1. How the European Electricity Market Works

1.1 The merit order and marginal pricing

European wholesale electricity prices are determined by the merit order: generators are dispatched in ascending order of short-run marginal cost, and the last unit called to meet demand sets the clearing price for every megawatt-hour sold in that hour, regardless of each generator's own production cost. A nuclear plant producing at €15/MWh receives the same clearing price as the gas-fired turbine that set it at €90/MWh. The logic is efficiency, at any moment, the system dispatches the cheapest available generation first, but the consequence is that a single high-cost marginal unit determines the revenue of every unit in the system.

In practice, the merit order in continental Europe runs from nuclear and run-of-river hydro at the bottom, through coal and lignite in the middle, to gas combined-cycle turbines at the top of the dispatchable stack. Renewables, wind and solar, bid at or near zero, because their short-run marginal cost is effectively nothing: no fuel, minimal variable operating cost. They are therefore dispatched before everything else whenever they are producing. Gas CCGTs, by contrast, carry a variable cost of approximately €85–100/MWh at current TTF and EUA levels, which means they only clear the market when demand is high enough, or renewable output low enough, that the system needs them. In those hours, the gas plant sets the price, and that price flows to every other generator in the system.

The practical consequence is a tight historical correlation between European gas prices and European electricity prices. When TTF averaged €131/MWh in 2022, European day-ahead electricity prices followed, averaging €276/MWh in France and comparable levels across the continent. When TTF fell back toward €40/MWh in 2023, electricity prices followed to €97/MWh. The gas plant is not just a generator, it is the price-setter for the majority of European electricity hours, and the economics of the entire system are therefore anchored to the cost of gas.

1.2 Market structure and the day-ahead auction

The price-setting mechanism operates through a day-ahead auction run by power exchanges, primarily EPEX SPOT for continental Europe, using a common algorithm called EUPHEMIA that clears prices simultaneously across interconnected bidding zones. At gate closure each afternoon, generators submit supply offers and retailers submit demand bids for each of the 24 hours of the following day. EUPHEMIA matches them at a clearing price that maximises social welfare across the coupled zones, subject to cross-border interconnection capacity constraints.

The result is a single clearing price per hour per zone, but prices can diverge sharply between zones when interconnection capacity is congested. The Iberian Peninsula historically cleared at prices significantly above or below the rest of continental Europe during periods of high renewable output or interconnector constraint, a live illustration of how physical infrastructure limits the theoretical uniformity of the single market. Interconnectors are therefore not just transmission assets; they are price arbitrage mechanisms, and their capacity determines how tightly correlated European electricity markets actually are in practice.

Intraday and balancing markets sit on top of the day-ahead structure, allowing generators and consumers to adjust positions as renewable output forecasts update throughout the day. Balancing costs, the cost of correcting mismatches between scheduled and actual generation, are socialised across market participants through TSO charges. As renewable penetration rises and forecast errors grow, these costs are increasing, though they remain relatively small compared to the day-ahead clearing price.

1.3 The structural tension this paper analyses

The merit order works as an investment coordination mechanism because of one specific condition: the marginal generator responds to price. When electricity is expensive, dispatchable plants run more. When it is cheap, they run less or shut down. The price signal is genuine, it reflects real scarcity, and investors, developers, and lenders can build 20-year financing decisions around it.^[3]

Renewable generation breaks this condition at scale. Wind turbines and solar panels produce when meteorological conditions allow, not when the market asks them to. A wind farm does not run more because electricity is expensive at 7pm on a winter evening, and it does not run less because prices are collapsing at noon on a sunny Sunday. Its output is determined by the weather, not the price signal. As renewable capacity grows to dominate supply during certain hours, the clearing price during those hours stops reflecting genuine scarcity and starts reflecting weather patterns. That is a fundamentally different kind of market, and the pricing mechanism was not designed for it.

The practical consequence is a price distribution that is becoming increasingly bimodal. During high-renewable hours, sunny afternoons, windy nights, supply floods the market and prices collapse toward zero or below regardless of whether the system actually needs more energy. During tight periods, cold, still winter evenings when demand peaks and renewables are not producing, residual demand falls entirely on dispatchable generators and prices spike sharply. Germany already sees this in the data: 457 negative-price hours in 2023 alongside periodic winter spikes above €200/MWh in the same year. The annual average of roughly €95/MWh in 2023 conceals a distribution that looks nothing like a stable market.

This creates two distinct and compounding problems, which are the subject of Sections 2 and 3 respectively.

For gas generators, the hours they are needed are becoming increasingly concentrated in extreme events. A gas plant that ran 3,000 hours a year at 35% renewable penetration might run 1,200 hours at 65% penetration, only during the coldest, stillest weeks when no other option exists. Its fixed costs are unchanged. Its fuel and carbon costs per MWh are unchanged. But those costs must now be recovered across 60% fewer running hours. Even a high spark spread during those hours may be insufficient to cover annual fixed costs. The plant is not broken, it is structurally uneconomic, a stranded asset produced not by technology failure but by the market structure around it.

For renewable developers, the problem runs in the opposite direction but through the same mechanism. They produce disproportionately during the hours when the price is lowest, because they produce when weather conditions allow, and weather conditions allow during the same hours for all generators of the same type simultaneously. A wind farm built in Germany in 2010 at 5% renewable penetration achieved roughly 92% of the average market price per MWh. At 35% penetration today, comparable assets achieve around 80%. At 65% penetration by 2030, the projection is 60–65%, before accounting for the indirect effect of lower gas demand on TTF, which compresses prices further during the remaining gas-dominated hours. The asset is physically identical. The market structure around it has changed.

There is a third tension that underlies both of these, and it is the most fundamental. The merit order was designed for a world where short-run marginal cost, the cost of producing one more MWh from a plant that already exists, and long-run marginal cost, the cost of building and operating a new plant, converge over time through the investment signal. When spot prices are consistently above the long-run cost of new capacity, developers build more until prices fall. When they are below, no new capacity is built until the gap closes. This self-correcting mechanism is what makes the merit order work as a market.

Renewables have a near-zero short-run marginal cost but a long-run marginal cost of €70–90/MWh for offshore wind once capex and financing are included. The spot market price during the hours they produce will structurally sit far below their long-run cost as penetration rises, the market is telling them they are worth very little at the precise moment they are producing. The market signal alone cannot finance new renewable capacity at scale. The gap between what the market pays and what the project needs to be bankable is exactly the gap that contracts for difference, power purchase agreements, and capacity contracts exist to fill, and it will widen as penetration rises, not narrow.

The 2023 EU electricity market reform attempted to address some of the consequences of this tension, particularly the distribution of windfall profits during gas price spikes. The pricing mechanism itself was left unchanged. Gas still dominates the marginal price in the majority of hours. The merit order is intact. And the structural forces pulling both gas and renewable economics toward dependence on administered prices are accelerating.^[4]

To understand where this goes, the first question is where gas prices are heading, and what rising renewable penetration does to European gas demand. That is Section 2.

2. Gas and LNG: Pricing Power and Its Limits

2.1 The LNG market

Natural gas arrives in Europe through two channels: pipelines, which are fixed in geography and capacity, and liquefied natural gas, which is not. LNG, natural gas cooled to −162°C, reducing its volume by a factor of 600 for maritime transport, has transformed European gas supply from a regionally constrained market into one directly connected to global production economics. That transformation is now complete, and its consequences for electricity prices are only beginning to show up in forward curves and project finance models.^[7]

The infrastructure underpinning this market is capital-intensive at every node. Liquefaction terminals, the largest of which, Qatar's Ras Laffan complex, handles 77 Mtpa from a single industrial city on the Persian Gulf, cost US$3–5 billion per train and take four to six years to build. Once operational, they run continuously at nameplate capacity regardless of price: the economics of the asset class do not permit throttling output in response to short-term market signals. Shipping requires dedicated cryogenic vessels costing US$250–255 million each, incapable of carrying any other cargo. At the receiving end, European regasification capacity expanded from near zero in Germany in 2021 to 19.3 Mtpa by end-2025, almost entirely through floating storage and regasification units deployed in under 18 months.

The result is a value chain in which fixed costs dominate at every stage, production decisions are driven by long-run contract economics rather than spot prices, and the marginal molecule reaching Europe is determined by where global supply exceeds global demand, not by European demand alone. As of 2025, Europe imported a record 126.2 Mt of LNG, with the United States accounting for 58.5% of that volume. The Atlantic Basin had become structurally long, and Europe was absorbing the surplus.

Global LNG supply chain — liquefaction, shipping, and European import terminals

↗ View full chart

2.2 How TTF is formed and why it matters

TTF, the Title Transfer Facility, the Dutch virtual trading hub that serves as the European wholesale gas benchmark, is set by the marginal clearing price across five drivers: the supply-demand balance in the short term, European storage dynamics and their seasonal premium, LNG import availability and the cost of the marginal cargo reaching Northwest Europe, carbon price expectations embedded in long-run dispatch economics, and tail-risk supply shock premiums of the kind the Hormuz crisis has injected into the 2026 spot price.

Of these, the power sector is the critical driver for this analysis. Gas CCGT plants are price-setting in roughly 40–60% of European electricity hours depending on season and renewable penetration. Power sector gas demand therefore feeds directly back into TTF formation, and changes in that demand are the primary mechanism through which rising renewable penetration suppresses gas prices. Norwegian pipeline supply of roughly 80–90 BCM per year provides the continental baseline; any shortfall below that level is filled by LNG imports, making LNG the marginal price-setter for European gas in all but the most supply-flush conditions.

The forward curve embeds the market's consensus view of how these drivers evolve. As of 17 July 2026, that consensus is striking in what it has already priced in. The spot price of approximately €52/MWh reflects the Hormuz crisis premium, US strikes on Iranian coastal defences have effectively closed the strait, locking Qatar's 77 Mtpa behind a 54-kilometre chokepoint and pushing current prices to three-month highs. But the forward curve looks through this: March 2027 futures trade at €45.84/MWh, April 2027 at €38.42, a €7.42 cliff in a single month, the market's best estimate of when the crisis normalises. By summer 2027 the curve settles near €35. The second structural cliff arrives at the Q1–Q2 2028 boundary, where March 2028 trades at €32.30 and April 2028 at €27.13, the market pricing in the first tranches of the US supply wave arriving in volume. By summer 2029, July futures trade at €21.91. By summer 2030, €20.60, within or marginally above the US Gulf Coast short-run marginal cost floor of €15–22/MWh.^[14]

The market has therefore substantially priced the supply wave. What it has not priced is what the supply wave, compounded by rising renewable penetration, does to electricity markets. That is the argument of this paper.

TTF natural gas forward curve — the market has priced the supply wave

↗ View full chart

2.3 The supply overhang

Global liquefaction capacity stood at 524.5 Mtpa at end-2025, up 30 Mtpa in a single year. It is projected to surpass 700 Mtpa by 2030, a 40% increase over five years. Every project driving that growth has already taken a final investment decision. The supply wave is not a forecast; it is a construction schedule.^[7]

The projects are predominantly American. Plaquemines LNG in Louisiana ramped to near-nameplate capacity of 20 Mtpa during 2025, built using a modular small-train design that enabled faster execution than conventional large-train facilities. Golden Pass LNG delivered its first cargo in April 2026 at 17.8 Mtpa. CP2 LNG and Port Arthur Phase 1, each adding 13–14 Mtpa, are scheduled for first trains in 2027. Rio Grande Phase 1 at 17.6 Mtpa and Port Arthur Phase 2 at a further 13 Mtpa follow in 2028. Woodside Louisiana LNG adds 16.5 Mtpa in 2029. US Gulf Coast capacity additions alone represent approximately 100 Mtpa by 2030, nearly one fifth of current global liquefaction capacity, arriving in a market that was already long.

Qatar's North Field expansion, which would have added 32 Mtpa across four new mega-trains, is now delayed at least one year from its original Q3 2026 target following Iranian missile strikes on Ras Laffan in March 2026, with two existing trains carrying 12.8 Mtpa nameplate capacity offline for an estimated three to five years. The delay provides a temporary reprieve for gas prices in the 2026–2027 window but does not alter the medium-term trajectory: the US supply alone is sufficient to drive the Atlantic Basin into structural oversupply by 2028–2029.

The price floor in that scenario is the short-run marginal cost of US Gulf Coast liquefaction, approximately US$5–7/MMBtu all-in, equivalent to roughly €15–22/MWh at the electricity level. At those prices, not a single gas CCGT in Europe covers its variable cost.

Full supply wave table (FID projects)

Project	Capacity	Start
Plaquemines LNG (US)	20 Mtpa	Ramped 2025
Corpus Christi Stage 3 (US)	6.0 Mtpa	2025
LNG Canada T1 (Canada)	7.0 Mtpa	2025
LNG Canada T2 (Canada)	7.0 Mtpa	Mid-2026
Golden Pass LNG (US)	17.8 Mtpa	First cargo Apr 2026
Corpus Christi Stage 3 remaining (US)	5.3 Mtpa	2026
CP2 LNG (US)	14.4 Mtpa	First trains 2027
Woodfibre LNG (Canada)	2.1 Mtpa	2027
Port Arthur LNG Phase 1 (US)	13.0 Mtpa	2027
Rio Grande Phase 1 (US)	17.6 Mtpa	2028
Port Arthur Phase 2 (US)	13.0 Mtpa	2028–2029
Woodside Louisiana LNG (US)	16.5 Mtpa	2029
Qatar NFE T8–11	32 Mtpa	Delayed ≥1yr from Q3 2026

Source: IGU World LNG Report 2026 / article notes §2.3 · All projects shown have already taken FID. US Gulf Coast alone ≈100 Mtpa by 2030.

2.4 Europe as the clearing market

When the Atlantic Basin is long, when more LNG is available than Northeast Asia can absorb at prevailing prices, the surplus flows to Europe by default. In 2025, the JKM/TTF spread averaged minus US$0.71/MMBtu: European delivery was systematically cheaper than Asian diversion for Atlantic-origin cargoes throughout most of the year. Europe absorbed a net 26.1 Mt increase in LNG imports, reaching a record 126.2 Mt, partly driven by the cessation of Russian pipeline gas transit via Ukraine in early 2025, which removed 17.7 BCM of annual supply and forced replacement from global LNG markets.^[7]

The structural consequence is that European gas prices are no longer primarily determined by European supply and demand. They are determined by the global LNG market clearing price, itself determined by the intersection of US production economics, Asian spot demand, and the routing decisions of destination-flexible cargo traders. NWE LNG averaged US$0.54/MMBtu below TTF in 2025, significantly wider than the 26-cent discount in 2024, reflecting the ease with which supply was finding its way to European terminals. By 2028–2030, as the full US supply wave clears through the market, that discount could widen further, pushing TTF toward the SRMC floor not as a tail scenario but as the central case.

2.5 The JKM/TTF spread and cargo routing

The mechanism through which global LNG supply reaches European gas prices is the JKM/TTF spread, the difference between the Japan-Korea Marker, the Asian spot benchmark, and TTF. When JKM exceeds TTF by more than the shipping differential of roughly US$1.0–2.0/MMBtu depending on route and vessel type, Atlantic cargoes divert east; when TTF exceeds JKM plus the differential, they flow to Europe. In 2025, the spread averaged minus US$0.71/MMBtu, meaning the arbitrage systematically directed US Gulf Coast cargoes to the closest and cheapest available market, Europe.

Destination-flexible contracts, the standard structure for US LNG offtake, where the buyer has no restriction on delivery terminal, are structurally equivalent to a shipping option. The offtaker holds the right but not the obligation to deliver to any accessible terminal, and the value of that option is a direct function of JKM/TTF spread volatility. Large portfolio traders, Shell, TotalEnergies, Vitol, Trafigura, hold these books and make routing decisions continuously based on real-time netback comparisons. When the Hormuz crisis pushed JKM to US$25.41/MMBtu on 19 March 2026 against TTF at US$22.50/MMBtu, at least eight Atlantic Basin cargoes diverted east within days, each representing approximately US$3–4 million of incremental netback value per vessel to traders with flexible positions. The speed of that response illustrates how tightly the two markets are now coupled, and how quickly European supply balances can shift when Middle Eastern routes are disrupted.

For the structural argument, what matters is the direction of the spread over the supply wave period. As US capacity adds ~100 Mtpa to the Atlantic Basin by 2030, the structural surplus persistently favours European delivery over Asian diversion, keeping TTF under sustained downward pressure regardless of short-term Hormuz volatility. The forward curve has priced this in. What follows analyses what it means for the electricity market.

JKM/TTF trading mechanics

JKM is the Asian LNG spot benchmark. The JKM/TTF spread decides where flexible Atlantic cargoes go: when JKM exceeds TTF by more than the shipping differential (~US$1.0–2.0/MMBtu), cargoes divert east; when TTF is richer, they flow to Europe.

In 2025, the spread averaged minus US$0.71/MMBtu, systematically directing US Gulf Coast cargoes to Europe. On 19 March 2026, Hormuz risk spiked JKM to US$25.41/MMBtu against TTF at US$22.50/MMBtu and pulled at least eight Atlantic Basin cargoes east within days.

Physical traders hedge with TTF and JKM futures/swaps; destination-flexible offtake contracts are effectively a shipping option whose value rises with spread volatility. As the US supply wave arrives, the structural bias is toward European delivery and sustained TTF pressure toward the US Gulf Coast SRMC floor.

Source: IGU 2026 / article notes §2.5 · Full JKM/TTF chart lands in Gas & LNG Stage 3 (§5.4).

2.6 Spark spread mechanics and the gas plant economics

For a standard CCGT operating at 55% thermal efficiency, a conventional reference assumption for illustrating the mechanism rather than the specific plant modelled later in this paper, which achieves a higher 65% and is analysed in full in Section 5, with a carbon intensity of 0.40 tonnes of CO₂ per MWh, the variable cost formula is:

Variable cost = TTF ÷ 0.55 + EUA × 0.40

At current spot TTF of approximately €52/MWh and EUA of €55/t, that yields a variable cost of roughly €94.5 + €22 = €116.5/MWh. The plant dispatches only when the electricity price clears above that level, which at current prices happens in relatively few hours. At the 2027 summer forward of €35/MWh TTF, variable cost falls to approximately €63.6 + €22 = €85.6/MWh, which makes the plant viable in more hours. At the 2029–30 summer forward of €22/MWh TTF, variable cost falls to roughly €40 + €22 = €62/MWh, cheap enough to set the electricity price in a significant share of hours.

The apparent paradox is that lower TTF improves the spark spread per running hour by reducing variable cost, while simultaneously reducing running hours as electricity prices fall with gas costs. The net effect on total annual profit is deeply negative. A plant that earned €40,000/MW net of fixed costs at 3,000 running hours and a €20/MWh spread earns approximately €7,000/MW at 1,800 running hours and a €15/MWh spread, an 83% collapse in economic profit before the full demand destruction from renewable penetration is incorporated. Fixed costs of roughly €20,000/MW per year are unchanged throughout.

The stranded asset trajectory does not require the gas plant to become technically inoperable. It requires only that its expected lifetime revenues fall below the sum of its fixed costs, a condition that is now arithmetically visible in the forward curve for merchant gas plants without long-term capacity contracts.

2.7 Who holds the risk

The structural compression of gas plant economics, LNG infrastructure returns, and utility contract books is not evenly distributed. Three groups are exposed in ways that are not yet fully reflected in asset valuations or disclosed as balance sheet risk.

Gas plant equity holders are running on a trajectory of declining running hours and compressed spreads. Fixed cost recovery is increasingly dependent on capacity market payments and ancillary services revenue, neither of which is structurally sized to compensate for the full loss of merchant energy revenue at high renewable penetration. The merchant gas plant business model is structurally impaired at 50–65% renewable penetration, and European markets are approaching that range.

LNG regasification infrastructure investors deployed capital at 2022–2023 throughput assumptions, when European LNG demand appeared structurally elevated. Global regasification utilisation was 39.2% in 2025, across 1,113.5 Mtpa of installed capacity, before a further 229.3 Mtpa of new terminals under construction come online. Assets with 20-year economic lives, financed on demand assumptions set at the peak of the post-Ukraine supply crisis, face the same structural revenue erosion as the gas plants they feed. The early signal is already visible: Deutsche Regas terminated the use of one FSRU at the Mukran facility in 2025 due to uncompetitively high transportation and marginal regasification costs.

Utilities holding unhedged oil-indexed LNG contracts sit on a less visible but potentially larger exposure. Long-term take-or-pay contracts indexed at Brent × 12–14%, the standard structure for contracts signed at the height of the 2022–2023 supply crisis, imply a delivered cost of approximately US$10.4/MMBtu at Brent US$80/bbl. The 2029–30 TTF summer forward implies a European market clearing price of US$6–7/MMBtu. A utility locked into oil-indexed supply at US$10.4/MMBtu, selling into a market that clears at US$6–7/MMBtu, is 30–50% out of the money on a take-or-pay obligation. That exposure sits across several major European utility balance sheets and is not transparently reported as a mark-to-market liability.

The forward curve has priced the gas supply side of this equation correctly. The electricity market implications, what cheap gas at €22–27/MWh TTF does to electricity prices, and what those electricity prices do to renewable project finance, are the subject of Section 3.

3. Renewables: The Cannibalisation Problem

3.1 The merit order and the cannibalisation mechanism

European electricity prices are set by the merit order: generators dispatched in ascending order of short-run marginal cost, the marginal generator sets the clearing price for all producers in that hour. Coal clears at roughly €40–60/MWh, gas CCGTs at €80–100/MWh, nuclear at €10–20/MWh, renewables at approximately zero. When renewable capacity is small relative to total demand, it displaces the cheapest thermal generators and has modest price impact, the marginal gas plant still sets the price in most hours, and the system functions as designed.

As renewable capacity grows, it begins to set the marginal price directly in the hours it produces. Since its short-run marginal cost is effectively zero, it drives the clearing price toward zero in exactly those hours. The self-defeating logic follows immediately: a renewable developer's own output, combined with the simultaneous output of every other generator responding to the same weather conditions, drives down the price in the hours it produces. Adding more capacity of the same type lowers revenue per MWh for every existing project of that type. This is not a market failure, it is the merit order functioning correctly. The problem is that the long-run cost structure of renewable energy, driven entirely by capital expenditure rather than fuel, cannot be recovered through a market mechanism that prices electricity at short-run marginal cost.

3.2 The load duration curve and the bimodal distribution

The load duration curve, demand ranked from highest to lowest across the 8,760 hours of a year, illustrates the structural shift directly. At low renewable penetration the curve is relatively smooth: high-demand periods clear at high prices, low-demand periods at lower prices, and the annual average price is a reasonable proxy for what a baseload generator captures over the year.

As penetration rises, the residual load duration curve, demand minus renewable output, becomes distorted into a bimodal shape. A large cluster of hours at near-zero residual demand accumulates at one end: daytime solar production in spring and summer, windy overnight periods, hours when renewable output routinely meets or exceeds demand. These clear at zero or negative prices. At the other end, a smaller but growing cluster of hours at very high residual demand, dark, windless winter evenings when dispatchable capacity is squeezed into covering the full load, clears at sharp scarcity prices of €200–500/MWh or above. The middle of the distribution, where gas plants historically ran profitably at moderate spreads across thousands of hours per year, compresses steadily as penetration rises.

The ENTSO-E hourly load data for 2024–2025 quantifies this compression in real time. In France, 85.1% of summer 2025 hours fell below 55% of annual peak demand, up from 73.4% in summer 2024. In Germany, the number of summer hours below 55% of annual peak rose from 544 in 2024 to 581 in 2025. These are not modelled projections. They are the empirical load duration curve narrowing year-on-year in live market data.^[5]

A merchant renewable developer's revenue is determined almost entirely by the left end of this distribution, because that is when their assets produce. They capture almost none of the high scarcity-price hours at the right end, those occur precisely when wind is calm and solar irradiance is low. The annual average electricity price of €70–80/MWh is therefore irrelevant as a planning assumption for a wind developer at 50–65% penetration. The relevant metric is the captured price during production hours, annual average multiplied by the market value factor, and that number is deteriorating structurally.

3.3 Negative price hours, the empirical acceleration

Negative electricity prices are the most direct empirical signal of cannibalisation in progress. They occur when renewable output is sufficient to meet demand and dispatchable generators, unable to shut down quickly or facing negative costs of curtailment, offer negative prices rather than reduce output. Their frequency is accelerating across every major European market.^[8][13]

Germany recorded 457 negative-price hours in 2023, 459 in 2024, and over 570 in 2025, approximately 6.5% of the year, up from roughly 3% in 2022. The Netherlands moved from 316 hours in 2023 to 458 in 2024 and 584 in 2025. Spain, where negative prices were not permitted by regulation until December 2023, recorded 247 hours in 2024 and over 500 in 2025, effectively doubling in a single year. France rose 45% year-on-year in 2025; Belgium exceeded 450 hours. Across European markets in aggregate, 2024 recorded 4,838 negative or zero-price hours, nearly double 2023, and 2025 exceeded 2024 levels in most markets.^[6]

The seasonality of these hours is telling. In 2025, 47% of all negative-price hours across several markets occurred in May and June alone, representing 18.4% of all hours in those two months. This is pure solar cannibalisation: midday surplus in peak solar season, with no dispatchable capacity able to absorb the output at positive prices. The average depth of negative prices in Germany was −€12.9/MWh in 2023, falling modestly to −€10.9/MWh in 2025 as the distribution widens, more hours at moderate negative prices rather than fewer hours at extreme ones. Aurora's forecasts project German negative-price hours reaching 5–10% of annual hours by 2030 without material storage deployment, and pan-European curtailment rising from 10 TWh in 2024 to approximately 33 TWh by 2030.^[9]

Negative electricity price hours by country, 2023–2025

↗ View full chart

3.4 The market value factor, empirical capture rates

The market value factor measures the ratio of a technology's average captured price to the market average in the same period. An MVF of 1.0 means the technology earns exactly the market average; below 1.0 means it systematically sells into below-average hours. The MVF is the single most important metric for understanding renewable project economics at scale, and it has been deteriorating faster than most project finance models assumed.

Hirth (2013) established the foundational empirical relationship: German wind MVF of approximately 0.92 at 5% penetration, declining to 0.80 at 35%, with projections of 0.50–0.60 at 65%. The curve is convex, the same increment in penetration has a disproportionately larger impact at high penetration than at low.^[1] The European Commission's 2023 discussion paper on renewable market value factors extends this analysis using six-country modelled data through 2030, and finds the same convex decay pattern holding across markets with materially different generation mixes, evidence that Hirth's original relationship generalises beyond the single-country case it was built on.^[16]

The live data confirms this. German solar captured €71.55/MWh in October 2025 against a market average of €84.40/MWh, an implied MVF of approximately 0.85 for that month, on a trajectory toward 50% by 2030 according to Veyt.^[10] Spanish solar averaged an MVF of 0.61 across 2025, with a monthly low of 0.40 in April 2024 during peak solar season. The volume-weighted average Spanish solar capture price fell 12% year-on-year to just under €40/MWh in 2025. German solar average capture prices dropped below €45/MWh in H1 2024; Spanish solar below €20/MWh in the same period.^[12] Offshore wind remains more resilient, German offshore achieved an MVF of approximately 0.95 in 2025, UK offshore 0.91, but the trajectory is set: Aurora projects solar discounts approaching 50% in Iberia and onshore wind discounts exceeding 25% in Germany by 2030.^[9]

The non-linearity matters for investment decisions. Spain passed its inflection point abruptly, zero negative-price hours before 2024, over 500 in 2025 in a single step. Germany is now on the steep part of the Hirth curve. The projects being financed today will operate in the market that exists in 2035–2045, not the market that exists now.

The cannibalisation curve — MVF vs penetration

↗ View full chart

3.5 PPA prices, the market's own signal

Power purchase agreements are bilateral contracts between a renewable developer and a corporate offtaker fixing a price for 10–15 years. They are the most direct observable market signal of cannibalisation expectations, sophisticated counterparties pricing what they believe renewable output will be worth over the contract term.

That signal is unambiguous. European solar PPA prices fell below €35/MWh in Q3 2025, down 19.4% year-on-year, a record low and a decline far in excess of any change in underlying generation costs.^[11] German and Spanish solar PPAs are now below €40/MWh. European onshore wind blended at €52.75/MWh in Q3 2025, down 3% quarter-on-quarter. The five-year German onshore wind PPA fair value stood at €76.20/MWh in mid-2024, 17% below the one-year value, with cannibalisation expectations embedded in the longer-dated price.^[10]

The gap between PPA prices and project breakevens is now structural. New onshore wind projects starting construction in Germany in 2027 require approximately €78/MWh to be bankable, against a maximum auction bid price under the EEG of just below €70/MWh, an €8/MWh gap that makes merchant and auction-route projects simultaneously unbankable. German, Dutch, and Danish offshore auction rounds have attracted zero bids. Lithuania secured a single bidder. These are not signals of temporary cost inflation, they are the market confirming that the DSCR cliff described in the following section has arrived in live pricing.^[9]

PPA prices vs breakeven

↗ View full chart

3.6 The DSCR cliff and project finance bankability

Offshore wind projects are financed as non-recourse special purpose vehicles: lenders have recourse only to project cash flows, with no claim on the developer's balance sheet. Revenue flows through a defined waterfall, operating costs, debt service, reserve account top-up, equity distributions, with cash lock-up covenants triggering at DSCR below 1.10–1.15x and technical default at approximately 1.00x. The DSCR is simply annual cash available for debt service divided by annual debt service obligations.

For a reference project, 400 MW offshore wind, €2.8M/MW capex, 75% leverage, 18-year tenor at EURIBOR+165bps, the arithmetic at different penetration levels tells a clear story. At 35% penetration with an MVF of 0.90 and a captured price of €74/MWh, annual revenue is approximately €284M, EBITDA €254M, and DSCR 3.2–3.4x. Equity IRR is approximately 15%. The project is comfortably bankable. At 50% penetration, MVF 0.75, captured price €60/MWh: DSCR falls to 2.5–2.7x, equity IRR to approximately 10%. Still bankable, but the margin is narrowing. At 62% penetration, Spain's actual 2025 MVF of 0.62, captured price falls to €42/MWh, DSCR to 1.6–1.7x, and equity IRR to near zero. Rational equity investors stop committing capital at this point, well before any covenant triggers. At 65% penetration, DSCR is 1.4–1.5x: one below-P90 wind year pushes the project into cash lock-up. At 70%+, systematic covenant breach.

The equity IRR is destroyed long before legal default. The market shuts down for new capital at the point where expected returns no longer justify the risk, which is precisely what German, Dutch, and Danish auction failures are confirming.

Offshore wind project finance — the bankability cliff

↗ View full chart

Full DSCR table

Penetration	MVF	Captured price	Annual revenue	EBITDA	DSCR	Equity
35%	0.90	€74/MWh	€284M	€254M	3.2–3.4x	~15% IRR
50%	0.75	€60/MWh	€230M	€200M	2.5–2.7x	~10% IRR
62%	0.62	€42/MWh	€161M	€131M	1.6–1.7x	IRR near zero
65%	0.55	€37/MWh	€142M	€112M	1.4–1.5x	Covenant risk
70%+	0.45	€30/MWh	€115M	€85M	~1.1x	Systematic breach

Reference project: 400 MW offshore, capex €2.8M/MW, 75% leverage, 18-year tenor · Source: Article notes §3.8

3.7 The P90 blind spot

The standard risk framework used across European project finance does not capture this deterioration. P90, the metric to which lenders size debt, is a probabilistic energy yield estimate: there is a 90% probability the project will produce at least P90 MWh in any given year. Banks lend to P90 revenues covering debt service at DSCR 1.10–1.20x, a methodology universally accepted across European infrastructure lending markets and correctly calibrated for the risk it measures: wind resource variability, turbine availability, wake losses, curtailment.

The problem is that P90 underwrites the left side of the revenue equation, megawatt-hours produced, and not the right side: the price per MWh achieved in the market. The full revenue equation is MWh produced multiplied by captured price. The captured price is not determined by the wind resource. It is determined by the market structure in the year of delivery, specifically by how many other projects of the same type are producing simultaneously in that hour, and what penetration level the market has reached by then. A project commissioned in 2024 will operate through a market that, by the mid-2030s, may be running at 60–65% renewable penetration. No current P90 methodology models what that does to the price the project receives per MWh.

The consequence is that a project can hit its P90 yield target exactly, producing precisely the volume the lender underwrote, and still miss its revenue forecast by 40–50% if the MVF has deteriorated over the asset's operating life. The yield risk crystallises as modelled; the revenue miss is entirely explained by a price distribution shift that the underwriting framework never captured. Lenders charging EURIBOR+150–175bps for investment-grade offshore wind are pricing yield risk correctly and captured-price risk not at all, a mispricing that will become visible in asset performance data before it shows up in default statistics.

P90 numerical example

2021 underwrite €85.0/MWh P90 AEP 1.43 TWh → revenue €121.6M

→

Year 12 realised (same P90 yield) €45.5/MWh MVF 0.65 × €70 market → revenue €65.1M (−46%)

P90 underwrites MWh produced, not €/MWh captured. A German offshore project can hit its P90 yield exactly and still miss underwritten revenue by nearly half once cannibalisation has moved the market value factor. Lenders pricing EURIBOR+150–175bps are pricing yield risk correctly and captured-price risk not at all.

Source: Article notes §3.9

3.8 CfDs, PPAs, and the limits of partial solutions

Contracts for difference and power purchase agreements are the market's response to the bankability problem, and they work, at the project level. A CfD eliminates cannibalisation risk for covered output by guaranteeing a fixed strike price regardless of the market clearing price during production hours. CfD projects finance at EURIBOR+100–130bps with leverage up to 80–85%. A PPA achieves similar revenue certainty bilaterally, without government backstop.^[3]

Both have structural limits that are now binding. CfD auctions in Germany, the Netherlands, and Denmark have attracted zero bids even at strike prices the government considered adequate, the gap between achievable strike price and project breakeven has widened beyond what the auction mechanism can close. The clawback obligation embedded in the two-sided CfD structure became politically toxic in 2022–2023 when market prices exceeded strike prices, triggering large transfers from developers to government and generating sufficient political pressure that retroactive modification of contract terms was discussed in several member states, an outcome that, if it occurred, would undermine the very revenue certainty the instrument exists to provide. And the fiscal contingent liability of running CfD auctions at the scale the EU's 2030 targets require, 400–500 GW of new offshore wind, represents tens of billions of euros of annual exposure across member state balance sheets.

PPA markets are sending a cleaner signal. Solar PPA prices at €35/MWh are below long-run cost for new projects, the market is pricing cannibalisation expectations into the offer price, and developers cannot build at those prices. The five-year German onshore wind PPA has fallen from roughly €80/MWh in 2022 to €70–72/MWh today, tracking the cannibalisation curve as buyers model the deteriorating captured price into their willingness to pay. The €25/MWh gap between what a German onshore wind project needs to be bankable (€78/MWh) and what the PPA market currently pays (€52.75/MWh blended Europe) is not a negotiating position, it is the market's aggregate assessment of where captured prices are heading.

CfDs and PPAs redistribute cannibalisation risk: from the developer to the government in the first case, from the developer to the corporate offtaker in the second. They do not remove it from the system. The 2023 EU electricity market reform left this systemic allocation entirely unaddressed, the uniform marginal pricing mechanism was preserved, gas remained the marginal price-setter in the majority of hours, and the structural incompatibility between short-run marginal cost pricing and long-run cost recovery for zero-marginal-cost generation was carried forward intact.^[4] Section 4 quantifies how the two channels, gas price compression and renewable cannibalisation, interact when modelled together.

4. The Interaction Model

4.1 Why the two channels must be modelled together

Section 2 showed that rising renewable penetration reduces power-sector gas demand and structurally lowers TTF. Section 3 showed that rising penetration floods renewable-heavy hours and depresses the spot price directly. Treated separately, these look like two distinct problems with two distinct magnitudes. They are not separate. Gas CCGT marginal cost is a direct function of TTF: (TTF ÷ 0.55) + (EUA × 0.40). When power-sector gas demand falls because renewables have displaced dispatchable running hours, TTF falls with it, and gas marginal cost falls across every hour in which gas remains the price-setter, not only the hours renewables are producing. The entire merit order stack shifts down. Even the revenue a wind farm earns during its low-output hours, when gas is setting the price, is compressed by a mechanism that has nothing to do with wind output.

How large that mechanism is in practice is itself contested territory. Zakeri et al. (2023), using econometric analysis of sub-hourly ENTSO-E data across the EU-27 plus Great Britain and Norway for 2015–2021, find that fossil generation set the marginal price roughly 58% of the time across the bloc in 2021, with gas alone responsible for 39 percentage points of that, despite gas supplying only 18% of electricity generated.^[15] The dispersion across countries is wide: in Belgium, Great Britain, Greece, Italy, and the Netherlands, gas set the price in more than 80% of hours the same year, while other markets with different generation mixes sat far lower. This is the empirical basis for treating gas as the dominant marginal technology in this model's price-formation logic, and it is also the reason a single-country calibration inevitably understates the range of outcomes a pan-European reader should expect, a limitation the model does not paper over.

The causal chain runs in one direction as follows: renewable penetration rises, electricity prices fall during renewable hours (direct), gas plants run fewer hours, European power-sector gas demand falls, TTF falls, gas CCGT marginal cost falls, electricity prices fall during gas hours too (indirect), merchant renewable revenue falls further than either channel alone would predict. The reverse direction produces the asymmetry that matters for financing structures: a gas price shock raises electricity prices during gas-dominated hours, producing windfall revenue for unhedged renewable generators, revenue that contracted structures like two-sided CfDs are specifically designed to claw back. The financing consequence of that asymmetry is not theoretical. Đukan, Keles and Kitzing (2025), modelling offshore wind cash flows under stochastic power prices, find that two-sided CfD protection allows lenders to size debt 15 to 27 percentage points higher than an equivalent merchant project, with the resulting leverage saving consumers an estimated €12–19/MWh in generation costs.^[2] That is the DSCR cliff from Section 3.6, expressed as a financing outcome rather than a coverage ratio, and it is the clearest quantitative evidence that contracted revenue is not a policy preference but a measurable difference in what a project can borrow.

4.2 Inside the model: architecture and method

The interaction model built for this analysis is a comparative-statics tool, not a forecasting engine. It reimplements a set of empirical relationships, calibrated to Zakeri et al. (2023) for gas price-setting frequency,^[15] the European Commission's 2023 discussion paper for market value factors,^[16] and ENTSO-E's negative-price-hour series, directly, rather than simulating hourly dispatch from first principles. It covers three technologies (merchant gas CCGT, solar, onshore wind) against a single reference market, Germany, chosen for data richness and liquidity rather than representativeness. The Zakeri cross-country dispersion cited above is the clearest illustration of why that choice matters: French nuclear baseload, Spanish solar profiles, and the UK's separate CfD regime would each pull the gas-price-setting share, and therefore the whole model's output, in a different direction. Germany sits closer to the EU-wide average than the >80% outlier markets, which is a defensible reason to use it as the reference case, but it is a choice, not a neutral default, a limitation returned to in Section 5.

Each scenario is defined by a small set of levers: a target year, solar and wind capacity as a percentage of Germany's locked EEG 2030 buildout targets (215GW solar, 115GW wind), a wind resource year, a demand-growth path anchored to the NEP Szenariorahmen, and an optional gas-price stress applied on top of the underlying TTF forward curve. Carbon enters via the locked ICE EUA Dec futures path (17 July 2026 settlement, Dec-26 through Dec-32, then 2%/yr), not a flat scalar. The model computes a single clearing price per scenario by weighting three mutually exclusive states, hours gas sets the price, hours the price is zero or negative, and a residual bucket calibrated to a real 2025 anchor point, and does not compute cannibalisation and TTF suppression as separate intermediate figures before combining them. This is a deliberate feature of the underlying mechanism, not a simplification introduced for this analysis: because gas SRMC and renewable penetration jointly determine the same weighted formula, any attempt to isolate one channel while holding the other's natural trajectory fixed produces an artificial number that the real market would never generate. A future iteration of the tool is expected to expose a cleaner isolation control for readers who want to stress-test the channels independently; the results presented here are the model's honest combined output.

4.3 Model outputs: the compound effect, quantified

Running the model at the German 2025 baseline, current renewable capacity, Scenario B demand growth, no supply disruption, produces an average clearing price of €80.4/MWh, in line with the calibration anchor the model was validated against. Advancing to 2030 on Germany's locked EEG trajectory, with every other input held at the same settings, drops that price to €42.2/MWh, a fall of 47.5%. This is the compound effect the previous two sections argued for, expressed as a single number: roughly half the electricity price disappears between today's penetration level and the 2030 target, before any supply shock is introduced at all.

The TTF path underlying each scenario shows why. In the 2025 case, the forward curve carries the current Hormuz-related premium before settling toward the mid-€20s by the early 2030s. In the 2030 case, the same curve has already rolled forward past that premium, and the lower gas marginal cost this implies feeds directly into a lower clearing price in every hour gas remains the price-setter, the indirect channel described in 4.1, visible here as a shift in the underlying cost curve rather than a separate output.

The price-duration curve makes the distributional shift concrete in a way the average alone cannot. At 2025 penetration, the sorted-hours curve is relatively flat across most of the year, with a modest tail. At 2030 penetration, the curve compresses sharply in its upper blocks and falls away faster toward zero, a larger share of hours cluster near the bottom of the distribution, consistent with the market value factor decay described in Section 3. The annual average captures the level of this shift; the duration curve captures its shape.

The supply mix chart underlying these scenarios explains why the shock's relative sensitivity holds steady rather than fading. Renewable generation rises roughly linearly across the horizon, from around 230 TWh in 2025 to nearly 700 TWh by 2045. Gas and other residual generation, the model bundles the two because it does not separately track smaller categories, does not correspondingly collapse. It rises modestly through the early 2030s, peaks in the high-2030s, and only then eases back toward its starting level by 2045. In absolute terms, gas remains a meaningful part of the German generation stack throughout the horizon; what changes is its share of a system that has roughly tripled in total output. A gas supply shock still transmits through the system at close to the same relative strength in 2030 as in 2025, because gas has not been displaced from the margin so much as diluted within a much larger whole.

Layering a supply shock, the same Hormuz/Qatar disruption used in Section 2, calibrated to a 22.6% gas-price stress, onto each penetration level shows the asymmetry the model was built to test. At 2025 penetration, the shock lifts the price to €93.4/MWh, a 16.2% increase. At 2030 penetration, the same shock lifts the price to €49.1/MWh, a 16.4% increase. In absolute terms the shock is worth less at higher penetration, €13.0/MWh in 2025 versus €6.9/MWh in 2030, simply because the base it is applied to has already fallen by half. But in relative terms, the market's sensitivity to a gas supply shock is essentially unchanged. Higher renewable penetration does not insulate the price from a gas shock; it only shrinks the shock's absolute euro footprint, because the structural decline in price has already done most of the work. This is a sharper and more defensible claim than the intuitive assumption that renewables dampen gas-price transmission, the model shows the transmission mechanism intact, just applied to a smaller base.

Across the four combinations tested, the 2025 baseline prices power at €80.4/MWh. Adding the Hormuz/Qatar disruption to 2025 lifts this to €93.4/MWh, an increase of 16.2%. Moving to 2030 penetration alone, with no supply disruption, lowers the price to €42.2/MWh, a fall of 47.5% versus the 2025 baseline. Adding the same Hormuz/Qatar disruption to the 2030 case brings the price to €49.1/MWh, still 38.9% below the 2025 baseline despite the shock.

The structural effect dominates the shock effect by an order of magnitude. Whatever happens to Gulf shipping lanes or Qatari export capacity in any given year, the larger and more certain move in European electricity prices over the rest of this decade is the one the model produces with the supply-scenario dial left untouched.

4.4 The 2023 EU reform: what it addressed and what it didn't

The EU's response to the 2021–2022 price crisis took final legislative form as Regulation (EU) 2024/1747 and its companion Directive (EU) 2024/1711, entering into force on 16 July 2024.^[4] Its core mechanism, per Article 19a of the Regulation, requires member states to use two-sided CfDs or an equivalent scheme whenever they provide public financial support for new low-carbon generation: a guaranteed strike price on the downside, and a clawback of revenue above strike on the upside. Alongside this sits continued promotion of PPAs, peak-shaving products, and non-fossil flexibility support, a broadening of the toolkit for contracting revenue, not a replacement of the spot market itself. The reform preserves marginal pricing on spot and intraday markets for operational efficiency, while generalising bidirectional CfDs for supported capacity from 2027 onward. The reform is, by design, an addition to the merit order, not an amendment to it.

That design choice is precisely what Sections 4.1–4.3 make legible. The reform's clawback mechanism is aimed at the asymmetric windfall problem: the Đukan, Keles and Kitzing (2025) finding that two-sided CfDs let lenders size debt 15–27 percentage points higher than an equivalent merchant structure is the reform's central justification made quantitative.^[2] On that count, the reform is a real improvement, it converts an ad hoc emergency measure (the €180/MWh ex-post inframarginal revenue cap imposed in October 2022 and formally allowed to lapse by 30 June 2024) into a standing, forward-looking mechanism generators can plan financing around.

What it does not touch is the mechanism this section has spent three subsections demonstrating: gas remains the marginal price-setter across the majority of hours the model examines, the merit order is unchanged, and the compound cannibalisation-plus-TTF-suppression effect quantified in 4.3, a 47.5% price fall between 2025 and 2030 penetration, before any supply shock, occurs identically whether the reform exists or not. The reform governs who bears the revenue swing once the spot price has moved. It has no bearing on why the spot price moves the way it does. A regulator reading this model's output would find nothing in the 2024 reform that changes a single number in Section 4.3.

There is a deeper version of this critique worth stating plainly, even though it runs against the reform's own stated intent. As more new-build capacity moves onto administered CfD revenue by construction, the regulation's explicit goal, a shrinking share of total generation is exposed to the marginal spot price the whole system still uses as its reference signal. A thinner spot market is a noisier investment signal, and the next generation of projects seeking financing will lean even more heavily on contracted structures precisely because the market they might otherwise reference is less informative than it used to be. Whether this becomes self-reinforcing, each successive capacity round justified partly by the previous round's success at removing supply from the price-forming market, is not something this model tests, and no serious proposal for the deeper redesigns (capacity markets, nodal pricing) that would address it directly is currently on the table in Europe. The reform solved the problem it was built to solve. It left the problem this paper is about untouched.

5. Conclusion

5.1 What the model tells us

The preceding sections have argued, in general terms, that merchant generation of either kind, gas or renewable, is becoming structurally difficult to finance as penetration rises. The plant economics model built alongside this paper, a full project-finance build for a 500MW merchant gas CCGT, priced directly against the same interaction model used in Section 4 rather than a separate assumption, converts that argument into a specific, falsifiable number, and the result is sharper than the qualitative case alone would suggest.

Priced as an existing asset already in operation, with no fresh construction financing and no project debt, the plant clears a positive equity NPV of €25.6m against the 2025 baseline price path. But equity IRR on that same cash flow is −4.8%, against a 6.0% cost of equity. The two figures are not in tension, they describe different things. The plant is worth marginally more than nothing to continue holding: whatever capital built it is sunk, and the residual cash flow it still generates has a small positive present value. But as a return on capital, it destroys value at every point across its operating life. Cash flow stays positive only through 2049, roughly two-thirds of the plant's modelled life, before degrading efficiency and rising fixed costs push it into sustained annual losses for the remainder of its tenor. This is the stranded-asset trajectory described qualitatively in Section 2.6, now expressed as a real number: not broken, structurally uneconomic on a forward-looking return basis, while remaining rational to keep running because the alternative, decommissioning a sunk asset, destroys more value than holding it.

Priced instead as new construction, financed with the phased capital draws, capitalised interest, and 60% leverage a real project would actually carry, the same plant is decisively unbankable. Equity NPV is −€482.9m. Minimum DSCR is −0.12x against a 1.20x covenant, not a marginal breach but a structural one: the project generates negative cash available for debt service in its worst year, and equity IRR is undefined because the cash flow never turns net positive across the full tenor. Debt service is a bullet in practice rather than an amortising schedule, none of the thirty pre-balloon years carry any scheduled principal, with the entire €327.7m obligation deferred to a single repayment at maturity the project's own cash flow gives no indication of being able to make.

The sensitivity analysis run against the operating case is the third result, and it is the one that most directly challenges the intuitive framing of this risk. Six levers were tested independently: fixed operating cost, plant efficiency, annual efficiency degradation, a gas price shock, a shift in how often gas sets the price, and leverage. Efficiency and degradation move the outcome further than any other lever, a two-point swing in nameplate efficiency moves cumulative cash to equity by more than €250m, and a faster degradation path alone is enough to erase the entire positive base case. Fixed operating cost matters almost as much: a €5,000/MW/year increase, well within the range of realistic maintenance cost inflation, cuts the base case by two-thirds. Gas price and gas-share shocks, the variables most commentary on this market treats as the central risk, move the outcome by comparatively less, and in the direction opposite to what a reader trained on Section 2's stranded-asset argument might expect: a higher gas price improves this specific plant's position, because it is more fuel-efficient than the reference marginal plant setting the market price in gas hours, so a rising gas cost widens its margin relative to the plant it is competing against. Leverage, tested only in the construction case since it is structurally irrelevant to an already-operating, debt-free asset, does not rescue the outcome at any tested level from 40% to 80%: DSCR stays broken across the entire range. This is not a capital-structure problem that better financing terms can solve. It is a cash-generation problem, and the model's own sensitivity output says so more precisely than the qualitative argument in Sections 2 and 3 could establish alone.

5.2 Who is positioned and who isn't

The asymmetry between the two financing cases sorts market participants cleanly, and the sort is not the one most coverage of this market assumes.

Owners of existing, already-depreciated merchant gas capacity are better positioned than the stranded-asset framing in Section 2 might suggest in isolation. Sunk construction cost is doing real work in the operating case's positive NPV: the same asset, financed today from scratch, fails outright. Integrated utilities holding generation alongside retail supply have a further partial offset the standalone model does not capture, a lower wholesale price that erodes generation margin simultaneously improves retail margin on the other side of the book. Developers and utilities with substantial CfD or long-dated PPA coverage, Ørsted, RWE, Vattenfall among the larger European players, are insulated from the cannibalisation dynamic on covered volumes for the contract term, with their remaining exposure concentrated in cost overrun and construction risk rather than the revenue risk this model quantifies.

Exposed positions cluster on the new-build and uncontracted side of the same asset classes. New merchant gas capacity, priced exactly as this model prices it, does not clear a financeable return under the price path this paper has established as the central case, not a downside scenario. Gas plant equity holding merchant tail exposure beyond an existing contract term faces the same arithmetic once that contract expires. LNG regasification infrastructure financed on the 2022–2023 throughput assumptions described in Section 2.7 carries a related but distinct exposure, capacity utilisation, not generation margin, but the underlying cause, demand assumptions calibrated to a supply crisis rather than the structural path this model implies, is the same. Merchant renewable developers without CfD or PPA coverage face the DSCR cliff quantified directly in Section 3.6. Equity holders in listed developers whose disclosed NAV assumptions embed a captured-price trajectory above what this paper's interaction model implies are carrying a valuation gap the market has not yet been forced to close.

5.3 What this means in practice

The practical value of this analysis is not a single market call but a reordering of what actually deserves underwriting attention. The sensitivity results in 5.1 invert the risk hierarchy most financing and investment discussions currently apply to merchant generation. Gas price and gas-share exposure, the variables that dominate commentary on this market and the ones a term sheet is most likely to negotiate covenants around, move this project's outcome by less than either its own operating efficiency or its degradation profile. A lender or equity investor spending diligence effort stress-testing commodity price paths while treating plant-level operating assumptions as fixed inputs is underwriting the wrong risk. The efficiency and degradation figures a technical advisor signs off on, not the gas price deck a trading desk provides, are what should decide whether a specific merchant gas asset clears its cost of capital.

The asymmetry in 5.1 between the operating and construction cases carries a second, more structural implication for how new capacity gets financed at all. An asset that clears a positive, if unremarkable, valuation once it already exists and is decisively unbankable to build from the same starting price path is not a financing problem a different capital structure solves, the leverage sensitivity in 5.1 already rules that out. It is a statement that new merchant generation, gas or renewable, needs contracted revenue in place before construction financing is sought, not as a risk-mitigation preference lenders might waive for a strong sponsor, but as the condition that determines whether the deal exists at all. Every merchant developer or gas plant sponsor currently modelling a construction-phase project against a spot-price forecast should ask whether that forecast, run through a model structured the way this one is, still produces a bankable DSCR, and should treat a negative answer as informative rather than as a reason to adjust the discount rate until it isn't.

Both the interaction model and the plant economics model built for this paper are available to test directly, rather than taken on the authority of the numbers printed here. A reader with a different view on renewable buildout, gas supply, or a specific plant's own cost assumptions can substitute those assumptions and see where the conclusion changes, and where it does not. That the leverage and gas-price sensitivities move the outcome by less than a reader might expect, while efficiency and degradation move it by more, is a claim worth verifying against a reader's own priors rather than accepting on the strength of this paper's presentation of it.

5.4 Limitations and further research

Three limitations are worth stating plainly rather than leaving implicit. The interaction model in Section 4 is calibrated to Germany, chosen for data richness rather than representativeness, and the Zakeri et al. dispersion cited in Section 4.1, gas setting the price in over 80% of hours in some markets against a far lower share in others, means a French, Spanish, or British reader should expect this paper's specific numbers to shift, even where the underlying mechanism holds. The plant economics model's operating case treats the plant as already built, with construction financing either fully phased and debt-financed or entirely absent, rather than modelling a real secondary-market acquisition transaction, a simplification that understates the transaction costs a real buyer of an existing asset would face without changing the underlying cash flow conclusion. Neither model endogenises gas plant retirement: as merchant economics deteriorate, capacity exit should tighten residual supply and partially offset the price compression this paper documents, meaning the compound effect quantified in Section 4.3 may be modestly overstated at the highest penetration levels, and the pace of demand-side flexibility deployment, storage, electrified heat, hydrogen, that could floor negative-price hours faster than assumed here is genuinely uncertain and not modelled at all.

Each of these is a direction for further work rather than a reason to discount the central finding: country-level disaggregation for France, Spain, and the UK; a closed-loop version of the interaction model incorporating retirement; and empirical validation against realised Spanish and German data, where the cannibalisation signal Section 3 documents is already detectable in live market data rather than model output.

5.5 Closing statement

The European electricity market is undergoing a structural transition that existing pricing mechanisms, and most of the financial models built around them, have not caught up to. The 2023 reform addressed the distribution of windfall revenue in a high-gas-price environment; it left the marginal pricing mechanism, and the compounding of direct cannibalisation with indirect TTF suppression, entirely unchanged. This paper's own project-finance model shows what that means in practice for a real asset rather than a stylised one: a plant that is worth holding once it already exists, and cannot be financed if it does not. As penetration continues toward the levels this paper has modelled, contracted revenue, CfDs, PPAs, long-term offtake, stops being a preference and becomes the only structure new merchant infrastructure can be built on, and the valuation gap between assets that have that protection and assets that do not will keep widening.

References

Academic

1. Hirth, L. (2013). "The market value of variable renewables: The effect of solar wind power variability on their relative price." Energy Economics, 38, 218–236.
2. Đukan, M., Keles, D., & Kitzing, L. (2025). "The Impact of Two-Sided Contracts for Difference on Debt Sizing for Offshore Wind Farms." The Energy Journal.
3. Joskow, P.L. (2011). "Comparing the costs of intermittent and dispatchable electricity generating technologies." American Economic Review, 101(3), 238–241.

Institutional and regulatory

4. European Union. Regulation (EU) 2024/1747 and Directive (EU) 2024/1711.
5. ENTSO-E Transparency Platform. Hourly load data, 2024–2025. transparency.entsoe.eu
6. RTE. (2024). Annual Electricity Review 2023.

Industry and market data

7. IGU. (2026). World LNG Report 2026.
8. IEA. (2025). Electricity 2025; IEA. (2026). Electricity 2026; IEA. (2025). Electricity Mid-Year Update 2025.
9. Aurora Energy Research. (2026). RESMOR European Renewable Market Outlook, January 2026.
10. Veyt. (2024, 2025, 2026). European PPA Market Review.
11. LevelTen Energy. (2025). European PPA Price Index, Q3 and Q4 2025.
12. S&P Global / Platts. (2025). (Various market data publications.)
13. FfE. (2025). "European day-ahead electricity prices in 2024." ffe.de
14. ICE / TradingView. TTF Natural Gas forward curve, July 2026 – December 2030. Data as of 17 July 2026.
15. ICE Endex / Barchart. EUA Futures Dec-26 through Dec-32 end-of-day settlements. Data as of 17 July 2026. Beyond Dec-2032: real-flat at Dec-32 × (1.02)^(year−2032).
16. Zakeri, B., Staffell, I., Dodds, P.E., Grubb, M., Ekins, P., Jääskeläinen, J., Cross, S., Helin, K., & Castagneto Gissey, G. (2023). "The role of natural gas in setting electricity prices in Europe." Energy Reports, 10, 2778–2792.
17. European Commission, Directorate-General for Economic and Financial Affairs. (2023). The Development of Renewable Energy in the Electricity Market. European Economy Discussion Paper 187.