 11 February 2026) [TWO-SOURCE].

  • SO WHAT: Any global forecast now rests on one country's policy reset, so a revision in China moves the whole total.

  • NOW WHAT: Check whether a forecast states China separately and on which basis, AC or DC, before comparing it with another.

 

Signal 2. Losses sit in manufacturing.

  • FINDING: Six Chinese makers lost about CNY 25 billion in the first nine months of 2026 (Yicai Global, 3 November 2026), and polysilicon inventory reached 600,000 t in August (BloombergNEF via The Edge Malaysia, 2 November 2026).

  • SO WHAT: Cheap equipment helps developers, but weak suppliers raise counterparty risk.

  • NOW WHAT: Compare the maker behind any module warranty with the loss figures in Table 5.1 before accepting long-dated claims.

 

Signal 3. Anti-involution has lifted prices more than it has removed capacity.

  • FINDING: China's anti-involution campaign, its push to end loss-making price competition, had cut under 10% of polysilicon capacity by March 2026 against a one-third target, yet prices recovered to RMB 40–43/kg by the end of Q3 2026 (Caixin, 5 March 2026; Daqo, 29 October 2026) [GFJ INFERENCE].

  • SO WHAT: Prices held by agreement rather than closures can break, reopening the loss cycle.

  • NOW WHAT: Test any investment case against a scenario in which the RMB 43,000/t silicon floor fails.

 

Signal 4. Revenue quality is the new risk.

  • FINDING: Pexapark reported April 2026 capture factors, the share of the average market price that solar output actually earns, of 0.10 in France and 0.26 in Germany, down from 0.42 and 0.40 a year earlier (12 May 2026) [SINGLE-SOURCE], while Sunrun reported a 75% storage attachment rate in the third quarter (PV Tech, 5 November 2026).

  • SO WHAT: A project's yield no longer sets its revenue; the share sold at fixed prices, and storage, now matter more.

  • NOW WHAT: Require the contracted-versus-merchant split and storage plan, and run the Section 4.4 worked example with your own inputs.

 

Signal 5. Growth is real but relocated and constrained.

  • FINDING: Ember estimates African solar installs at 17 GW in 2026 [SINGLE-SOURCE], yet only 18% of ASEAN solar and wind projects had multilateral or concessional finance [SINGLE-SOURCE], and Brazil's grid operator curtailed 17.8% of potential wind and solar output up to 10 May 2026.

  • SO WHAT: In new markets, grid access and the cost of finance limit returns, not sunshine or module prices.

  • NOW WHAT: Check grid-connection status and curtailment terms in every offtake agreement before comparing tariffs across markets.

 

Table 1.1: Five signals at a glance

#

Signal

Headline number

Source

Evidence label

1

China drives the 2026 fall

110.83 GW (AC), January–September 2026, -54%

NEA via TaiyangNews, 27 Oct 2026

[SINGLE-SOURCE]

2

Losses sit in manufacturing

About CNY 25bn nine-month net loss, six makers

Yicai Global, 3 Nov 2026

Press-reported results, not checked against filings

3

Prices lifted, capacity mostly intact

Under 10% of capacity cut by March 2026

Caixin, 5 Mar 2026

[SINGLE-SOURCE] and [GFJ INFERENCE]

4

Revenue quality is weakening

Capture factor 0.10 (France), 0.26 (Germany), April 2026

Pexapark, 12 May 2026

[SINGLE-SOURCE]

5

Growth is relocated and constrained

17 GW Africa (2026 estimate); 18% ASEAN MDB share

Ember, 2 Sep and 7 Oct 2026

[SINGLE-SOURCE]

Methodology note: The signals are an editorial selection, not a scored ranking. We chose five findings that (a) carry a verified figure from at least one named source, (b) change what an investor would do with capital, and (c) are not simply restatements of the installation forecast. Each signal carries an evidence label: TWO-SOURCE where two independent data owners report it, SINGLE-SOURCE where one does, and "inference" where Green Fuel Journal draws a conclusion the sources do not state.

Sources as listed in each row; full URLs are in References & Strategic Sources.


1.2 How to read the evidence

  • FINDING: SolarPower Europe, Ember and BloombergNEF count capacity in direct current (DC), while IRENA and China's NEA count in alternating current (AC), which is why 664 GW (SPE) and 511 GW (IRENA) both describe 2025 global additions (SPE, 22 June 2026; IRENA, 2 April 2026).

  • SO WHAT: A comparison across bases shows a gap of 153 GW (our arithmetic, 664 minus 511) that is a definition, not a market difference.

  • NOW WHAT: Confirm the basis, DC or AC, before using any installation figure in a model.

 

Three labels run through the report:

[TWO-SOURCE] for two independent data owners,

[SINGLE-SOURCE] for one owner even when several outlets repeat it, and

[GFJ INFERENCE] for a conclusion we draw that the sources do not state. Most figures reached us through trade press, so the Research Limitations statement before the Legal Disclaimer applies throughout. The report claims no global levelised cost, breakeven price, investor return or scenario probability, because no verified source supplies them.


2. Macro Context & Strategic Drivers

Direct answer: The global solar market is contracting in 2026 mainly because China's additions are falling after a policy-driven record year. SolarPower Europe expects 612 GW (-8%) and BloombergNEF 584 GW (-10.8%), both on a DC basis. China added 110.83 GW (AC) in January–September 2026, 54% below 2025. US credit expiry and weaker European rooftop demand add smaller drags, while higher gas prices after 1 March 2026 supported first-half demand.


2.1 Why the forecasts disagree

  • FINDING: Four publishers put 2025 global additions at 664 GW (SolarPower Europe, 22 June 2026), 655 GW (BloombergNEF, 16 October 2026), 647 GW (Ember, as reported by TaiyangNews on 21 April 2026) and 511 GW (IRENA, 2 April 2026), and only IRENA reports in AC.

  • SO WHAT: The spread is mostly definition, so the useful comparison is each publisher's change from 2025 to 2026, not the absolute level.

  • NOW WHAT: Compare growth rates within one publisher, and treat any cross-publisher gap as a basis check first.

 

The 2027 outlook is less settled. TaiyangNews reported on 16 October 2026 that BloombergNEF expects 618 GW (DC) in 2027. On 2 November 2026, Bloomberg's syndicated report, carried by The Edge Malaysia, said BloombergNEF had cut its 2026 figure to 585 GW from 624 GW and described 2027 as "weaker than previously estimated". The sources do not say whether 618 GW is the revised or the earlier number, and the 2026 figure differs too (584 GW against 585 GW). SPE expects growth to resume in 2027, and CPIA expects recovery without giving a figure, so a 2027 recovery is a forecast, not a consensus fact.



2.2 China: the policy reset and the demand drop

  • FINDING: China added 72 GW (AC) in the first half of 2026, 66% below the first half of 2025, then recovered from July, with 14.08 GW in July (+28%) and 13.64 GW in September (+41.2%) (Caixin, 24 July 2026; NEA via PV Tech, 25 August 2026; NEA via TaiyangNews, 27 October 2026) [TWO-SOURCE].

  • SO WHAT: Monthly gains come off a weak base, so they show stabilisation, not a return to 2025 volumes.

  • NOW WHAT: Track monthly NEA additions against the CPIA range before accepting any full-year forecast above it.

 

China set a record of 315.07 GW (AC) in 2025 [TWO-SOURCE]. One headline from DealStreetAsia links the 2026 fall to the unwinding of a rush before the May 2025 pricing reform. That reform is Document 136, a notice from the National Development and Reform Commission and the NEA on market-based pricing for new-energy feed-in tariffs, effective 1 June 2025. Its text is not cited here, so this report does not describe how it works, and the link between the reform and the rush rests on that one headline. CPIA's February 2026 forecast of 180–240 GW (AC) for 2026 would be the first annual decline since 2019. Forecasts conflict: PV Tech cited 160–200 GW in August 2026, and a pv magazine headline mentioned "up to 287 GW", with no supporting text cited.

Ember recorded record Chinese exports of 68 GW in March 2026, the month before the export VAT rebate on PV products ended on 1 April 2026; no source we read ties the two together.


2.3 United States: credit cliff, tariffs and security rules

  • FINDING: SEIA and Wood Mackenzie put 2025 US installs at 43.2 GWdc (-14%), then reported a second-quarter 2026 rebound to 11.4 GWdc (+45%), led by 9.6 GW of utility-scale solar [SINGLE-SOURCE].

  • SO WHAT: The rebound hides a split, because residential installs fell to 995 MW, a five-year low, while utility developers race a credit deadline.

  • NOW WHAT: Read utility and residential data separately, and ask for the safe-harbour status of any US project.

 

The 2025 reconciliation law (OBBBA) ended the residential 25D credit after 2025. The 4 July 2026 begin-construction deadline for the 45Y and 48E credits has passed, so projects that missed it must be in service by 31 December 2027. Treasury Notice 2025-42 (15 August 2025) tightened what counts as beginning construction, and Notice 2026-15 (12 February 2026) set foreign-entity material-assistance rules. A Section 232 proclamation signed 6 August 2026 takes effect on 4 December 2026, adding a 15% tariff on ingots, wafers, cells and modules and minimum import prices that include $0.38/W for modules, as reported by White & Case [SINGLE-SOURCE]. The proclamation text itself is not cited here, so the figure rests on that one summary.


Wood Mackenzie's base case holds US installs near 44 GWdc a year through 2031, with a safe-harboured utility pipeline of 216–240 GW. All of these US volumes trace to one data owner, SEIA and Wood Mackenzie, and outlets that repeat the report do not confirm it.

"This stability reflects resiliency amidst structural market constraints,"

said Caitlin Connelly, Senior Analyst US Solar at Wood Mackenzie, on 10 September 2026.


2.4 EU and UK: volume holds, value falls

  • FINDING: SolarPower Europe reported EU installs of 65.1 GW in 2025 against 65.6 GW in 2024 (-0.7%), while the rooftop share of new capacity fell from 28% in 2023 to 14% in 2025 (SPE, 11 December 2025) [SINGLE-SOURCE].

  • SO WHAT: Volume is holding while the mix moves toward utility-scale assets that depend on market prices rather than rooftop support.

  • NOW WHAT: Check the rooftop-to-utility split of any European platform, because the two segments face different policy risk.

 

SPE's June 2026 global report gives a different EU-27 figure for 2025, 67.2 GW (about +1%); the two numbers come from the same publisher and the sources do not reconcile them. In the first half of 2026 the EU installed 33.8 GW against 33.2 GW, which SPE attributed to heatwaves and energy-supply concerns rather than policy.

"Europe should not need an energy crisis to make the case for solar,"

said Walburga Hemetsberger, Chief Executive Officer of SolarPower Europe, on 8 September 2026. SPE still expects a small contraction in 2026 and 2027.


Policy is trimming demand support and adding supply-side conditions. Germany's cabinet approved an EEG reform on 29 July 2026 that phases out small-PV feed-in tariffs, though it is not yet law. The EU's funding ban on high-risk inverters could affect about 14% of 2026–2030 demand and add 2–8% to project costs, according to Wood Mackenzie, a single source [SINGLE-SOURCE].


Wider EU rules on renewable fuels are covered in our analysis of the RED III RFNBO targets to 2030. In the UK, the CfD Allocation Round 7 awarded 4.9 GW of solar at £65.23/MWh (2024 prices) on 10 February 2026, and cumulative capacity reached 23.2 GW.


2.5 Where demand is moving

  • FINDING: Wood Mackenzie said on 4 June 2026 that India would be the only top global market with higher solar deployment in 2026 than in 2025.

  • SO WHAT: Growth capital is moving toward markets with domestic-content mandates and large unmet demand, not toward the largest historical markets.

  • NOW WHAT: Read Sections 3 and 7 for the market-by-market evidence behind that shift.

 

Wood Mackenzie gave no gigawatt figures because its report is paywalled. Section 3 tests the India claim, and Section 7 compares India with three other markets.


Table 2.1: Regional snapshot

Region

Installs and basis

Direction

Key policy

Evidence label

United States

43.2 GWdc (2025); 11.4 GWdc (Q2 2026)

2025 -14%; Q2 2026 +45%

OBBBA credit cliff; Section 232 from 4 Dec 2026

[SINGLE-SOURCE]

European Union

65.1 GW (2025); 33.8 GW (H1 2026)

2025 -0.7%; H1 2026 up from 33.2 GW

EEG reform (not yet law); inverter funding ban

[SINGLE-SOURCE]; conflicts with 67.2 GW

United Kingdom

23.2 GW cumulative (Aug 2026)

291,000 new installations in 12 months

CfD AR7: 4.9 GW at £65.23/MWh

AR7 price [TWO-SOURCE]; installs secondary

China

315.07 GW AC (2025); 110.83 GW AC (9M 2026)

9M 2026 -54%

Document 136 (text not retrieved); export VAT rebate ended 1 Apr 2026

2025 [TWO-SOURCE]; 9M [SINGLE-SOURCE]

India

171.05 GW cumulative (30 Sep 2026); 35.2 GW (9M 2026)

9M 2026 +19.5%

ALMM Lists I–III; PM Surya Ghar

[SINGLE-SOURCE]

Emerging markets

Africa 17 GW (2026 estimate); Brazil 10.6 GW (2026 forecast)

Africa +45%; Brazil 2025 -24%

Brazil curtailment rules

[SINGLE-SOURCE]; estimate and forecast

Sources: SEIA/Wood Mackenzie via Mercom; SolarPower Europe; GOV.UK; NEA via TaiyangNews; MNRE via SolarQuarter; Ember; ABSOLAR via Canal Solar. URLs in References.


Table 2.2: Where sources disagree

Item

Version A

Version B

Why it matters

BloombergNEF 2027 global

618 GW DC (TaiyangNews, 16 Oct 2026)

2027 "weaker than previously estimated"; 2026 cut to 585 GW (Edge Malaysia, 2 Nov 2026)

A 2027 recovery is not a consensus fact

EU 2025 additions

65.1 GW, -0.7% (SPE, 11 Dec 2025)

67.2 GW, about +1% (SPE via TaiyangNews, 22 Jun 2026)

Same publisher, opposite direction

China 2026 forecast

180–240 GW AC (CPIA, Feb 2026)

160–200 GW (PV Tech, Aug 2026); "up to 287 GW" (headline)

Range sets the size of the fall

India module capacity

233 GW (IEEFA/JMK, Jun 2026)

261.7 GW (Mercom); 225.5 GW ALMM-enlisted

Different definitions change utilisation

India solar-plus-storage tariff

INR 2.86/kWh (auction, Oct 2025)

INR 3.52–3.53/kWh (CERC-approved)

Unexplained gap in the same package

 Sources as listed; no reconciliation exists in the evidence, so both versions stand.


3. India-Specific Analysis

Direct answer: India's solar market is both growing and oversupplied. Cumulative solar capacity reached 171.05 GW on 30 September 2026, and about 35.2 GW was added in January–September 2026. Yet India's module factories, at 233 GW of capacity, ran at only 35–40% utilisation in June 2026, according to IEEFA and JMK Research. Demand is real, but manufacturing capacity has outrun it, and unsigned power sale agreements and transmission limits cap the returns on new projects.


3.1 Deployment: a record year, then slowing utility additions

  • FINDING: India added about 44.6 GW in fiscal year 2025-26 (+87.2%) and 35.2 GW in January–September 2026, but utility-scale additions slowed from 12.12 GW in the first quarter to 6.87 GW and then 5.4 GW (JMK Research, 10 April and 9 October 2026) [SINGLE-SOURCE].

  • SO WHAT: Rooftop and utility segments are diverging, so a headline total hides the segment that sets a large project's risk.

  • NOW WHAT: Compare utility-scale additions on their own when sizing demand for ground-mounted projects.

 

JMK splits the January–September 2026 total into 24.4 GW of utility-scale, 9.8 GW of rooftop and 1.04 GW off-grid solar. Gujarat (12.9 GW) and Rajasthan (9 GW), counting solar and wind together, supplied about 55% of additions. The Ministry of New and Renewable Energy (MNRE) data, reported by SolarQuarter on 21 October 2026, show 3.011 GW added in September alone. Calendar 2025 totals differ by source and definition: JMK 37.9 GW, Mercom 36.6 GW (a headline figure), JMK utility-only 28.6 GW, and SolarPower Europe 45.7 GW on a DC basis.



3.2 Module overcapacity and the upstream gap

  • FINDING: IEEFA and JMK Research put India's module capacity at 233 GW in June 2026, running at 35–40% utilisation against the 50–65% they say factories need, while Mercom counts only about 36.6 GW of cell capacity [SINGLE-SOURCE].

  • SO WHAT: Revenue assumptions for Indian module makers should reflect utilisation of roughly a third to two-fifths of nameplate, not full capacity.

  • NOW WHAT: Check which capacity definition a company uses, installed, annualised or ALMM-enlisted, before comparing utilisation.

 "India has added module capacity faster than the market can absorb it,"

said Prabhakar Sharma, Senior Consultant at JMK Research, in the IEEFA and JMK release of 1 September 2026. About 135 GW more module capacity is planned, and the report counts incremental demand to 2030 of only 17–22 GW from data centres, green hydrogen and ammonia, and exports.


Our coverage of global green ammonia bunkering infrastructure tracks one of those outlets separately.


Charith Konda, Lead Energy Specialist at IEEFA, said in the same release on 1 September 2026:

"Sustained export growth depends on closing the cost and technology gap with China."

The upstream gap matters because policy now requires domestic content. The Approved List of Models and Manufacturers (ALMM) List-I covers modules, List-II covers cells with effect from 1 June 2026 and the Ministry did not grant a blanket extension, and List-III will cover ingots and wafers from 1 June 2028. Table 3.1 shows why the capacity figures differ.


Table 3.1: India module and cell capacity: three definitions

Definition

Value

Source

Date

What it measures

Installed capacity, IEEFA/JMK

233 GW modules; utilisation 35–40%

IEEFA/JMK

Jun 2026 (report 1 Sep 2026)

Installed factory capacity and how much of it runs

Annualised capacity, Mercom

261.7 GW modules; 36.6 GW cells

Mercom via PV Tech

Jun 2026 (28 Sep 2026)

Capacity counted on an annualised basis

ALMM-enlisted capacity

225.5 GW List-I modules; about 35.5 GW List-II cells

ALMM enlistment as reported by PV Tech and IEEFA

Sep 2026

Capacity eligible for domestic-content projects

Sources: IEEFA (1 Sep 2026); PV Tech reporting Mercom (28 Sep 2026). Evidence label: SINGLE-SOURCE for each definition; definitions are unreconciled.


3.3 US trade exposure

  • FINDING: Commerce's final determinations, announced on 11 September 2026, set combined US duties of 249.13% on Indian solar cells (123.04% antidumping plus 126.09% countervailing), and the US took about 97% of Indian module exports in fiscal 2026 (PV Tech; IEEFA/JMK) [SINGLE-SOURCE].

  • SO WHAT: Export-led Indian capacity has lost most of its largest market, which pushes utilisation lower still.

  • NOW WHAT: Separate export-dependent revenue from domestic-offtake revenue in any Indian manufacturer's order book.

 

IEEFA records Indian exports to the US down 44–47% from the fiscal 2024 peak, with combined duties above 200% on most makers. Adani and its Mundra unit are named in the determinations. The International Trade Commission's final vote (scheduled for 14 October 2026) and the final orders (scheduled for 2 November 2026) could not be found, so the outcome is unverified. The 249.13% figure is Commerce's announced determination, not a confirmed final order.


3.4 Offtake, grid and curtailment risk

  • FINDING: About 43.9 GW (43,942 MW) of awarded capacity had power sale agreements (PSAs) pending at 30 September 2025, the latest figure available and more than 12 months old, while 24,928 MW of PSAs have been signed since April 2023 [SINGLE-SOURCE].

  • SO WHAT: Awards without signed PSAs carry offtake risk that auction tariffs do not reveal.

  • NOW WHAT: Require the PSA status of any awarded project, with the date of that status.

 

Awards linked to PSAs can be cancelled case by case, and the Solar Power Developers Association cited more than 50 GW as stranded in August 2025, a figure older than 12 months. No update after September 2025 was found. Grid limits add to the risk. Ember reported about 470 GWh of renewable curtailment, meaning generation switched off, in the first quarter of 2026, with about 300 GWh from transmission limits.

"Transmission constraints accounted for nearly two-thirds of all renewable energy curtailment,"

said Duttatreya Das, Energy Analyst Asia at Ember, on 19 May 2026. India lost around 34 GWh on 30 March 2026 alone.


The interstate transmission (ISTS) charge waiver is also phasing down, to zero after 30 June 2028, according to an advisory blog that contradicts itself on one step and cites no Ministry order.

Pralhad Joshi, Union Minister of New and Renewable Energy, told PTI on 4 November 2026:

"We are on track; I do not see any problem,"

on the 500 GW target for 2030. S&P Globa

l's view that the target arrives by 2032 stands as the counterview.


3.5 Tariffs and storage economics

  • FINDING: Plain solar tariffs cleared at INR 2.42–2.54/kWh in 2026, while firm supply costs far more: FDRE-IX bids came in at INR 5.99–6.00/kWh (24 August 2026) and round-the-clock (RTC) bids from INR 5.25/kWh in two headline-level reports [TWO-SOURCE].

  • SO WHAT: Buyers pay roughly double the plain-solar price for reliability (our arithmetic on the two ranges), which sets the premium that storage-backed projects can earn.

  • NOW WHAT: Compare tariffs only within the same product type.

 

FDRE stands for firm and dispatchable renewable energy. CERC adopted tariffs of INR 2.42–2.54/kWh for Azure Power's SECI projects on 2 November 2026 (SolarQuarter, 4 November 2026). The Ministry of Power reported standalone battery bids of about INR 2.1 per unit without viability-gap funding at two cycles a day, against INR 10.18 in fiscal 2022-23 [TWO-SOURCE]. For solar plus storage, the SECI auction cleared at INR 2.86/kWh in October 2025 while CERC approved INR 3.52–3.53/kWh for the same package; the gap is unexplained. Wood Mackenzie expects Indian utility-scale system prices to rise about 20% by the fourth quarter of 2026 on domestic cell supply [SINGLE-SOURCE].


3.6 What it means for investors

  • FINDING: PM Surya Ghar passed 50 lakh installations (14.8 GW) in August 2026, while PM-KUSUM stood at 13.11 GW on 31 March 2026 against a 34.8 GW target (Business Standard, 4 August 2026; Energetica India, 26 July 2026).

  • SO WHAT: Government-backed rooftop demand is delivering while the farm-pump scheme lags, so scheme exposure is not uniform.

  • NOW WHAT: Check which scheme or tender sits behind a company's order book before treating India's demand as one market.

 

PM Surya Ghar targets 1 crore households by 2026-27. For PM-KUSUM, official pump counts are unreconciled (1.1 million against 28.4 lakh). Two manufacturers show how the oversupply and export risks play out in company results, Waaree and Premier Energies, and Section 5.4 sets out their figures. The risk register in Table 3.2 pulls the India evidence into one view. What we could not verify: the ITC outcome, current distribution-company (DISCOM) dues, and any official 2027 forecast.


Table 3.2: India risk register

Risk

Evidence

Date

Evidence note

Module overcapacity

233 GW at 35–40% utilisation; about 135 GW more planned

1 Sep 2026

[SINGLE-SOURCE]

US trade exposure

249.13% combined duty on cells; US took about 97% of module exports in FY2026

11 Sep 2026

ITC outcome not found

Unsigned PSAs

43.9 GW of awards with PSAs pending

30 Sep 2025

More than 12 months old

Transmission and curtailment

About 470 GWh curtailed in Q1 2026, about 300 GWh from transmission

19 May 2026

[SINGLE-SOURCE]; state-level data not found

Cost inflation

System prices to rise about 20% by Q4 2026 on ALMM-II cell supply

26 Sep 2026

[SINGLE-SOURCE]; secondary

Tender participation

A Power Management Institute official, as quoted by PV Tech, said tender cost estimates were too low to draw broad participation

17 Nov 2025

Second-hand; older than 12 months

Methodology note: The register lists India's main risks with the evidence and date for each. It assigns no score and no ranking. The last column shows either the evidence label or the limit on the data, such as age or source type, so a reader can judge how far each line can be relied on.

Sources: IEEFA/JMK; PV Tech; Saur Energy; Ember; GreentechLead. URLs in References.


4. Operational / Technical Deep-Dive

Direct answer: Overcapacity reaches investment returns in three steps. Polysilicon inventory hit a record 600,000 t in August 2026, and six Chinese makers lost about CNY 25 billion in nine months, which keeps equipment cheap, with global modules at $0.11–0.12/W against a $0.09/W record low in 2025. Cheap equipment then feeds more solar into the same hours, and April 2026 capture factors fell to 0.10 in France, cutting what each megawatt-hour earns.


4.1 The manufacturing chain: capacity, output and inventory

  • FINDING: China's polysilicon capacity was about 3.5 million t a year at the end of 2025, with Tongwei, GCL-Tech, Daqo and Xinte holding 56.7%, while 2025 output fell 26.4% for polysilicon and 12.3% for wafers (PV Tech, 16 October 2026; Caixin, 5 March 2026) [SINGLE-SOURCE].

  • SO WHAT: Output restraint exists, yet inventory kept building from above 500,000 t in March 2026 to 600,000 t in August, more than six months of global wafer production.

  • NOW WHAT: Compare the inventory trend, not the price alone, when judging whether supply is tightening.

 

The four largest producers hold roughly 900,000 t (Tongwei), 480,000 t (GCL-Tech), 305,000 t (Daqo) and 300,000 t (Xinte) of capacity. Caixin put Chinese utilisation at about 40–50% in 2025 [SINGLE-SOURCE]. BloombergNEF noted that weak demand limits price increases (via The Edge Malaysia, 2 November 2026). No verified source gives a segment-by-segment global comparison of capacity and demand for wafers, cells and modules, so this report does not construct one.


4.2 Price transmission

  • FINDING: Chinese polysilicon traded near RMB 43/kg at 30 September 2026, Daqo reported n-type prices recovering to RMB 40–43/kg from RMB 30.5–33.0/kg by the end of Q3 [TWO-SOURCE], and global modules sold at $0.11–0.12/W against a 2025 record low of $0.09/W [SINGLE-SOURCE].

  • SO WHAT: Developers still buy equipment near record lows, so the gains from the slowdown reach buyers while the damage falls on sellers.

  • NOW WHAT: Test project economics at the 2025 low and at the proposed price floors, not only at today's spot price.

 

Chinese module prices rebounded to about RMB 0.9/W from a RMB 0.6/W low but stayed below a cost of roughly RMB 1/W, according to Caixin in March 2026 [SINGLE-SOURCE]. Producers discussed floors of RMB 0.79/W (TOPCon) and RMB 0.86/W (back-contact) on 11 October 2026 (PV Tech) [SINGLE-SOURCE]. US prices sit higher: SEIA and Wood Mackenzie report $0.37/W for distributed modules (-16% year on year) and $0.33/W for utility modules in the second quarter [SINGLE-SOURCE]. The Section 232 minimum import price for modules, $0.38/W from 4 December 2026, sits above both US figures [SINGLE-SOURCE].


4.3 Standalone solar versus solar-plus-storage

  • FINDING: Lazard's July 2026 US unsubsidised ranges are $40–98/MWh for utility solar, $61–156/MWh for solar plus storage, $37–99/MWh for onshore wind and $51–129/MWh for new gas combined cycle [SINGLE-SOURCE].

  • SO WHAT: The ranges overlap, so site and financing assumptions decide the ranking, and the figures apply to the United States only.

  • NOW WHAT: Treat the ranges as US inputs to check against local capex and cost of capital, never as a global levelised cost.

 

Lazard's base case assumes 60% debt at 8% and 40% equity at 12%, a 7.7% after-tax cost of capital.

"Renewables remain the lowest-cost and quickest to deploy resource,"

said Samuel Scroggins, Managing Director at Lazard, on 13 July 2026. New gas faces what Lazard calls a 15-year high in levelised cost on turbine constraints. Gas prices also rose after the Middle East conflict began on 1 March 2026, which SPE and PV Tech say lifted first-half solar demand; our analysis of LNG as a transition fuel to 2030 covers the gas side of that comparison.


Two cost signals point in different directions. ESS News reported Lazard's levelised cost of storage up about 27% since 2020 on a $/MWh basis, for a 100 MW, four-hour system, while Ember reports battery costs falling 20% in 2024 and 45% in 2025 [SINGLE-SOURCE]. The first measures the full cost of delivered storage and the second measures battery costs, so the two are not directly comparable. India's solar-plus-storage tariffs also conflict (Section 3.5). The evidence contains no payback period, operating cost or non-US levelised cost, so no breakeven price between standalone and hybrid projects can be derived.



4.4 Merchant revenue erosion

  • FINDING: In April 2026, France recorded 139 negative-price hours and Germany 123, up from 90 and 75 a year earlier, while LevelTen's second-quarter PPA benchmarks were $61.40/MWh in North America (-4.8% on the quarter) and EUR 56.59/MWh in Europe (+2.8%) (Pexapark, 12 May 2026; LevelTen via PV Tech, 23 July 2026) [SINGLE-SOURCE].

  • SO WHAT: Fixed-price contracts now carry the revenue that merchant sales cannot.

  • NOW WHAT: Require the contracted share and strike price before valuing any European solar asset.

 

Spain saw 138 negative-price hours in April (117 a year earlier), and Germany curtailed about 3.1% of its solar output in 2025, up from 1.9% (Ember). The box below shows how a falling capture factor changes revenue for a fictional project.

Worked example: what a falling capture factor does to project revenue (illustrative)


Sourced inputs. Pexapark April capture factors, 2025 to 2026: France 0.42 to 0.10; Germany 0.40 to 0.26 [SINGLE-SOURCE]. LevelTen Q2 2026 European solar PPA benchmark: EUR 56.59/MWh [SINGLE-SOURCE].


Illustrative inputs (not sourced). A 100 MWp plant producing 1,000 MWh per MWp a year, so 100,000 MWh annually; a baseload reference price of EUR 80/MWh; in the hedged case, 70% of output sold at the PPA benchmark and 30% sold at the market.


Calculation 1.

Captured price = capture factor × baseload price. France: 0.42 × 80 = EUR 33.60 falls to 0.10 × 80 = EUR 8.00. Germany: 0.40 × 80 = EUR 32.00 falls to 0.26 × 80 = EUR 20.80.


Calculation 2.

Merchant revenue = 100,000 MWh × captured price. France: EUR 3.36m falls to EUR 0.80m, a drop of EUR 2.56m (76%). Germany: EUR 3.20m falls to EUR 2.08m, a drop of EUR 1.12m (35%). This holds the April factors for a full year, which is a stress case, not a forecast.


Calculation 3.

Hedged Germany case, using the European average PPA benchmark. Blended price = (0.70 × 56.59) + (0.30 × 20.80) = 39.61 + 6.24 = EUR 45.85/MWh. Revenue = 100,000 × 45.85 = EUR 4.59m, against EUR 2.08m fully merchant: a difference of EUR 2.51m.


How to use it.

Replace the yield, reference price and contract share with your project's numbers. The percentage drop depends only on the two capture factors, so it holds at any baseload price.


Table 4.1: From polysilicon to project: price and volume markers

Stage

Metric

Date

Source

Evidence label

Polysilicon

Inventory 600,000 t; price RMB 40–43/kg

Aug 2026; end-Q3 2026

BloombergNEF via Edge Malaysia; Daqo via TaiyangNews

Inventory [SINGLE-SOURCE]; price [TWO-SOURCE]

Manufacturers

Six makers lost about CNY 25bn in 9M 2026

3 Nov 2026

Yicai Global

Press-reported results

Modules (global)

$0.11–0.12/W against $0.09/W record low

16 Oct 2026

TaiyangNews

[SINGLE-SOURCE]

Modules (US)

$0.33–0.37/W; Section 232 floor $0.38/W from 4 Dec 2026

Sep 2026; 10 Aug 2026

SEIA/WM via Mercom; White & Case

[SINGLE-SOURCE]

Project revenue

Capture factor 0.10 (France), 0.26 (Germany)

12 May 2026

Pexapark

[SINGLE-SOURCE]

Sources as listed. Evidence labels are defined in Section 1.2; URLs in References.


5. Named Company Case Studies

Direct answer: The most exposed companies are the Chinese module and wafer makers. Tongwei, LONGi, TCL Zhonghuan, JA Solar, Jinko and Trina lost CNY 6.8bn, 4.7bn, 4.6bn, 3.6bn, 3.4bn and 1.9bn in the first nine months of 2026 (Yicai Global, 3 November 2026). In the United States, First Solar cut its 2026 sales guidance to $4.6–4.9bn on 30 October 2026. Debt and liquidity data were not available, so this ranks exposure by reported losses and demand signals, not solvency.


5.1 First Solar (US)

  • FINDING: First Solar reported third-quarter 2026 net sales of $1.23bn (-23%) and a backlog of 44.0 GW, and cut its 2026 sales guidance to $4.6–4.9bn from $4.9–5.2bn (PV Tech, 30 October 2026) [SINGLE-SOURCE].

  • SO WHAT: Even the US supplier with the strongest policy protection faces slower orders, so protected demand does not guarantee volume.

  • NOW WHAT: Compare backlog against revised guidance, since the two moved in different directions.

 

The company's net income was $571m, helped in the quarter by 45X manufacturing-credit receipts and IEEPA tariff refunds. It guides 2026 module volume to 16–17 GW and US module capacity to about 17 GW by 2027. Reports tie the guidance cut to slower orders amid tariff uncertainty, a single detailed source [SINGLE-SOURCE].

Mark Widmar, First Solar's Chief Executive Officer, said on 30 October 2026 that the company is

"structurally positioned to benefit from this environment".

5.2 Sunrun and Enphase (US residential)

  • FINDING: Sunrun installed 204 MW of solar in the third quarter of 2026 (-15%) while its storage attachment rate reached 75% (70% a year earlier) and revenue rose 52% to nearly $1.1bn (PV Tech, 5 November 2026).

  • SO WHAT: Residential value is moving from panels to batteries, which is the revenue-quality argument at household scale.

  • NOW WHAT: Compare storage attachment rates when judging residential platforms.

 

Sunrun's direct installations rose 9%, and it lowered its cash-generation guidance to about $200m from $200–375m. Enphase reported revenue of $293.3m (-28.5% year on year), US sell-through down 35% year on year and fourth-quarter guidance of $225–255m, citing the end of the 25D credit and interest rates. US residential installs of 995 MW in the second quarter of 2026 were the lowest in five years, and Rhodium Group found residential storage outpacing residential solar for a second consecutive quarter.


5.3 LONGi, Tongwei, JA Solar and Daqo (China)

  • FINDING: Tongwei lost CNY 6.759bn on revenue of CNY 58.337bn (-9.69%), LONGi lost CNY 4.713bn on CNY 43.864bn (-13.85%) and JA Solar lost CNY 3.64bn on CNY 28.113bn (-23.62%) in the first nine months of 2026 (Yicai Global, 3 November 2026; pv magazine brief, 9 November 2026).

  • SO WHAT: Scale has not shielded the largest producers, so supplier concentration should not be read as supplier strength.

  • NOW WHAT: Compare each producer's loss with its revenue change, as Table 5.1 does, instead of ranking on size.

 

Tongwei's loss compares with about CNY 5.27bn in the comparison period the report cites. LONGi shipped 48.61 GW in the nine months, including 31.64 GW of back-contact modules, and 60.92% of JA Solar's shipments went overseas. TCL Zhonghuan, Jinko and Trina lost CNY 4.6bn, 3.4bn and 1.9bn; revenue and shipment figures for Jinko and Trina were not found. Chia Chen, an analyst at Bloomberg Intelligence, said on 2 November 2026 that LONGi's loss "suggests a long road to recovery".


Daqo is the polysilicon case. Its third-quarter output was 53,469 t against guidance of 40,000–45,000 t, sales were 46,910 t, the average selling price was $4.12/kg, utilisation was about 70%, and revenue was $194.2m compared with $244.6m. It raised full-year output guidance to 175,000–185,000 t but guided fourth-quarter output to 35,000–45,000 t. Output above guidance while industry inventory rose is the pattern Section 6 examines.



5.4 Waaree and Premier Energies (India)

  • FINDING: Waaree Energies reported an order book of INR 59,200 crore, second-quarter fiscal 2027 revenue of about INR 7,517.7 crore (+24%) with profit down about 14%, and plans to expand its US module plant to 4.8 GW (headline-level reports from Mercom and pv magazine India, 28 October 2026).

  • SO WHAT: Orders are growing while profit falls and US exposure persists, since Waaree's shares fell by up to 14.6% on 25 February 2026 on US duty news.

  • NOW WHAT: Verify the figures against BSE filings, because these reports come from aggregators and trade press.

 

Premier Energies reached module capacity of 11.1 GW and cell capacity of 10.6 GW after a 7 GW plant at Naidupeta began in September 2026, plans 10 GW of ingot-wafer capacity by the first quarter of 2028, and booked second-quarter fiscal 2027 orders of INR 40.01bn (US$419m, 2,308 MW) (PV Tech, 12 October 2026). Both firms are building integrated capacity during a domestic module surplus, and their results come from secondary sources only.


5.5 ACWA Power and Masdar (Gulf)

  • FINDING: Masdar and the Emirates Water and Electricity Company (EWEC) reached financing of US$6.1bn for a 5.2 GW round-the-clock solar project with 19 GWh of batteries, made up of US$5.1bn of debt from 13 banks and US$1bn of equity (Energy-Storage.News).

  • SO WHAT: Large firmed solar-plus-storage projects are being financed at scale, which is the hybrid model Section 7 tests.

  • NOW WHAT: Compare the contracted structure, not only the headline size, when assessing firmed projects.

 

Groundbreaking was in October 2025 and operations are expected in 2027. ACWA Power reported a 98.2 GW portfolio that includes 52.3 GW of renewables (not solar only), with 29 projects under construction totalling 46.2 GW and 3.5 GWh, and 11.6 GWh of battery capacity. Net income reports conflict, so we do not state it.


5.6 Manufacturer Stress Screen

  • FINDING: Six Chinese makers reported nine-month net losses between CNY 1.9bn and CNY 6.8bn, and only three of them published revenue changes we could verify (Yicai Global, 3 November 2026).

  • SO WHAT: The screen ranks relative stress on available metrics, so it shows who is weaker within the group, not who will fail.

  • NOW WHAT: Treat a "Low" rating as lower stress relative to peers, since every company in the group lost money.

 

Debt, liquidity and cash-cost data were not available from verified sources, and Canadian Solar, SolarEdge and Qcells are not covered. Table 5.1 shows only what the reported results support.


Table 5.1: Manufacturer Stress Screen

Company

9M 2026 net loss (stress)

Revenue change, year on year (stress)

Volume or utilisation as reported

Tongwei

CNY 6.8bn (High)

-9.69% (Low)

Polysilicon capacity about 900,000 t

LONGi

CNY 4.7bn (High)

-13.85% (Medium)

48.61 GW shipped, 31.64 GW back-contact

TCL Zhonghuan

CNY 4.6bn (Medium)

Not scored

Not found

JA Solar

CNY 3.6bn (Medium)

-23.62% (High)

60.92% of shipments overseas

Jinko

CNY 3.4bn (Low)

Not scored

Not found

Trina

CNY 1.9bn (Low)

Not scored

Not found

Daqo

Not found

Revenue $194.2m vs $244.6m

Utilisation about 70%

Methodology note: The screen is structured analyst judgement, not a financial model. It compares companies on the metrics available for all or most of them from reported results: nine-month net loss, year-on-year revenue change, shipments or output, share of revenue from overseas, and utilisation where the company reports it. Each metric is shown as reported; we then mark each company High, Medium or Low stress for that metric by its position within the group, not against an absolute threshold. No composite score is calculated. The screen does not include debt, liquidity, cash cost or margin data, because these were not available from verified sources, and it should not be read as a solvency or survival forecast. Companies without sufficient verified data are shown as "Not scored".

Sources: Yicai Global, 3 Nov 2026; pv magazine brief, 9 Nov 2026; TaiyangNews on Daqo, 29 Oct 2026. Press-reported results, not checked against company filings. No composite score is calculated.


6. Friction, Risk & Systemic Bottlenecks

Direct answer: Evidence so far shows output restraint and price floors in China's polysilicon industry but little capacity retired. Caixin counted under 10% of capacity cut by March 2026 against a one-third target, inventory reached 600,000 t in August 2026, and producers discussed floors of RMB 43,000/t for silicon on 11 October 2026. No source reports capacity retired after March 2026. Our reading, labelled as inference, is price support rather than capacity removal.


6.1 Critical unresolved issue: does anti-involution remove capacity, or only lift prices?

  • FINDING: Polysilicon prices rose from about RMB 34,400/t at the end of June 2025 to RMB 53,200/t at the end of September 2025 on anti-involution expectations, then fell back to about RMB 40,000–43,000/t by the end of Q3 2026, while inventory reached 600,000 t (PV Tech, 6 November 2025; PV Tech, 16 October 2026) [SINGLE-SOURCE].

  • SO WHAT: Whether this is closure or coordination decides if loss-making supply returns when demand stays weak.

  • NOW WHAT: Monitor three indicators: capacity retirements in CPIA or MIIT data, monthly polysilicon output and inventory, and whether the RMB 43,000/t floor holds.

 

This question matters because Chinese capacity is the main variable in the overcapacity thesis. The IEA put supply-chain concentration above 90% in key segments through 2030 in October 2025, and BloombergNEF expected overcapacity to persist through at least 2027 in April 2025. Table 6.1 sets out the timeline.


Table 6.1: Anti-involution evidence timeline

Date

Event

Source

Aug 2025

Producers announce a plan to cut output by one-third; Reuters reports a RMB 50bn (about $7bn) fund to buy and shut capacity, with funding unclear

PV Tech, 5 Aug 2025

Dec 2025

Nine producers and CPIA form a platform with RMB 3bn capital (Tongwei 30.35%), small against expected costs of tens of billions

Caixin, 5 Mar 2026

9 Jan 2026

State Administration for Market Regulation (SAMR) warns Tongwei, GCL, Daqo and Xinte against coordinating capacity, volumes and prices

PV Tech; OPIS, 12 Jan 2026

Mar 2026

Caixin counts under 10% of capacity cut; 2025 utilisation about 40–50%; inventory above 500,000 t

Caixin, 5 Mar 2026; OPIS, 24 Mar 2026

Aug 2026

Record inventory of 600,000 t, over six months of global wafer output

BloombergNEF via Edge Malaysia, 2 Nov 2026

End-Q3 2026

Daqo output 53,469 t, above guidance; price RMB 40–43/kg

TaiyangNews, 29 Oct 2026

11 Oct 2026

Producers discuss price floors (RMB 43,000/t silicon; RMB 0.79/W and RMB 0.86/W modules); October output projected down 12.9% month on month

PV Tech, 16 Oct 2026

Sources as listed; each row is [SINGLE-SOURCE]. SAMR action is described as a halt (pv magazine headline), an urging to suspend (OPIS) and a warning (PV Tech); we use "warned".

Reading A: prices lifted, capacity largely intact. The one-third cut target met a platform capitalised at RMB 3bn and a regulator warning. Caixin counted under 10% cut by March, inventory rose, and Daqo's third-quarter output beat its own guidance. The October response is a price floor, which keeps plants running. Prices have since fallen back from RMB 53.2/kg to RMB 40–43/kg.


Reading B: output discipline is real even if capacity is not retired. Output fell 26.4% (polysilicon) and 12.3% (wafer) in 2025, January and February 2026 output fell month on month, and one leading producer halted plants in three provinces. Daqo guided fourth-quarter output down. Output cuts are not capacity removal, and none of the sources reports capacity retired or acquired after March 2026.

Our reading. The evidence fits a price-lift story better than a capacity-removal story: capacity cut under 10% against a one-third target, a regulator warning, price floors and rising inventory all point the same way [GFJ INFERENCE]. This is Green Fuel Journal's inference, not a sourced finding.


What cannot be answered. No source gives a post-March 2026 figure for capacity retired. No named analyst has tested whether the floors hold through 2027, when BloombergNEF expects 2026 global installs down 10.8% and China's first nine months of 2026 were 54% lower. No source examines how price floors sit with SAMR's January guidance against price coordination. The Pricing Law amendment outcome is unknown.


6.2 Other friction by region

  • FINDING: Brazil's grid operator curtailed 17.8% of potential wind and solar output in 2026 to 10 May (wind 16.5%, solar 20.6%), after 20.7% in 2025, and ANEEL withdrew 509 authorisations (22 GW) in 2025 (ONS and ANEEL via S&P Global, 11 May 2026).

  • SO WHAT: Curtailment without compensation turns a generation asset into a revenue risk that tariffs cannot fix.

  • NOW WHAT: Require curtailment and compensation terms in any Brazilian offtake or grid agreement.

 

Rodrigo Sauaia, executive president of ABSOLAR, said on 11 May 2026 that the main problem is not curtailment itself but "the absence of any refund to generators". Elsewhere the constraints differ, but each cuts returns in a similar way. In the US, Wood Mackenzie says permitting delays touch about 30% of the early-stage utility pipeline, and Rhodium Group recorded about $17bn of energy and industry project cancellations in the second quarter of 2026, with solar about $3bn [SINGLE-SOURCE]. In the EU, the high-risk inverter ban could add 2–8% to project costs and shift 4–5 GW a year away from Chinese vendors, while about 80% of demand flows through channels the ban does not cover [SINGLE-SOURCE].

In India, unsigned PSAs and curtailment weigh on returns (Section 3.4). In ASEAN, Ember estimates a need of about $281bn a year by 2035, 23–26 GW of storage by 2030 against about 7 GW announced, and finance that is thin. In Africa, Ember's installs are import-derived estimates that official statistics cannot yet see, and the World Bank's claim of more than 50 million connections under Mission 300, announced on 16 June 2026, is disputed by ODI.


Table 6.2: Risk register by region

Region

Risk

Evidence

Evidence note

Brazil

Curtailment without compensation

17.8% of potential wind and solar output curtailed, 2026 to 10 May

Secondary (ONS via S&P Global)

United States

Permitting and cancellations

About 30% of early-stage utility pipeline delayed; about $3bn solar cancellations in Q2 2026

[SINGLE-SOURCE]

European Union

Merchant erosion; inverter costs

Capture factors 0.10 (France), 0.26 (Germany); 2–8% cost increase

[SINGLE-SOURCE]

India

Offtake and grid

43.9 GW PSAs pending (Sep 2025); about 470 GWh curtailed in Q1 2026

Latest PSA figure older than 12 months

ASEAN

Finance and grid

18% of solar and wind projects with MDB or concessional finance

[SINGLE-SOURCE]

Africa

Statistics and contested claims

17 GW import-derived estimate; Mission 300 claim disputed

[SINGLE-SOURCE]; contested

Methodology note: The register lists the main friction in each region with the evidence behind it. It assigns no score and no ranking. The last column shows either the evidence label or the limit on the data, and regions are ordered by the sequence of the section, not by severity.

Sources: S&P Global; Wood Mackenzie; Rhodium Group; Pexapark; Ember; ODI. URLs in References.


6.3 Source conflicts investors should know

  • FINDING: Sources describe China's January 2026 SAMR action as a halt (pv magazine headline), an urging to suspend (OPIS) and a warning (PV Tech).

  • SO WHAT: The legal weight of the action changes how much enforcement risk investors should attach to producer coordination and price floors.

  • NOW WHAT: Treat the action as a warning, the description PV Tech supports, until a primary text is read.

 

Table 2.2 lists the five conflicts with the largest effect on forecasts. Others to keep in view: ACWA Power's income reports and India's total renewable capacity (299.2 GW against 308.02 GW and 310 GW). The conflicts are shown, not resolved.


7. Capital & Investment Implications

Direct answer: Outside China, India is the best-evidenced market for solar capital in 2027, with 171.05 GW installed at 30 September 2026, though 43.9 GW of awards still lacked signed power sale agreements at 30 September 2025. The Philippines and Nigeria can be scored only on thin evidence, and Malaysia could not be assessed because no verified data were found. Markets are compared on demand evidence, offtake and finance, grid, policy and trade exposure, and data confidence.


7.1 Standalone versus hybrid assets

  • FINDING: Sunrun reported a 75% storage attachment rate in the third quarter of 2026, Australia's Clean Energy Regulator reported battery attachment above 80% of small-scale installs in April 2026, and EU large-scale batteries passed 10 GW from 4 GW in 2023 (PV Tech, 5 November 2026; ESS News, 31 August 2026; Ember).

  • SO WHAT: Where solar revenue is eroding, storage is the main lever that installers and buyers are already using.

  • NOW WHAT: Screen any solar asset for revenue quality: contracted share, storage configuration and exposure to negative-price hours.

 

This is the revenue quality argument. A standalone project earns the price at the hours it generates, which is falling where solar is abundant. A hybrid project can move output to later hours, at a higher cost: Lazard's US range for solar plus storage is $61–156/MWh against $40–98/MWh for solar alone [SINGLE-SOURCE]. No verified source gives returns, payback or operating cost for hybrid against standalone assets, so this report does not rank them. Investors who compare solar with other long-dated renewable assets may find our coverage of offshore wind investment intelligence and next-generation geothermal investment useful for comparison.


7.2 Four-country market-risk matrix

  • FINDING: India has the deepest evidence base of the four markets, with 171.05 GW installed, 35.2 GW added in January–September 2026 and tariffs of INR 2.42–2.54/kWh, while Malaysia has none that we could verify.

  • SO WHAT: Best-evidenced is not risk-free, because India's offtake and grid risks sit in the same evidence as its demand.

  • NOW WHAT: Ask each counterparty for PSA status, grid-connection date and curtailment terms before comparing markets.

 

Table 7.1 rates five criteria as High, Medium or Low in opportunity-to-risk terms, relative to the other markets. "Not scored" means no verified evidence was found.


Table 7.1: Four-country market-risk matrix

Criterion

India

Philippines

Nigeria

Malaysia

Demand evidence

High: 171.05 GW at 30 Sep 2026; 35.2 GW in 9M 2026

Medium: GEA-4 subscribed 9,423.622 MW of 10,653 MW; solar-only figure not given

Medium: 1.7 GW import-derived estimate for 2026 [SINGLE-SOURCE]

Not scored

Offtake and financing

Medium: tariffs INR 2.42–2.54/kWh; 43.9 GW PSAs pending (Sep 2025)

Low: single-buyer procurement; modelled tariff $76/MWh; thin MDB finance (ASEAN 18%)

Not scored: no tariff, offtaker or currency data

Not scored

Grid and curtailment

Low: about 470 GWh curtailed in Q1 2026

Low: grid named as a constraint for ASEAN generally

Not scored: no grid data

Not scored

Policy and trade exposure

Medium: ALMM protects domestic demand; 249.13% US duty hits exporters

Medium: auction delivery 2026–2029; no trade evidence

Not scored: no policy evidence

Not scored

Data confidence

Medium: several sources, but conflicts and secondary reporting

Low: auction and model data only

Low: one estimate; ODI disputes related claims

Not scored

Methodology note: The matrix is structured analyst judgement based on the evidence gathered for this report. Five criteria are rated High, Medium or Low opportunity-to-risk: demand evidence, offtaker and financing conditions, grid and curtailment, policy and trade exposure, and data confidence. Each rating is shown with its supporting figure and source. Ratings are relative to the other markets in the matrix and are not forecasts of returns. A market is marked "Not scored" where we found no verified evidence; this applies to Malaysia unless supplementary evidence is added. The matrix does not account for currency risk or sovereign credit, which were not covered by the sources.

Sources: MNRE via SolarQuarter; JMK Research; Ember (7 Oct 2026, 2 Sep 2026); REGlobal on GEA-4; ODI (June 2026). URLs in References.


India. The questions to put to a developer are the PSA status of each awarded megawatt, the connection date, and the share of output exposed to transmission curtailment.

Philippines. The GEA-4 auction delivers from 2026 to 2029, and Ember's model of a 20 MWp project with a 12% equity return and 7% debt needs about $76/MWh, a model output rather than an auction result [SINGLE-SOURCE]. Ask who the offtaker is and which lenders are in the structure.

Nigeria. Ember's estimate is among four African markets (South Africa 3.3 GW, Egypt 2.0 GW, Nigeria 1.7 GW, Algeria 1.4 GW) and rests on import data. The ODI reports that the Nigeria DARES programme had 1.046 million solar home systems and 67 isolated mini-grids by June 2026 and no interconnected mini-grids. Ask for tariff, offtaker and currency terms, none of which the sources provide. Malaysia. We found no verified data on installs, grid access, power purchase agreements or foreign-investment limits, so it is not scored.


7.3 Reference markets

  • FINDING: Pakistan had about 38 GW of distributed solar by June 2025, supplying 28% of fiscal 2025 generation, up from 10% in fiscal 2023 (Ember, 25 June 2026) [SINGLE-SOURCE].

  • SO WHAT: Demand can scale on household economics without utility tenders, and the sources we could verify do not report grid performance.

  • NOW WHAT: Check net-billing rules and grid data before extrapolating Pakistan's pattern elsewhere.

 

Brazil is the reference case for curtailment risk. ABSOLAR projects 11.4 GW installed in 2025 (-24%) and 10.6 GW in 2026, and both are forecasts, not actuals [SINGLE-SOURCE]. Cumulative capacity was 60 GW in August 2025, of which 42.1 GW was distributed. Table 7.2 sets the two markets side by side; it carries no ratings, so no methodology note applies, but the Table 7.1 note governs any judgement drawn from it.


Table 7.2: Reference markets

Market

Demand signal

Constraint

Evidence label

Pakistan

About 38 GW distributed by Jun 2025; 28% of FY25 generation; 47 GW of panels exported from China by Jun 2025

Data vintage June 2025; grid-strain reports not verified

[SINGLE-SOURCE]

Brazil

11.4 GW (2025, -24%); 10.6 GW (2026 forecast); 60 GW cumulative (Aug 2025)

Curtailment 17.8% in 2026 to 10 May; no refund to generators

[SINGLE-SOURCE]; forecasts

Methodology note: This table carries no ratings. Any judgement drawn from it follows Methodology Note 3, shown directly under Table 7.1.

Sources: Ember with Renewables First, 25 Jun 2026; Canal Solar, 10 Dec 2025; S&P Global, 11 May 2026. URLs in References.


7.4 Manufacturing exposure

  • FINDING: Capital in manufacturing sits at risk in two places, Chinese segments reporting about CNY 25bn of nine-month losses and an Indian module base running at 35–40% utilisation, and Rhodium Group recorded about $1.1bn of US solar manufacturing cancellations in the second quarter of 2026 [SINGLE-SOURCE].

  • SO WHAT: Optionality in manufacturing depends on whether price floors hold and whether consolidation follows, and neither is yet visible.

  • NOW WHAT: Watch capacity retirements and floor durability before treating any manufacturing stake as a recovery position.

 

Rhodium also recorded nearly $1bn of solar manufacturing investment (-19% against the second quarter of 2025) and more than $10bn of newly announced manufacturing, about $8bn of it solar. US module capacity reached 65.5 GW at the end of 2025, but production was about 70% of 2025 installs and domestic cell capacity was 3 GW. We name no acquisition targets, because the sources give no debt, liquidity or cash-cost data for any company.


8. Future Scenarios & Forecast (2026–2030, with 2031–2035 Direction)

Direct answer: Three scenarios frame 2027: prolonged oversupply, gradual consolidation and faster recovery. SolarPower Europe's 2026 cases range from a low of -25% through 612 GW in the medium case to 724 GW in the high case, and CPIA's 2030 range for China is 270–320 GW (AC). Triggers include whether the RMB 43,000/t silicon floor holds. No scenario carries a numeric probability, because no source supports one.


8.1 Scenarios 2026–2030

  • FINDING: SolarPower Europe's 2026 cases run from a low of -25% to 612 GW (medium) and 724 GW (high), and CPIA's China outlook is 265–300 GW for 2028 and 270–320 GW for 2030, both in AC (SPE, 22 June 2026; PV Tech, 11 February 2026).

  • SO WHAT: The ranges are wide enough that planning on one number would hide most of the risk.

  • NOW WHAT: Test each scenario against its trigger indicators rather than a probability.

 

Bear: prolonged oversupply. Floors fail or inventories keep building, China's additions stay near the low end of the CPIA range, and manufacturers' losses persist. Base: gradual consolidation. Output restraint continues, some capacity exits, global installs stabilise near the 2026 level, and growth resumes slowly. The BloombergNEF conflict sits here: one report gives 618 GW for 2027, another calls 2027 weaker than previously estimated, so the Base case uses neither number. Bull: faster recovery. Demand follows SPE's high case, China returns toward the upper CPIA range, and storage and grid investment absorb more solar. SPE expects growth to resume in 2027, which supports the Base and Bull cases, while the Bloomberg Intelligence view of a long road to recovery supports the Bear.


Table 8.1: Scenarios, 2026–2030

Scenario

Assumptions

Trigger indicators

Investment implication

Evidence anchor

Bear: prolonged oversupply

Price floor breaks; inventory above 600,000 t persists; China near low end of CPIA range

Floor fails; output rises while demand is flat

Manufacturer counterparty risk stays high

Caixin, Mar 2026; BloombergNEF via Edge Malaysia, 2 Nov 2026

Base: gradual consolidation

Output restraint holds; some capacity retired; global installs near the 2026 level

Retirements appear in CPIA or MIIT data; inventory falls

Equipment stays cheap; selectivity on revenue quality

SPE medium case 612 GW; CPIA 180–240 GW (2026)

Bull: faster recovery

Demand follows SPE high case; China toward upper CPIA range

Monthly NEA additions beat 2026 run-rate; PSAs signed in India

Volume-linked assets benefit; margin recovery depends on floors

SPE high case 724 GW; CPIA 270–320 GW (2030)

Methodology note: The three scenarios are analyst constructions anchored on published forecast ranges (SolarPower Europe's low, medium and high cases for 2026, and CPIA's 2030 range). They carry no probability weights, because no source provides a basis for assigning them. Each scenario states its assumptions and the observable indicators that would move the outlook toward it. The 2031–2035 paragraph is analyst judgement only, with no supporting forecast data.

Sources: SolarPower Europe, 22 Jun 2026; CPIA via PV Tech, 11 Feb 2026; Caixin, 5 Mar 2026. No probabilities are assigned.


8.2 Trigger indicators

  • FINDING: Dated checkpoints fall close together: Section 232 tariffs and price floors take effect on 4 December 2026, the next SEIA/Wood Mackenzie report is expected in December 2026, and China's battery export VAT rebate falls to zero and a European Union clawback requirement begins on 1 January 2027 (White & Case; SEIA; TaiyangNews; Clean Energy Wire).

  • SO WHAT: The next two quarters will show which scenario the data favour, so the checkpoints matter more than long-range numbers.

  • NOW WHAT: Build a monitoring list around the six indicators below and review it at each checkpoint.

 

The six indicators are polysilicon output and inventory, whether the RMB 43,000/t floor holds, China's monthly additions, the effect of the US Section 232 measures, India's PSA signings and storage attachment rates. India's PM Surya Ghar target and PM-KUSUM timeline both run to March 2027.


8.3 2031–2035 direction

  • FINDING: The sources contain no verified forecast for 2031–2035, and the longest data are CPIA's China range to 2030 (270–320 GW) and SEIA/Wood Mackenzie's US outlook of about 44 GWdc a year through 2031.

  • SO WHAT: Long-dated models therefore rest on judgement, not on sourced forecasts.

  • NOW WHAT: Label any input beyond 2030 as an assumption and test it against the Bear case.

 

Our judgement, with no supporting forecast data [GFJ INFERENCE]: if oversupply clears, the constraint on solar growth shifts from module prices to grid capacity, storage and revenue design. Other renewable asset classes compete for the same capital, and our offshore wind investment intelligence, part II offers a comparison point.


9. Strategic Recommendations

Direct answer: Revenue quality, counterparty strength and offtake readiness should carry more weight than any single forecast. April 2026 capture factors were 0.10 in France and 0.26 in Germany. Six Chinese makers lost about CNY 25 billion in the first nine months of 2026. In India, 43.9 GW of awards lacked signed agreements at 30 September 2025. This is analysis for energy investors, not advice to buy, sell or hold.


9.1 For industry

  • FINDING: Daqo guided fourth-quarter 2026 output to 35,000–45,000 t, below its third-quarter 53,469 t, and said lower utilisation was under consideration (TaiyangNews, 29 October 2026).

  • SO WHAT: Cash and utilisation discipline now matter more to manufacturers than volume.

  • NOW WHAT: Compare utilisation and cash position, not nameplate capacity, when benchmarking producers (Section 5).

 

Manufacturers face a choice between output discipline and market share, and Section 6 shows why price floors alone leave inventory in place. Vertical integration is one response: Wood Mackenzie's Caitlin Connelly noted that the new US tariff structure widens the advantage for fully integrated domestic suppliers. Developers can favour hybrid configurations and contracted offtake, as the FDRE and round-the-clock tenders in Section 3.5 show (Section 4.4).


9.2 For investors

  • FINDING: Revenue quality, not irradiance, sets what a solar asset earns in markets with high solar penetration, as the 0.10 French capture factor shows (Pexapark, 12 May 2026) [SINGLE-SOURCE].

  • SO WHAT: The contracted share of output and the counterparty behind it decide value more than yield does.

  • NOW WHAT: Weigh a revenue-quality screen first, then offtake status, then curtailment terms (Sections 4.4, 3.4 and 6.2).

 

Three factors follow from the evidence.

First, test revenue quality using the worked example in Section 4.4.

Second, treat awards without signed PSAs as unpriced risk until the agreement is signed (Section 3.4).

Third, verify counterparty strength against the Manufacturer Stress Screen and confirm curtailment terms before committing capital (Sections 5.6 and 6.2).

Treat Malaysia and Nigeria as pending diligence, because the sources do not support a score (Section 7.2). These are factors to weigh, not instructions to buy or sell.


9.3 For policymakers

  • FINDING: Francisco Silva, director of regulatory matters at ABEEólica, said on 11 May 2026 that companies would stop investing in Brazil until curtailment is resolved (S&P Global).

  • SO WHAT: Unresolved curtailment compensation can halt new investment even when demand exists.

  • NOW WHAT: Weigh compensation rules for curtailed generators alongside tariff design.

 

Four evidence-backed priorities emerge. Brazil could set clear compensation for curtailed output. India could speed signing of the PSAs behind its 43.9 GW of pending awards. China's regulators could publish capacity-removal data so that price-floor claims can be tested. African statistics offices could close the gap between import-derived estimates and official figures.


9.4 The first diligence step for energy investors

Before committing capital to any solar asset this quarter, an energy investor can test it with three requests: the contracted-versus-merchant revenue split with strike prices, the signed PSA or grid-connection date, and the counterparty's latest reported loss and utilisation. Then rerun the Section 4.4 calculation at the 0.10 and 0.26 capture factors and revisit the result after the Section 232 measures take effect on 4 December 2026. This is a diligence sequence, not a recommendation to buy, sell or hold.


Table 9.1: Recommendations at a glance

Recommendation

Reader

Evidence anchor

Section

Prefer cash and utilisation discipline over volume

Industry

Daqo Q4 guidance 35,000–45,000 t; losses of about CNY 25bn

5.3, 9.1

Favour hybrid configurations and contracted offtake

Industry

Capture factors 0.10 and 0.26; FDRE tariffs INR 5.99–6.00/kWh

3.5, 4.4

Apply a revenue-quality screen first

Investors

Worked example: 76% and 35% revenue falls at April factors

4.4, 9.2

Treat unsigned PSAs as unpriced risk

Investors

43.9 GW pending at 30 Sep 2025

3.4

Verify counterparty and curtailment terms

Investors

Table 5.1; Brazil 17.8% curtailment

5.6, 6.2

Publish capacity-removal and curtailment-compensation data

Policymakers

Under 10% of capacity cut by Mar 2026; Brazil refund gap

6.1, 6.2

Recommendations are factors to weigh, not instructions to buy, sell or hold.


10. Executive FAQ

Direct answer: China explains the 2026 fall, with 110.83 GW (AC) added in nine months, 54% lower, and Chinese manufacturers carry the losses of about CNY 25 billion. India leads outside China on evidence, with 171.05 GW installed. In the US, Lazard's range for solar plus storage is $61–156/MWh against $40–98/MWh for solar alone, and no breakeven analysis exists. The 2027 outlook turns on three scenarios and on whether the RMB 43,000/t floor holds.


Why is the global solar market shrinking in 2026, and which countries are behind it?

China drives the fall: its additions dropped 54% to 110.83 GW (AC) in January–September 2026 after a record 315.07 GW in 2025, and SolarPower Europe and BloombergNEF forecast global declines of 8% and 10.8%. The United States and Europe add smaller shifts: US residential installs reached a five-year low of 995 MW in the second quarter, and the EU's rooftop share of new capacity fell to 14% in 2025.


Which solar manufacturers are under the most financial pressure from overcapacity?

Tongwei and LONGi carry the largest reported losses. Tongwei lost CNY 6.759bn and LONGi CNY 4.713bn in the first nine months of 2026, and TCL Zhonghuan, JA Solar, Jinko and Trina also lost money. Debt and liquidity data were not available, so the ranking reflects losses rather than solvency.


Where can investors look for solar opportunities outside China in 2027?

India is the best-evidenced market outside China, followed by Africa and Pakistan on demand. India has 171.05 GW installed and is the one top market Wood Mackenzie expects to grow in 2026, while Ember estimates African installs at 17 GW. The limits are unsigned PSAs in India, finance in ASEAN and curtailment in Brazil. Malaysia could not be assessed.


How is China's solar overcapacity affecting module prices and company profits?

Overcapacity is pressuring both prices and profits. Polysilicon inventory reached 600,000 t in August 2026, six Chinese makers lost about CNY 25 billion in nine months, and global modules sold at $0.11–0.12/W against a $0.09/W low. Producers are discussing a silicon floor of RMB 43,000/t, yet Caixin counted under 10% of capacity cut by March 2026.


Is solar plus storage a better investment than standalone solar right now?

The evidence does not show that it is. Lazard's US ranges put solar plus storage at $61–156/MWh and solar alone at $40–98/MWh, but they cover only the United States. Capture factors of 0.10 in France and Sunrun's 75% attachment rate favour storage on revenue quality. No payback or breakeven analysis exists in the evidence.


What could 2027 look like for the solar industry, and what would change the outlook?

The 2027 outlook is uncertain and splits into three scenarios. SPE expects growth to resume, BloombergNEF's 2027 view conflicts between 618 GW and "weaker than previously estimated", and CPIA expects recovery without giving a figure. Watch the RMB 43,000/t floor, China's monthly additions and the US measures effective 4 December 2026.


Research Limitations

This report draws on research compiled on 10 November 2026. Most figures reached us through trade press such as PV Tech, TaiyangNews, SolarQuarter, Mercom, Yicai Global, Caixin and S&P Global, reporting the work of SEIA, Wood Mackenzie, SolarPower Europe, BloombergNEF, IEA, IRENA, Ember, CPIA, NEA and MNRE. We could not open the primary documents of most of those bodies, and some pages were blocked.


Source pages were read in summary form, so quotations and some figures may differ from the original wording and should be checked against the live page. Several figures conflict between sources and are shown unreconciled. No cost comparison outside the United States, apart from India's auction tariffs, meets our two-source standard, and we obtained no current Chinese module-cost, global levelised-cost or capital-cost benchmark. Coverage of Southeast Asia, Latin America outside Brazil, Africa and the Middle East is partial.


We found no debt, liquidity or cash-cost data for any company, no outcome for the US International Trade Commission's vote on Indian cells, no current Indian distribution-company dues, no verified data for Malaysia, and no verified forecast for 2031–2035. No figure in this report is estimated.


References & Strategic Sources

Format: Source Name │ Document Title │ Date │ URL. Grouped by region. Many figures reached us through trade-press reports of the underlying institutions; see Research Limitations.


Global


United States and Canada


European Union and United Kingdom


China


India


Emerging markets


This report is backed by authoritative research, institutional analysis, industry intelligence, and strategic data sources.


© 2026 Green Fuel Journal (GreenFuelJournal.com). All rights reserved. Reproduction or redistribution of this report, in whole or in part, requires prior written permission. Short excerpts may be quoted with a clear credit and a link to the original article. Third-party data remains the property of its original sources, as cited in the References.

bottom of page