Unlocking the Spanish Grid
A Diagnostic Report on Spain’s Electricity Distribution Crisis
Version 19.0 | March 2026 | Author: IB
All data sourced from: REE, CNMC, BOE, DSO capacity maps, and Ley 24/2013 as amended through 2025.
Executive Summary
In Spain, 88% of new grid connection applications are rejected, blocking an estimated €30 billion in stalled investment (UNEF, the Spanish solar industry association, 2025). Grid capital is flowing abroad — Iberdrola directed €5 billion to US and UK networks in 2025, its chairman citing Spain’s regulated returns as “a clearly negative signal.” The government has committed €13.6 billion in grid investment for 2025–2030. But is new infrastructure effective against a grid that is full on paper but half-empty on the cables?
In the last five years, Spain added over 35 GW of renewable generation capacity. The measured system peak reached 40.07 GW in January 2025 — still below the 44.8 GW record set in 2007. Yet 83–85% of substations are reported as unable to accept new connections. At transmission level, REE identifies 78 thermally constrained nodes out of ~700 — roughly 10% physical congestion on a network reported as 75% saturated. No DSO has published thermal loading data for any distribution substation.
To understand why, it is necessary to understand what fills the DSO capacity register — the administrative ledger that determines which connections are possible and which are blocked. The register records every existing connection at the maximum power it is permitted to draw, every permit granted to projects not yet built, and every application under review. A substation is declared “full” when the sum of these entries reaches the rated capacity of its cables and transformers. The critical point: the register is filled by administrative entries, not by measured physical load. A substation can be 85% full on paper while its cables carry 40% of their thermal capacity. This gap — between administrative saturation and physical utilisation — is the allocation trap.
The DSO capacity register is filled by three distinct populations:
- Population A — Operating connections: measured consumption exists. Before 2025, DSOs counted these at maximum contracted or permitted capacity, at coincidence factor 1. The CNMC’s 2025 reform mandates actual hourly measurement data for this population. This is real and correct.
- Population B — Valid permits, not yet connected: no metering history. Counted at full permitted capacity with simultaneity coefficients. A 10 MW unbuilt permit occupies the register for up to five years. The reform does not reduce this population’s footprint.
- Population C — The replenishment flow: Spain’s FCFS system is binary — no formal queue exists. But at saturated nodes, new applications arrive continuously, each blocked by the existing Population B stock. Rejected applications reappear at adjacent nodes or reapply after minor modifications. Population C is not a separate stock of capacity in the register; it is the mechanism by which Population B self-replenishes. The reform does not address this dynamic.
When REE published the first transmission demand capacity maps in February 2026 under the new methodology, Spain’s DSOs collectively claimed over 90 GW of reserved capacity at transmission-distribution interfaces. A hostile reader will note that the sum of regional maxima exceeds national coincident peak by definition — regional peaks do not occur simultaneously. Even allowing a generous non-coincidence factor that would push the 40 GW system peak up to a theoretical sum-of-regional-peaks of approximately 50–55 GW, the 90 GW DSO claim still exceeds that ceiling by 65–80%. The harder evidence is independent: REE could only agree with 45% of interface values. The CNMC was forced to delay full map publication by three months. This combination — a magnitude that cannot be reconciled with non-coincidence alone, and a TSO unwilling to validate the majority of submitted figures — strongly indicates that Populations B and C dominate the register.
The verdict: Spain’s grid congestion is systematically amplified — and at many nodes dominated — by administrative allocation. Two instruments would address this directly. Neither requires new primary legislation if the regulatory path is used correctly. Implementation details — transitional provisions, audit procedures, compensation rate-setting — can be addressed via CNMC binding circulars under existing authority. Legislative clarification of CNMC’s authority to enforce guarantee forfeiture would strengthen enforceability but is not a precondition for the instruments to operate:
- Mandatory permit expiry with real teeth: utilisation threshold below 50%, timeline below five years, no renewal at the same node, guarantees forfeited not returned. RD 997/2025 exists but is too weak on all four dimensions.
- Mandatory flexible access as default: converting Circular 1/2024’s optional flexible connections into the required access modality for new connections at saturated nodes. Current voluntary adoption is below 10% — not a policy, a gesture. Full Article 32 transposition delivers this.
Infrastructure investment is necessary at genuinely thermally constrained nodes. It is not sufficient — and without allocation reform, new capacity fills on paper within one application cycle. Worse: administrative over-allocation actively hides where the real thermal constraints are, making it impossible to direct investment correctly until the paper-full problem is resolved first.
Chapter 1: Players, Institutions, and Legal Framework.
Chapter 2: Grid Infrastructure and Fees — five voltage levels, regional saturation, tariffs, firm-capacity criterion, permit hoarding.
Chapter 3: Queue Backlog — FCFS mechanism, 88% rejection rate, 78 critical nodes, permit pipeline.
Chapter 4: Policy Responses evaluated against Populations A, B, and C.
Chapter 5: Verdict, solutions, and testable prediction.
Introduction
Spain closed 2025 with 142.5 GW of installed generation capacity — 7.3% more than 2024 — after adding 10 GW of renewable power (8.8 GW solar PV, 1.2 GW wind). Of total installed capacity, 68.9% is renewable.
Solar PV is the largest technology by installed capacity at approximately 48 GW (33% of the installed mix), having grown from near zero in 2015 to Europe’s second-largest solar fleet. Wind follows at approximately 33 GW (22% of installed capacity). Despite having less installed capacity than solar, wind leads generation output — contributing 21.6% of total electricity produced in 2025 against solar PV’s 18.4%. The reason is capacity factor: wind turbines in Spain run at approximately 28–30% of nameplate on an annual basis, while solar panels average 18–22%. Nuclear contributes 19% of generation from a small but highly utilised fleet.
Spain also has approximately 17 GW of conventional hydroelectric capacity — rivers, reservoirs, and run-of-river plants built largely in the 1960s and 1970s, their capital costs fully amortised decades ago. On top of this sits 3.3 GW of pumped storage hydropower across 18 plants, including La Muela on the Júcar river (1,482 MW — the largest pumped hydro plant in Europe). Pumped storage contributed 9,213 GWh to the system in 2025. These assets represent almost free storage at the margin: the civil infrastructure — dams, tunnels, turbine halls — was paid for fifty years ago. Operating costs are low; the only variable cost is the electricity used to pump water uphill during cheap overnight or midday solar hours. When the April 2025 blackout struck, it was Iberdrola’s pumped storage plants — La Muela, Aldeadávila II, Villarino, and Puente Bibey — that provided the synchronous inertia and fast-response generation used to restore the grid. Spain has underinvested in expanding this asset class: the PNIEC 2030 target of 22.5 GW total storage requires approximately 4 GW of new pumped capacity, against a current base of 3.3 GW.
Spain and Portugal operate as an “energy island.” The Iberian Peninsula maintains only 2% interconnection capacity relative to installed generation — far below the EU targets of 10% by 2025 and 15% by 2030. A subsea cable across the Bay of Biscay, expected operational by 2027, will double exchange capacity to 5 GW. Until then, surplus generation in southern Spain cannot easily reach European demand centres, driving the solar evacuation congestion documented in this report.
Spain’s wholesale electricity market operates through OMIE within the joint Iberian market (MIBEL), following standard European marginal pricing at ~€62/MWh (2024 average). The policy target is 81% renewable electricity by 2030, requiring rapid grid expansion and storage deployment.
The number that frames this report: Spain’s peak electricity demand is approximately 40 GW against 142.5 GW of installed capacity. Spain does not have a generation shortage. It has a delivery problem and, as the chapters below document, predominantly an administrative one.
On April 28, 2025, a peninsula-wide blackout left 47 million people without power across Spain and Portugal. The cause was a frequency stability failure, not grid congestion or an overloaded cable. But the political response treated it as proof that the grid was insufficient — triggering €13.6 billion in investment commitments, emergency legislation, and the policy debate this report examines.
Chapter 1: Players, Institutions, and Legal Framework
This chapter maps the institutional landscape: who operates what, who regulates whom, and what legal framework constrains the system. Understanding the actors and their incentives is essential because Spain’s grid crisis is institutional before it is technical.
1.1 The Institutional Map
1.1.1 Ministry for Ecological Transition
The government (Ministry for Ecological Transition) sets system charges (cargos), approves grid development plans, and can legislate via Royal Decree-Law (RDL) in urgent circumstances — subject to Congressional ratification within 30 days. RDL 7/2025 (June 2025), the emergency decree following the blackout, was derogated when Congress refused ratification after 28 days.
1.1.2 CNMC
The regulatory authority is the CNMC (Comisión Nacional de los Mercados y la Competencia), Spain’s independent markets and competition regulator. CNMC sets access tolls (peajes), approves DSO remuneration methodology, and issues binding circulars on capacity allocation, tariff design, and grid access conditions. It operates independently of the government but within the legislative framework set by Ley 24/2013. Critically, CNMC can reform methodology and require transparency but cannot compel DSOs to grant connections where its own capacity maps show headroom — that would require legislative change.
1.1.3 TSO: Red Eléctrica de España (REE)
Red Eléctrica de España (REE) operates 46,155 km of transmission lines at 400 kV, 220 kV, and 132 kV. REE serves as both transmission system operator (TSO) and system operator (SO): it owns the infrastructure, manages national system balance, and operates the grid second-by-second for frequency stability (50 Hz) and voltage support. REE is majority state-controlled through Redeia, with the government holding approximately 20% directly.
1.1.4 Five DSOs control ~90% of MV/LV distribution
- e-distribución (Endesa/Enel): South and Central Spain
- i-DE (Iberdrola): East and North (Madrid, Mediterranean coast)
- UFD (Naturgy): Scattered areas
- EDP: North and West
- Plus ~330 small municipal or cooperative DSOs
DSOs are regulated utilities. Revenue comes from access tolls (peajes) set by CNMC and system charges (cargos) set by the government. The same corporate groups — though legally separated — are often both grid operator and power supplier. E-distribución = Endesa = Enel (Italian); i-DE = Iberdrola (Spanish, 70%+ free-float, BlackRock as top shareholder); UFD = Naturgy (Spanish, CriteriaCaixa + GIP).
1.1.5 Legal Framework: Ley 24/2013
Spain’s electricity sector is governed by Ley 24/2013, del Sector Eléctrico, enacted December 2013 and amended extensively through December 2025. Key articles:
- Article 6: defines actors — including, since November 2025, storage operators, independent aggregators, and energy communities.
- Article 7: universal access to electricity supply — conditioned on capacity availability as determined by the grid operator.
- Article 12: first-come-first-served (FCFS) allocation. No queue position, no wait-list, no tariff mechanism to manage excess demand. Rejection is final for that application at that node.
- Article 14: Sustainability Principle — any regulatory action increasing system costs must be offset by equivalent revenue. Introduced after Spain’s €26 billion tariff deficit crisis. Constrains policy responses that increase costs, but not all proposed responses do so equally. Mandatory permit expiry imposes no new system cost — it removes Population B from the register at zero DSO cost. Mandatory flexible access with compensation does increase costs (compensation paid to curtailed users), which Article 14 would require offsetting via peaje adjustments or efficiency savings elsewhere. The distinction matters: Article 14 is a genuine legal constraint on flexible access compensation design, but it is not a barrier to permit expiry reform.
- Article 4bis (Ley 9/2025, November 2025): demand position planning mechanism allowing REE to reassign transmission capacity between generation and demand at specific nodes.
1.2 Consequences of the Legal System
Spain’s grid congestion is produced by a specific administrative choice. Ley 24/2013 does not require DSOs to distinguish between the grid connection rating and contracted capacity when determining available grid capacity. DSOs use the connection rating — the maximum the hardware can deliver — instead of contracted capacity, the maximum the user can draw. In Spain, a company can hold a grid connection twice the size of its capacity contract, at zero cost.
Contracted capacity is the real operational ceiling. A factory with a 10 MW grid connection rating and a 5 MW capacity contract will never load the cable beyond 5 MW, but the DSO registers 10 MW. The register fills with connection ratings; the cables do not fill with electrons.
1.2.1 The CNMC’s 2025 Reform: What Changed
To understand what the reform corrects — and what it leaves untouched — it is necessary to separate three distinct layers in every grid connection.
Layer 1 — Physical connection rating: the hardware maximum the cable, transformer, and meter installation can deliver. A factory with a 10 MW physical connection can draw up to 10 MW without damaging the infrastructure. Before the CNMC reform, DSOs used this layer as the default register entry, counting each connection at its full physical rating with coincidence factor 1.0. Ten factories each with 10 MW physical ratings occupied 100 MW in the register regardless of how much they actually drew.
Layer 2 — Contracted capacity: what the user has formally agreed to pay for and may draw. This is the legal entitlement — the billing ceiling. A factory might have a 10 MW physical connection but only a 6 MW capacity contract. The contracted capacity is what appears on the invoice and determines the maximeter billing band.
Layer 3 — Actual measured usage: what the cable physically carries. The same factory may consistently peak at 5 MW despite a 6 MW contract and a 10 MW physical connection.
The CNMC reform (BOE-A-2025-12396, June 2025) changes how DSOs assess capacity by introducing two distinct corrections. First, it removes the Layer 1 to Layer 2 gap from the register — the difference between physical connection rating and contracted capacity. This is a firm and permanent correction: the user has no contractual right to draw more than their contracted capacity, so the DSO correctly stops reserving that headroom. In the example above, the 4 MW gap between the 10 MW physical rating and the 6 MW contract is permanently removed from the register. No future action by the user can reclaim it without renegotiating their contract.
Second, the reform introduces measured consumption data (the most unfavourable hour from a five-year representative period) to assess the Layer 2 to Layer 3 gap — the difference between contracted capacity and actual usage. Simultaneity coefficients are applied: 0.7 for industrial connections, 0.85 at the MV feeder, 0.95 at the AT/MT transformer. This produces a soft and statistical correction: the factory consuming 5 MW against a 6 MW contract is now assessed closer to 5 MW × 0.7 = 3.5 MW. But the user retains the contractual right to draw up to 6 MW whenever they choose. If the factory expands production and starts drawing 5.8 MW, the DSO has no mechanism to prevent it — the soft relief erodes at the next five-year measurement cycle. Spain has no equivalent of the Dutch CBC (Congestie Beheer Contract) that would convert this soft statistical reduction into a hard contractual obligation with compensation.
Third, DSOs must now justify any denial with specific data — the demand and generation situation in the most unfavourable hour, growth assumptions used, and technical criteria not met. The era of rejecting applications with “no capacity available” without supporting evidence is formally over.
1.2.2 What the Reform Does Not Change — and the Evidence
The reform is real and its direction is correct. But it addresses only one of the three populations that fill the DSO capacity register.
Population A — Operating connections. Facilities already connected and drawing power, with a metering history. The reform changes how these are counted: actual hourly measured consumption replaces contracted or permitted maximum, with simultaneity coefficients applied at each network level. This is the reform’s genuine contribution.
Population B — Valid permits, not yet connected. Projects that obtained access rights but have not built their installations. There is no metering history. The new methodology counts these at full permitted capacity with simultaneity coefficients applied — but against the permitted value, not a measurement. A 10 MW unbuilt permit still occupies its full permitted value in the node register for up to five years. The reform cannot reduce this population’s footprint because there is nothing to measure.
Population C — The replenishment flow. Population C is not a separate stock in the DSO register. Spain’s FCFS system is binary — no formal queue exists. But at saturated nodes, new applications arrive continuously, each blocked by the existing Population B stock. Rejected applications reappear immediately at adjacent nodes or reapply after minor modifications. Population C is the mechanism by which Population B self-replenishes. The reform does not address this dynamic.
The practical consequence is visible in the first demand capacity maps published by REE in February 2026. When forced under the new methodology to declare their reference values at transmission-distribution boundary points, Spain’s DSOs collectively submitted figures totalling more than 90 GW. A 90 GW sum-of-interface-values cannot be directly compared against the 40 GW national coincident peak, because regional peaks do not occur simultaneously. Applying a generous non-coincidence factor of 0.7 to a national peak of 40 GW produces a theoretical regional sum-of-maxima of approximately 50–57 GW. The 90 GW DSO claim exceeds that ceiling by roughly 60–80%. REE — applying its own load flow analysis rather than relying on the comparison to coincident peak — could only agree with 45% of interface values. The CNMC was forced to delay full map publication by three months.
This evidence — a magnitude that cannot be reconciled with non-coincidence adjustments alone, plus a TSO that refuses to validate the majority of submitted figures — demonstrates that Populations B and C dominate the register. Correcting how operating connections are counted does not move the total significantly when unbuilt and queued permits dwarf measured consumption.
A worked example at a single node — before and after the CNMC reform
Before the reform: Ten industrial users each hold a 10 MW physical connection rating (Layer 1) with a 6 MW capacity contract (Layer 2) and actual consumption averaging 5 MW (Layer 3). The DSO registers all ten at their full 10 MW physical rating with coincidence factor 1.0. Register occupation: 100 MW. The node is paper-full.
After the reform: The reform removes the Layer 1 to Layer 2 gap firmly — the 4 MW difference between physical rating and contracted capacity per user is no longer reservable. The remaining contracted capacity (6 MW per user) is assessed using measured consumption data at a 0.7 industrial coincidence factor: 10 × 6 MW × 0.7 = 42 MW. The node now shows 42% register occupation. Approximately 58 MW appears freed — partly firm (the Layer 1 to Layer 2 correction, permanently removed) and partly soft (the Layer 2 to Layer 3 reduction, which erodes if users expand within their contracts).
Within one or two application cycles: A solar developer with a 50 MW AC nameplate connection application and a data centre developer requesting 8 MW both apply at the node under FCFS. Both are within the newly visible headroom. Both are granted. Register occupation returns to approximately 100 MW — now occupied by a different mix of measured Population A connections and new unbuilt Population B permits. The node is paper-full again. No physical infrastructure has changed.
This is not a hypothetical failure mode — it is the mechanical consequence of correcting Population A without simultaneously constraining Population B intake. The firm Layer 1 to Layer 2 correction is real and durable. But the freed space is immediately recaptured by new Population B applications at full nameplate. The reform improves the accuracy of the register; it does not prevent the register from refilling.
The reform therefore has a hard structural limit. For Population A it substitutes measured reality for assumed maximum — a genuine improvement. For Populations B and C it changes nothing. Spain’s capacity maps will continue to show widespread saturation as long as unbuilt permits occupy the register at full permitted value. The tools to address that are evaluated in Chapter 4.
1.3 Who Benefits from the Status Quo
The allocation trap persists because every major actor in the Spanish electricity system benefits from the current arrangement. The CNMC’s 2025 reform corrects how Population A connections are measured — a genuine improvement. But the incentive structures described below predate the reform, were not addressed by it, and continue to govern how every newly freed register space is allocated. The reform changed the measurement; it did not change the incentives.
Large Industrial Users
Industrial users holding existing grid connections benefit in three ways. First, the connection permit costs nothing to hold and the maximeter dampens the cost of over-contracting, preserving the option value of future expansion. Second, in saturated zones, the grid connection becomes a competitive moat — new entrants cannot replicate the production capacity of an incumbent who holds a scarce grid permit. Third, the FCFS mechanism means incumbents face no pressure to release unused capacity.
Distribution System Operators
Spain’s DSOs earn regulated returns under the RAB model established in CNMC Circular 6/2019. The return (5.58% current period, proposed 6.46–6.58% for 2026–2031) is earned on assets in service, regardless of utilisation rate. A substation at 20% physical utilisation generates the same regulated return as one at 80%. DSOs have no financial incentive to maximise utilisation of existing infrastructure. Their revenue grows by expanding the asset base — not by connecting more users to existing capacity. The “saturation crisis” narrative, while genuine in thermally constrained locations, also serves the DSOs’ investment case.
Investment spending is capped at 0.13% of GDP for distribution. This creates a convenient equilibrium: the DSO cannot over-invest (capped), does not need to optimise utilisation (no incentive), and can justify future investment by pointing to reported saturation.
Vertically Integrated Utilities
Spain formally unbundled distribution from generation and retail under Ley 24/2013. In practice, the same corporate groups control all three activities. Endesa (Enel) owns e-distribución (DSO), Endesa Energía (retail), and generation assets. Iberdrola owns i-DE (DSO), Iberdrola Clientes (retail), and generation assets. Naturgy owns UFD (DSO) and retail operations.
Vertical integration creates aligned incentives across the value chain without requiring coordination: the DSO arm earns regulated returns regardless of utilisation; the retail arm benefits from reduced competition in its territory; the generation arm profits from curtailment compensation and higher wholesale prices during supply shortfalls. Each entity independently benefits from the status quo.
The Political Dimension
The government attempted to break the allocation trap with RDL 7/2025 (June 24, 2025), introducing use-it-or-lose-it provisions for demand permits — automatic expiry if the holder did not contract at least 50% of allocated capacity within five years. Congress derogated it on July 22, 2025 — 28 days after enactment.
The government recovered some provisions through RD 997/2025 (November 2025), which does not require Congressional ratification. RD 997/2025 reintroduces permit expiry for connections above 1 kV and the 50% utilisation threshold. But the five-year clock means earliest expiries in 2030, the threshold is too low, and CNMC has not yet established enforcement procedures.
The political lesson: even after the worst power system failure in Spanish history, Congress could not sustain reform requiring industrial permit holders to use or lose their grid allocations. The constituencies that benefit from the allocation trap are politically stronger than those harmed by it.
1.4 Connection Rights vs. Connection Duty
Spain’s Ley 24/2013 guarantees universal access to electricity supply (Article 7). But this right is conditioned on capacity availability as determined by the grid operator — and the grid operator’s determination is based on the allocation methodology described above.
In some European systems, DSOs have an affirmative connection obligation (aansluitplicht in Dutch law): they must connect any requesting party within a defined timeframe, and if capacity is insufficient, they must invest to create it. In Spain, the DSO can deny access based on its own capacity assessment. Under Article 12, access is granted on technical criteria only — meaning the DSO’s calculation of available capacity is the binding constraint.
The CNMC’s reforms introduce transparency (published maps), measurement-based forecasting, and justification requirements. These are necessary steps. But they do not create an obligation to connect. A DSO that publishes a capacity map showing 40% headroom is not, under current law, required to proactively offer that capacity to waiting applicants. The maps inform; they do not compel.
The distinction between transparency and obligation is where Spain’s regulatory reform reaches its current limit.
The common thread: decision power delegated to self-interested parties
Across every layer of Spain’s grid access framework, the decisive choice is made by the party with the greatest private incentive to optimise against the public interest. DSOs self-declare their capacity — and benefit from showing saturation. Developers self-select their connection size — and benefit from maximising permit value. Permit holders self-determine their utilisation timeline — and benefit from holding capacity as a competitive moat or tradeable asset. Industrial incumbents self-report their consumption — and benefited until 2025 from the gap between physical rating and actual usage being invisible to the regulator.
Each of these behaviours is individually rational within the framework Spain created. Collectively they produce the allocation trap. This is not a market failure in the conventional sense — prices are functioning, contracts are honoured, rules are followed. It is a regulatory architecture failure: a framework designed to trust market participants, in a sector where private and public interests systematically diverge.
The solutions proposed in Chapter 5 work precisely because they replace self-declaration with external enforcement: permit expiry removes the developer’s discretion over whether to hold or build; mandatory flexible access removes the DSO’s discretion over whether to offer conditional connections; the peak connection cap for solar removes the developer’s discretion over connection size. Each instrument shifts a decision from the self-interested participant to the regulator. That is the architectural correction the reform of 2025 did not make.
Chapter 2: Grid Infrastructure and Fees
The April 2025 blackout — a frequency stability failure, not a congestion event — triggered the €13.6 billion investment commitment this chapter examines. This matters analytically, not just historically. A frequency stability failure is caused by insufficient reactive power and inertia in the generation mix — by too much inverter-based solar and wind displacing synchronous generators. It is not caused by cable congestion or substation saturation. The correct response is investment in grid-forming storage, synchronous condensers, and interconnection — not in distribution feeder expansion. Yet the political response allocated the majority of the €13.6 billion to transmission and distribution infrastructure justified by the saturation figures this chapter documents. Understanding why those figures cannot be trusted to direct investment correctly is the analytical task of Chapters 1 through 4.
Spain reports 85.7% distribution saturation nationally. Eight provinces report 100%. This chapter describes the physical grid — how it is built, who operates it, where congestion is concentrated, and what is driving it. All saturation figures are administrative: they measure allocated capacity as a percentage of rated capacity, as reported by DSOs and REE. This does not measure physical thermal loading.
2.1 Grid Architecture: Five Voltage Levels
Level 1: Transmission (HV, ≥145 kV) — REE operated
The transmission network is meshed — multiple paths exist between any two points. Congestion can be rerouted via alternative paths, reactive power can be adjusted, or generation can be redispatched. This redundancy makes transmission congestion operationally manageable — expensive but solvable without building new lines. Saturation: ~75% nationally. Specific nodes are at 100% and require competitive demand access auctions (Chapter 3).
Level 2: HV-to-MV Transformation — REE operated above 36 kV
Large power transformers step voltage down from 400/220 kV (transmission) to 66/45/36 kV (upper distribution). REE owns assets above 36 kV; DSOs own assets below. This boundary — where REE’s transmission assets connect to DSO distribution assets — is where coordination disputes stalled the CNMC’s capacity map publication in early 2026.
Generation-focused zones face their first bottleneck here. A solar farm in Andalusia must inject power through the local HV-MV transformer. If saturated, the farm cannot inject even if nationwide grid frequency is stable. Transformer replacement costs ~€5M per large unit with 18-month lead times.
Level 3: MV Primary (below 36 kV) — DSO Regional Feeders
Most primary MV feeders are radial — a single trunk from the transformer station, with branches. If the trunk saturates, there is no alternate path. Unlike meshed transmission, MV radial congestion cannot be rerouted. It requires physical infrastructure. Administrative saturation: 85.7% nationally (50–100% regional variation).
Level 4: MV Secondary (10–30 kV) — Local Feeders
This is where residential demand growth and distributed solar impact the grid most directly. Traditional distribution was designed for unidirectional flow (generation → transmission → distribution → consumption). Rooftop solar reverses the flow. A neighbourhood with high solar penetration and low daytime demand creates reverse power flow back through the MV secondary transformer. Saturation here involves both thermal limits and voltage stability.
Level 5: LV Distribution (<1 kV) — Household
Spain now has ~1.3 million residential solar systems injecting power into LV networks. Heat pumps and EV chargers draw from LV. The saturation that blocks household connections is not caused by household over-contracting — it is caused by industrial and generation over-allocation at MV and HV levels that fills the capacity register above.
| Level | Voltage | Operator | Network | Key Constraint | Saturation (2026) |
|---|---|---|---|---|---|
| Transmission | ≥145 kV | REE | Meshed | Node bottlenecks | ~75% |
| HV-MV Transform | 400/220 → 66/45/36 kV | REE (>36 kV) / DSOs (<36 kV) | Point assets | Generation overflow | High in renewable zones |
| MV Primary | Below 36 kV | DSOs | Radial | Feeder saturation = local dead end | 85.7% avg |
| MV Secondary | 10–30 kV | DSOs | Radial | Reverse flows; voltage limits | Rising |
| LV Household | <1 kV | DSOs | Radial | Transformer capacity | Rising |
2.2 Saturation by Region
National saturation (85.7%) masks vast geographic variation.
Fully saturated (95–100%): Basque Country (99–100%): all three provinces fully saturated — high industrial demand (~3 GW peak), limited France interconnection. Navarra (~99%): excellent wind resources (~2.5 GW), local demand only ~1.2 GW, export corridors congested. La Rioja (~99%): ~5 GW generation, ~0.8 GW demand. Eight provinces at 100%: Guipúzcoa, Vizcaya, Álava (Basque), Guadalajara, Albacete (Castilla-La Mancha), Teruel (Aragón), Soria (Castilla-León), Jaén (Andalusia).
Severely saturated (85–95%): Andalusia (85–95%): ~10–12 GW solar, local demand ~3.5 GW. Aragón (85–92%): wind zones (Teruel) at 95%+. Castilla-La Mancha (85–90%): ~8 GW solar, ~1.5 GW demand.
Moderately saturated (75–85%): Madrid (75–85%): demand-driven — peak demand has grown sharply since 2015, driven by data centres, electrification, and population growth. Valencia (80–85%): coastal electrification. Barcelona/Catalonia (75–85%): mixed industrial and renewable.
Lower saturation (50–75%): Galicia (60–75%): ~2 GW wind, well-integrated. Extremadura (65–75%). Canary and Balearic Islands (75–85%): insular grids with seasonal tourism peaks.
2.3 Drivers of Congestion
Driver 1: Renewable Generation Mismatch
Spain added 8.85 GW of generation capacity in 2025 (7.9 GW solar PV). Solar capacity stands at 48 GW — the largest single generation source. Renewable generation is concentrated where resources are best (southern and central Spain), not where demand is highest (Madrid, Barcelona, Mediterranean coast). Andalusia generates ~10–12 GW solar against ~3.5 GW local demand; Castilla-La Mancha generates ~8 GW against ~1.5 GW demand. This surplus must flow north through limited transmission corridors or be curtailed. In 2024, 1.4% of renewable generation was curtailed nationally; in saturated zones, 5–8%.
Driver 2: Electrification Load Growth
Heat pumps: ~500,000 installed (2024), drawing 2–10 kW during heating season. EVs: ~1.2 million (2024), uncontrolled home charging at 7 kW concentrated at 6–7 PM. Data centres: Madrid and Barcelona emerging as European hubs — a single facility consumes 50–200 MW continuously, and Madrid’s peak demand has grown sharply over the last decade as a result.
Driver 3: Infrastructure Investment Lag
DSO distribution capex is ~€400–600M/year across Spain. Permitting takes 2–4 years for major lines. Transformer lead times are 12–18 months. Historical execution delays of 2–3 years relative to plan are the norm.
2.4 How Spain Bills for Grid Capacity
Households (2.0TD, up to 15 kW) pay capacity charges based on contracted power (potencia contratada). Holding excess contracted capacity costs money — creating right-sizing pressure. Households are not the source of the allocation trap.
Industrial and commercial users (3.0TD, 6.1TD, 6.2TD, 6.3TD, 6.4TD) bill via the maximeter — the highest 15-minute average demand recorded in each of six billing periods. The billing rules create an asymmetric incentive:
- Below 85%: billed at 85% of contracted — a floor creating some cost for over-contracting.
- 85–105%: billed on the maximeter reading — paying only for actual peak demand.
- Above 105%: maximeter reading plus double the excess — a steep penalty for undersizing.
6.1TD tariff rates (2026), CNMC Resolución RAP/DE/009/25 (hour counts approximate; exact distribution varies year by year with calendar configuration of Type A/B/B1/C/D days):
| Period | Typical Hours (approx.) | €/kW/yr | Physical Grid Condition |
|---|---|---|---|
| P1 | Peak winter weekday mornings (~220 hrs) | 23.95 | Genuinely heavily loaded at specific nodes |
| P2 | Weekday shoulders (~500 hrs) | 12.69 | High loading; some physical stress |
| P3 | Weekday mid-afternoon (~1,000 hrs) | 5.94 | Moderate loading; mostly headroom |
| P4 | Weekday evenings (~1,000 hrs) | 3.80 | Moderate loading; mostly headroom |
| P5 | Late nights (~1,250 hrs) | 1.42 | Low loading; substantial headroom |
| P6 | Nights, weekends, holidays (~4,790 hrs) | 0.06 | Lightly loaded; cables far below limits |
The P1:P6 spread (380× for 6.1TD) incentivises demand shifting. But the critical feature for the allocation trap is the billing asymmetry: the penalty for contracting too little is far steeper than the cost of contracting too much.
The critical distinction: Contracted power has a cost floor (85% rule). The grid connection rating does not. A factory can have a 10 MW connection rating, contract 6 MW, peak at 5 MW, and pay only on maximeter readings. The 4 MW gap between connection rating and actual use occupies the DSO’s register at zero cost.
| Layer | Value | What DSO register records | Annual cost to holder |
|---|---|---|---|
| Grid connection rating | 20 MW | 20 MW — fills node on paper | Zero |
| Contracted power | 10 MW | Not visible to the register | 85% floor if maximeter < 8.5 MW |
| Actual peak (maximeter) | 7 MW | Not visible to the register | Actual usage charges |
2.5 The Firm Capacity Criterion: Overstatement by Design
Spanish DSOs have historically assigned capacity under a criterion of absolute firmness. Each user is assumed to consume their full allocated capacity for every hour of the year (8,760 hours), simultaneously with every other user. The standard European coincidence factor for industrial MV connections is 50–70% (IEC 60439, CIGRÉ TB 566). Spain’s methodology assumes 100% — overstating actual coincident demand by a factor of approximately 1.4–2×.
The consequence: a substation with a physical thermal rating of 100 MW, serving users whose actual coincident peak is 50–70 MW, shows as “100% saturated” because allocated connection ratings sum to 100 MW. The 30–50 MW of physical headroom is invisible to the allocation system.
Based on REE’s published load duration data, at the specific nodes that are genuinely thermally constrained — the 78 confirmed transmission nodes and equivalent distribution hotspots — the network approaches its physical capacity ceiling during approximately 800–900 hours per year, concentrated in P1 and upper P2. This is a local figure at those specific nodes, not a national average; the majority of Spain’s network runs well below thermal limits during the same hours. The remaining ~7,900 hours even at constrained nodes, cables run well below thermal limits. See Section 3.7 for the full thermal constraint analysis.
The generation-side nameplate problem
Spain applies one capacity reduction at the application stage for solar PV: under RD 1183/2020, the access permit is granted at the lesser of the DC panel capacity and the AC inverter continuous rating. A 100 MWp DC solar field with a 75 MW AC inverter receives a 75 MW permit. This DC/AC ratio typically runs 1.2–1.4× in Spain, producing a 15–25% reduction from panel nameplate.
This reduction is real but has a critical limitation: it is bypassable. The developer chooses the inverter size, not the DSO. Installing a 100 MW inverter on a 100 MW DC field eliminates the reduction entirely. In a saturated grid where permit capacity is worth €80,000/MW or more, the cost of a larger inverter is trivially justified by the larger permit obtained. Spain therefore has no effective nameplate reduction mechanism for solar generation permits — only a voluntary one.
The deeper problem is what the inverter cap does not address: the relationship between the permitted connection capacity and the actual output profile. Spain’s utility-scale solar is now almost entirely deployed on horizontal single-axis trackers (HSAT) — panels rotating from east to west through the day. By 2024, approximately 99% of new utility-scale solar installations used single-axis tracking, as the combination of tracking and bifacial modules delivers the lowest levelised cost of electricity. A tracked 100 MW AC plant produces more than 80 MW for only approximately 200–300 hours per year — roughly 2–3% of annual hours. For the remaining 97–98% of the year, output is well below the permitted connection capacity.
The Netherlands applies a hard peak connection cap reflecting this reality: a 100 MW PV park receives a firm connection of 50–70 MW. The remaining 30–50 MW is structurally curtailed — always, by design, as a condition of the connection agreement. The developer accepts this upfront. The economic logic is sound: the top 2–3% of peak output hours represent a small fraction of annual energy; sizing the feeder and substation to serve those hours wastes infrastructure capacity for 97% of the year. Spain applies no equivalent cap. A 100 MW tracked solar farm with a 100 MW AC inverter holds 100 MW in the DSO register despite producing above 80 MW for only 2% of annual hours.
The simultaneity coefficient applied to multiple demand connections (0.7 for industrial, 0.85 at MV feeder, 0.95 at transformer) correctly addresses the coincidence across multiple users. For a single large solar park — which is the dominant utility-scale configuration in Spain — the coincidence factor with itself is 1.0 by definition. The simultaneity coefficient is the wrong instrument for the single-plant case. The correct instrument is the peak connection cap, applied at the point of permit grant.
Chapter 3: Queue Backlog
3.1 The FCFS Allocation Mechanism
Spain allocates grid access on a first-come-first-served basis under Article 12 of Ley 24/2013. There is no formal queue. Applications are evaluated as they arrive: if capacity exists at the requested node, the permit is granted; if not, the application is rejected. There is no waiting list, no priority order, no allocation by social cost or system value. Rejection is final for that application at that node.
88% of applicants are removed from the system before any economic analysis occurs. Better tariff design, capacity markets, and flexibility mechanisms are irrelevant to projects that never pass the access gate. Spain’s most powerful policy tools operate on the 12% of projects that survive FCFS — the rest are invisible to the policy toolkit.
3.2 The Denial Rate: 88% and Rising
| Project Type | Requests (GW/yr) | Approval Rate | Primary Denial Reason |
|---|---|---|---|
| Solar PV | 18–22 | 10–15% | MV saturation in southern generation corridors |
| Wind | 6–8 | 12–18% | Regional transmission bottlenecks |
| Battery storage (BESS) | 8–10 | 5–10% | No firm grid access for charge/discharge cycles |
| Industrial/commercial | 6–8 | 15–25% | Geographic concentration in saturated urban zones |
| Distributed (HP, EV) | 2–3 | 40–60% | Some rural capacity; urban constrained |
| National weighted total | ~40 | ~12% | — |
Source: REE and DSO data, 2025. During 2025, 40 GW of access and connection requests were submitted to the network. Only 4.5 GW obtained permits; 25 GW were rejected for lack of capacity; 8.5 GW remain under review.
The denial rate correlates directly with DSO territory saturation. i-DE (Iberdrola, Madrid and Mediterranean coast) reports 88–92% denial — the highest among the Big Five. e-distribución (Endesa, Andalusia and Aragón) reports 85–90%. EDP (Galicia, Asturias) reports 60–70%, reflecting lower saturation in the northwest.
3.3 Transmission: 78 Critical Nodes
At the transmission level, REE has identified 78 nodes — out of approximately 700 total transmission nodes, roughly 11% — where FCFS capacity is exhausted and competitive demand access auctions are required. The remaining ~89% of transmission nodes are administratively saturated, not thermally constrained. These are concentrated geographically: 19 in Andalusia (solar overflow), 11 in Castile-León, 10 in Aragón (wind), 9 in Castilla-La Mancha (solar), and 7 each in Madrid (demand-driven) and Extremadura.
The Tagus-Júcar Bottleneck
The Tagus-Júcar corridor is the most critical transmission constraint — the primary 400 kV and 220 kV route moving power from Andalusia’s solar generation north to Madrid’s demand. During peak solar hours (noon to 15:00), the corridor operates at 75–95% of thermal capacity. Southern generation of 8–12 GW meets local Andalusian demand of 3.5 GW, leaving 4.5–8.5 GW requiring northward export through a corridor rated at approximately 6 GW. The result is structural curtailment during midday peaks — this is a genuine thermal constraint, not a paper-full phenomenon. Reinforcement of the Tagus-Júcar corridor is planned for 2027–2029.
3.4 The Permit Pipeline
The aggregate permit pipeline at the transmission network level (REE, February 2026):
| Category | Volume | Status |
|---|---|---|
| Renewable gen. (wind + solar) operational | ~81 GW | Operational (48 GW solar + 33 GW wind, end-2025) |
| Renewable gen. permits granted (transmission) | 129 GW | Total permits — includes operational |
| Unbuilt renewable gen. permits (est.) | ~48 GW | 129 GW total minus ~81 GW operational |
| Storage permits (transmission) | 16 GW | Permitted, largely not built (96 MW operational as of Feb 2026) |
| Demand permits since 2022 | 11.8 GW | Granted, none operational |
| Total demand permits (transmission) | 19 GW | All demand permits on REE books |
Source: REE capacity access publication, February 20, 2026. Note: the 129 GW of wind/solar permits is confirmed by REE. The operational/unbuilt split is estimated from REE 2025 installed capacity data (48 GW solar + 33 GW wind = ~81 GW) against the total permit stock.
The permit bubble context: Between 2018 and 2021, 130 GW of access rights were granted — three times total system capacity at the time. The majority was generation. This 130 GW bubble created the paper-full condition that now occupies most of Spain’s grid allocation register. Milestone extension regimes (RDL 29/2021, RDL 8/2023, RD 997/2025) have repeatedly deferred expiry rather than enforcing it.
3.5 Why Permits Don’t Expire
The anti-speculation milestone regime (RDL 23/2020) designed to reclaim permits from non-compliant developers has been extended rather than enforced:
- RDL 29/2021: added 9 months to milestones
- RDL 8/2023: allowed further extensions
- RDL 7/2025: attempted to extend fifth milestone — derogated in 28 days
- RD 997/2025: introduced 50% utilisation threshold over five years for connections above 1 kV. First expiries no earlier than November 2030.
The pattern: each time a major milestone cohort approaches its deadline, the government extends the timeline rather than enforcing expiry. Developers facing non-compliance are organised, politically connected, and have deposited financial guarantees they lobby hard to recover.
3.6 The SPV Market: Consolidation, Not Release
Grid connection permits in Spain are held by legal entities — typically Special Purpose Vehicles (SPVs). When permit expiry approaches, the rational response for a speculative holder is not to surrender the permit but to sell the SPV. The permit transfers with the company.
Financially, ready-to-build assets trade at €80,000/MW or above, with the grid connection permit as the dominant value component. Operational assets reached €1.1 million/MW at peak in early 2024. The financial guarantees raised specifically to deter speculation were simply absorbed into SPV sale prices. When guarantees quadrupled, resale prices rose. The speculation continued.
SPV sales convert speculative Population B holders into operators who will actually build — a private efficiency gain. But from the DSO register’s perspective, the permit at the node is held before the sale and after it. The node remains paper-full. The SPV market is a secondary market for Population B positions, not a release mechanism. One important nuance: if mandatory expiry is enforced, developers will rationally accelerate SPV sales to ready-to-build operators before expiry deadlines arrive. This is a gaming response but ultimately self-correcting — a ready-to-build operator who acquires an SPV specifically to avoid expiry will actually build, converting Population B permits into operating connections (Population A). The permit bubble shrinks even if developers game the transition, because the gamers are forced to become builders.
3.7 The Residual Thermal Constraint
Stripping away the paper-full phenomenon does not eliminate all congestion. At specific thermally constrained nodes — the Tagus-Júcar corridor, certain urban distribution feeders, the 78 confirmed transmission nodes — the network approaches its physical capacity limits during approximately 800–900 hours per year. This is a local figure, not a national one. During those same hours, the majority of Spain’s network runs well below thermal limits. The 800–900 hour constraint at a congested solar evacuation feeder in Andalucía is not representative of a national grid approaching saturation — it is evidence of localised thermal stress at specific nodes in a network that is predominantly paper-full rather than physically loaded. Specific feeders face thermal stress for substantially longer: a solar evacuation feeder in Andalucía may be at thermal limits for 1,500+ midday hours; an urban distribution feeder in central Madrid serving data centres may be thermally stressed for 2,000+ hours.
A new industrial customer seeking a firm 24/7 connection cannot be served by a feeder that is thermally full for even 876 hours — unless new infrastructure is built or the customer accepts flexible access that curtails during constrained hours.
The solution set after allocation reform is layered: reclaim the paper-full headroom first, then manage the residual thermal constraint through infrastructure at genuinely constrained nodes and flexible connections for the thermal peak hours. The allocation trap must be addressed first — without it, the thermal constraint cannot even be accurately measured.
A critical asymmetry between transmission and distribution deserves explicit acknowledgement. At transmission level, REE’s 78 confirmed thermally constrained nodes are based on actual load flow modelling with independent measurement capability — that identification is credible. At distribution level, no DSO has published thermal loading data for any substation. The capacity maps show administrative saturation; they do not show physical cable loading. Spain does not know where its distribution thermal constraints are. This is not a minor data gap — it means the €4.4 billion distribution investment component of the national plan cannot be correctly targeted until allocation reform clears the paper-full noise from the register.
Chapter 4: Policy Responses — Evaluated Against Populations A, B, and C
Spain’s congestion is not a single problem — it is a tripartite allocation trap. As established in Section 1.2: Population A (operating connections with measured consumption), Population B (valid unbuilt permits occupying the register at full permitted value), and Population C (the replenishment flow — the mechanism by which Population B self-replenishes through continuous new applications).
The CNMC’s 2025 methodology reform (BOE-A-2025-12396) addresses only Population A. The February 2026 capacity maps confirm that Populations B and C dominate the register (see Section 1.2.2). No policy currently in force meaningfully reduces the footprint of B or C.
This chapter evaluates each instrument against all three populations.
4.1 CNMC Methodology Reform (BOE-A-2025-12396)
What it does: Replaces firm-capacity assumptions with actual hourly measurement data for Population A. Applies simultaneity coefficients (0.7 for industrial, 0.85 at MV feeder, 0.95 at transformer) to measured consumption. Requires DSOs to justify denials with specific data.
| Population | Effectiveness | Reason |
|---|---|---|
| A | High (firm Layer 1) / Moderate (soft Layer 2–3) | Firmly removes the physical rating to contracted capacity gap — permanent and non-reclaimable. Softly reduces the contracted to actual usage gap via measured consumption and simultaneity coefficients — but this relief erodes if users expand within their contracted capacity, and the DSO has no mechanism to prevent it. |
| B | Marginal | Unbuilt permits are counted at full permitted value — there is no measured data to substitute. Simultaneity coefficients (0.85 feeder, 0.95 transformer) are applied at aggregation levels above the connection point, providing marginal relief. But individual permits remain at full nameplate at the connection point level. A single 50 MW unbuilt solar permit still occupies 50 MW at the node. The aggregate coefficient helps fractionally; it does not address the fundamental problem. |
| C | None | Queue priority applications block capacity prospectively. Reform does not address this. |
Status: Implementing. First capacity maps delayed by REE due to implausible DSO claims (see Section 1.2.2 for the magnitude analysis).
Verdict: Necessary for Population A. Does nothing for B or C — the dominant populations. Strongly supports the diagnosis: paper-full congestion is structural, not measurement-based.
4.2 €13.6 Billion Infrastructure Investment (2025–2030)
What it does: €9.2B for transmission (REE), €4.4B for distribution (DSOs). Targets 9,500 km of new/reinforced 400/220 kV lines and MV/LV upgrades. Execution delayed 2–3 years historically; full delivery may extend to 2032–2033.
| Population | Effectiveness | Reason |
|---|---|---|
| A | High | Directly addresses physical bottlenecks at operating nodes (e.g., Tagus-Júcar corridor). |
| B | Transient | New transmission capacity genuinely creates headroom that some Population B permits can access — a new 400 kV line feeding a saturated region enables MV connections that were previously blocked. But new headroom is allocated via FCFS at distribution level. First-comers request maximum connection ratings at nameplate. New capacity fills on paper within one or two application cycles. Infrastructure helps Population B transiently; without allocation reform, the register saturates again at a higher physical ceiling. |
| C | Low | Queue priority applications block new capacity before it is even built. |
DSO Incentive Conflict: DSOs earn regulated return on assets in service, regardless of utilisation — returns are tied to the RAB, not to effective utilisation. Investment grows the RAB, aligning with DSO incentives but not with public need for utilisation.
Verdict: Necessary for Population A at genuinely thermally constrained nodes. Ineffective for B and C without allocation reform. The investment plan and the allocation framework are structurally misaligned — expanding capacity without correcting allocation rules amplifies paper congestion.
There is a deeper problem. The €13.6 billion plan is being allocated against capacity maps that are unreliable for distinguishing thermal from administrative saturation. Those maps show 83–85% administrative saturation — but they cannot distinguish between a substation that is genuinely thermally loaded 2,000 hours per year and one that is 30% physically loaded but has permits summing to 95% of rated capacity. Both look identical in the DSO register. Both trigger the same investment justification. Without solving the paper-full problem first, Spain cannot know where the real thermal constraint resides. The substations that genuinely need new iron — the Tagus-Júcar corridor, Madrid data centre feeders, Basque industrial zones — compete for the same distribution budget as hundreds of substations that need a permit audit, not a new cable. A significant and unquantifiable fraction of the €4.4 billion distribution investment is being directed by noise. The fraction cannot be quantified precisely — and that inability is itself significant: the same maps that prevent correct diagnosis of where investment is needed are the maps guiding the investment plan. The pragmatic response is not to pause the entire plan but to sequence it: continue transmission investment at nodes where thermal constraints are confirmed independently of the capacity maps (Tagus-Júcar corridor, urban demand feeders serving data centres and EV charging), while deferring distribution feeder expansion at nodes that are administratively saturated until permit expiry has cleared sufficient Population B stock to make a reliable thermal assessment possible.
4.3 P1–P6 Tariff Structure + Flexible Connections (Circular 1/2024 + Article 32 EU)
What it does: The P1–P6 tariff creates extreme time-of-use differentiation (P1 €23.95/kW/yr vs P6 €0.06/kW/yr — a 380× ratio for 6.1TD), incentivising industrial users to shift consumption from peak to off-peak hours. Circular 1/2024 introduced conditional (flexible) access — connections granted with curtailment obligations during P1 and upper P2 hours, in exchange for access that would otherwise be denied. Article 32 of EU Electricity Directive 2019/944 requires Spain to transpose a framework enabling DSOs to procure flexibility services from distributed resources. Transposition remains partial as of early 2026.
| Population | Effectiveness | Reason |
|---|---|---|
| A | Moderate | P1–P6 reduces coincident peak at operating nodes by shifting load to off-peak. Reduces physical thermal stress. Does not change the DSO register. |
| B | Low | Unbuilt permits are not yet consuming. No demand to shift. |
| C | None | Queue-blocked applications have no connection and no consumption to shift. |
The combined mechanism — where it works: P1–P6 pricing is the necessary incentive; flexible connections are the necessary access instrument. Together they create a viable path for new connections in saturated zones: the applicant accepts curtailment during the ~800–900 hours per year when cables are genuinely thermally stressed, in exchange for firm access during the remaining ~7,900 hours.
Why adoption is below 10%: Project financiers require revenue certainty. Curtailment during P1 hours — the hours with highest electricity prices — creates revenue risk that flexible access tariff savings do not offset. The maximeter already bills on actual consumption, so a contractual curtailment obligation does not reduce the capacity bill. There is no financial upside to accepting curtailment. Furthermore, accepting flexible access forfeits the option on future expansion in a grid where obtaining additional capacity is nearly impossible. DSOs have no incentive to offer flexible connections proactively — conditional access complicates operations without increasing the RAB.
What Article 32 transposition would change: Full Article 32 transposition would require DSOs to make flexible access the default for new connections in saturated zones, and to establish compensation mechanisms for curtailed users. This converts a voluntary instrument with below-10% adoption into a mandatory framework. The European Commission can pursue infringement proceedings against Spain for incomplete transposition — external pressure that bypasses domestic political blockage.
Verdict: P1–P6 is an effective demand-shifting signal but does not change the DSO register for any population. Flexible connections address Population A nodes that are thermally stressed — unlocking the ~7,900 hours of physical headroom. Ineffective for B and C. The gap between the mechanism’s design and its adoption is structural, not informational. Mandatory default for new connections in saturated zones is the missing step.
4.4 SPV Permit Market
What it does: Grid connection permits are held by Special Purpose Vehicles. When a permit holder cannot or will not build, the rational response is to sell the SPV rather than surrender the permit. Ready-to-build assets trade at €80,000/MW or above, with the grid connection permit as the dominant value component.
| Population | Effectiveness | Reason |
|---|---|---|
| A | None | Operates on existing connections, not permit transfers. |
| B | Partial | SPV sales convert speculative holders into operators who will actually build. Capacity moves from inactive to active within Population B — but stays in Population B. The DSO register does not change. No capacity is freed. |
| C | None | Queue applications are not yet permitted and cannot be sold. |
The critical limitation: SPV sales consolidate the permit bubble — they do not reduce it. A speculative developer selling an SPV to a ready-to-build operator is a private efficiency gain. From the DSO register’s perspective, the permit at the node is held before the sale and after it. The node remains paper-full.
Verdict: Economically rational private market that accelerates project delivery within Population B. Does not reduce Population B’s footprint in the DSO register. Cannot substitute for permit expiry enforcement.
4.5 Demand Access Auctions (BOE-A-2025-14863)
What it does: Competitive auctions for demand capacity at specific transmission nodes where FCFS capacity is exhausted. The first round awarded 928 MW across five nodes (Brazatortas 400 kV, Cristóbal Colón 220 kV, Francolí 220 kV, Nuevo Vigo 220 kV, Palos 220 kV) to industrial electrification and green hydrogen projects. REE has identified 75 nodes potentially eligible for future tender rounds.
| Population | Effectiveness | Reason |
|---|---|---|
| A | None | Auctions are for new demand, not existing connections. |
| B | Partial | Auction winners displace or absorb Population B positions at auctioned nodes — competitive selection replaces FCFS priority. At those specific nodes, permit quality improves. |
| C | Moderate | Auctions provide an alternative access path for projects currently blocked in Population C at the auctioned nodes. |
Scope limitation: 928 MW awarded in the first round against 40 GW of annual applications — roughly 2% coverage. The mechanism is transmission-only. REE’s 75-node pipeline, if fully tendered, could unlock 5–10 GW of demand capacity at transmission level. It does not address the 83.4% of distribution nodes showing administrative saturation.
Verdict: Well-designed for its scope — competitive allocation at genuinely congested transmission nodes is more efficient than FCFS. Useful for Population C at specific high-value nodes. Does not address distribution congestion or the permit bubble. A necessary complement to allocation reform, not a substitute for it.
If competitive auctions scale to REE’s full 75-node pipeline at 50–100 MW per node, annual capacity awarded could reach 3.75–7.5 GW — potentially making auctions the primary allocation mechanism at transmission level within three to five years, complementing FCFS at distribution level where auctions are less operationally practical.
4.6 Summary Assessment
| Policy Instrument | Population A | Population B | Population C | Timing |
|---|---|---|---|---|
| CNMC methodology reform | High (Layer 1 firm) / Moderate (Layer 2–3 soft) | None | None | Implementing |
| €13.6B infrastructure | High (thermal) | Low (re-hoarded) | Low | 2027–2033 |
| P1–P6 + flexible connections | Moderate | None | None | Active / <10% uptake |
| SPV market | None | Partial (consolidates, doesn’t reduce) | None | Active |
| Demand access auctions | None | Partial (specific nodes) | Moderate (specific nodes) | 2026+ |
| Gap: Populations B + C | — | No instrument reduces the permit bubble | No instrument clears the queue | — |
4.7 Why the Reforms Cancel Each Other Out
The pattern across the summary table is consistent and damning: every instrument currently in force addresses Population A or specific high-value transmission nodes. No instrument reduces the total footprint of Populations B and C in the DSO register.
The CNMC reform reduces Population A’s footprint. That freed space is immediately occupied by new Population B permit applications — because the same FCFS mechanism that created the bubble continues operating unchanged. A node that was paper-full on operating connections becomes paper-full on unbuilt permits within one application cycle. The net effect of the reform on total registered capacity is approximately zero. The bubble does not shrink — it rotates: operational over-allocation is corrected; speculative over-allocation fills the gap.
The SPV market accelerates rotation within Population B — permits move from speculative to operational holders faster — but the total volume of Population B permits at each node does not fall. Infrastructure investment adds new physical capacity that is immediately claimed by the same FCFS queue, restoring paper saturation at a higher physical ceiling.
The fundamental problem is that no instrument currently in force imposes a cost on holding unbuilt permits or creates a mechanism to convert blocked Population C applications into active connections. Until Population B’s volume falls below the level that produces paper-full conditions at distribution nodes, Population A corrections and thermal infrastructure investment operate on a register that will continue to show near-total saturation. Spain can build, measure, and optimise simultaneously — and still not move the 88% rejection rate if the permit bubble self-replenishes faster than it is corrected.
Chapter 5: Verdict and Solutions
5.1 The Verdict
Spain’s grid congestion is systematically amplified — and at many nodes dominated — by administrative allocation. The DSO capacity register is filled by three populations — operating connections counted at maximum assumed capacity, valid permits held by unbuilt projects, and queue-priority applications blocking capacity before a permit is even issued. The CNMC’s 2025 methodology reform correctly addresses the first population. It leaves the second and third untouched. The February 2026 capacity maps confirm that Populations B and C dominate the register (Section 1.2.2). Infrastructure investment is necessary at the specific nodes and corridors that are genuinely thermally constrained. It is not sufficient for the 83.4% of nodes that are administratively saturated (administrative saturation — nodes reporting no available capacity in the DSO register — is distinct from the ~11% of transmission nodes confirmed as thermally constrained by REE load flow analysis; both figures are accurate but measure different things). And without solving the paper-full problem first, the investment plan cannot even correctly identify which nodes genuinely need new iron — because the capacity maps guiding the €13.6 billion are contaminated by Populations B and C. Spain is investing at scale against a signal it cannot read.
The situation is urgent now, not in 2030. Eighty-eight percent of grid connection applications are rejected. An estimated €30 billion in industrial investment is stalled. Industrial electrification projects, green hydrogen developments, and data centre expansions cannot connect. Spain’s manufacturing base cannot expand its grid footprint. This is not a forecast — it is the current state of the system as documented by REE, CNMC, and the government’s own BOE preambles.
Who can move this — and who cannot
The political economy is unambiguous. Spain’s vertically integrated utilities — Iberdrola, Endesa, Naturgy — hold simultaneous positions as DSOs, generators, and retailers. They have lawyers, lobbyists, and congressional access. The 88% of blocked applicants — renewable developers, industrial electrification projects, green hydrogen developers, data centre operators — are scattered, uncoordinated, and have no equivalent political leverage in Madrid. The derogation of RDL 7/2025 in 28 days demonstrated this asymmetry precisely: the government proposed reform, the utilities organised against it through their congressional allies, and the reform died. No coalition of blocked developers will replicate that power.
There are exactly two realistic reform vectors. The first is CNMC acting within its existing regulatory authority — issuing binding methodology circulars, redefining how permits are assessed, changing the rules of the game rather than attempting to police DSO compliance through financial penalties that will be litigated for years. The instruments proposed in this chapter are deliberately designed for this path: permit expiry rules and peak connection limits for solar generation are methodology changes, not enforcement actions. They cannot be gamed once they define the measurement. The second vector is the European Commission initiating Article 32 infringement proceedings — bypassing Madrid entirely. Both paths operate outside the legislative process that the utilities have demonstrated they can block. Everything else — voluntary industry commitment, government white papers, stakeholder consultation — has been available for years and has produced the system documented in this report.
5.2 The Two Instruments That Would Work
Two instruments address Populations B and C directly. Both are within reach. Neither requires new primary legislation if the regulatory path is used correctly. They can be implemented simultaneously, but the practical sequencing matters: mandatory permit expiry should be activated first, because it reduces the Population B stock already in the register — freeing space that flexible access can then fill with productive new connections. Mandatory flexible access as default should follow within six to twelve months, once the expiry regime is operational and the register has begun to clear. Deploying flexible access into a still-bloated Population B register risks granting new flexible connections at nodes that are paper-full rather than thermally constrained, which wastes the instrument’s capacity to unlock genuinely available physical headroom.
Instrument 1 — Mandatory permit expiry with real enforcement
RD 997/2025 introduced permit expiry but got four parameters wrong. The correct version:
- Utilisation threshold: 20% of permitted capacity within 24 months — not 50% within five years. A project that cannot demonstrate 20% utilisation within two years is not a real project.
- No renewal at the same node: expiry must mean permanent loss of the specific node position. The current framework allows immediate re-application, which is not expiry — it is a reset.
- Guarantees forfeited, not returned: financial guarantees must be non-refundable on expiry. Under RD 997/2025, holders who formally renounce their permit recover their guarantee. This removes the financial cost of speculative holding entirely.
- Population C included: queue-priority applications must carry the same utilisation obligations from the date of priority grant, not from the date a permit is eventually issued.
These four changes convert a nominal expiry mechanism into one that actually reduces Population B and C footprints. Without all four, the permit bubble self-replenishes. A grid connection permit is a conditional allocation, not a property right. The condition implicit in every permit is demonstrated utilisation: a position in the grid queue is granted in exchange for a commitment to connect and use. A permit holder who cannot demonstrate utilisation within the prescribed period is not having property expropriated — they are failing to meet the condition that made the permit valid in the first place. This reframing is legally significant: the question is not “can the DSO take away property?” but “can the DSO enforce the utilisation condition that always defined the permit?” The answer under existing Spanish administrative law is yes, provided the condition is stated clearly in the permit terms. RD 997/2025 already establishes this principle; the problem is that the parameters are too weak to produce real enforcement.
Speculative permit holders who lose their positions under mandatory expiry are absorbing the risk they accepted when they obtained a permit with no obligation to build. That is not a policy problem — it is the point. The current system allows capacity to be reserved indefinitely at near-zero cost and traded as a financial asset at €80,000/MW or more. Ending the system means that arbitrage ends. Developers who held Population B permits speculatively should not expect compensation.
Legitimate project delays are a genuine concern and are addressed within the instrument design: the 24-month utilisation threshold and the interim permit mechanism at 50% of requested capacity both preserve queue position for projects that are genuinely under development. The line between speculative holding and legitimate delay is drawn by demonstrated project readiness — signed offtake contract, environmental permits, land acquisition — not by self-declaration. Implementation requires a litigation-safe design that ties forfeiture to objective utilisation measurements, and staged rollout beginning with new permits before applying retrospectively to existing ones.
Mandatory expiry addresses accumulated Population B permits. It does not prevent the register from refilling with new permits issued at the same inflated values. For expiry to durably reduce register footprint, it must be accompanied by a capacity reduction at the application stage for generation permits. Spain’s current inverter AC cap rule (RD 1183/2020) is a hardware definition, not a grid planning tool — it is bypassable by the developer choosing a 1:1 inverter-to-panel ratio, and it applies no simultaneity discount across multiple plants at the same node. What is needed is a simultaneity coefficient applied at the feeder and node level when assessing new generation permit applications — analogous to the 0.7 coefficient now applied to industrial demand connections under BOE-A-2025-12396. For utility-scale solar generation permits, the simultaneity coefficient is the wrong instrument — a single large PV park has a coincidence factor of 1.0 with itself by definition. The correct instrument is a peak connection cap: permits for utility-scale solar should be granted at the output level covering approximately 98% of actual annual generation hours — typically 50–70% of AC nameplate capacity — with the remainder structurally curtailed as a condition of the connection agreement. This reflects the physical reality that Spain’s tracked solar fleet exceeds 80% of AC nameplate output for only approximately 200–300 hours per year. Sizing the grid connection at 100% of nameplate to capture those hours wastes node capacity for 97–98% of the year. The Dutch DSO applies this peak cap in practice. Spain does not.
For multiple smaller distributed solar installations sharing a feeder — community solar, industrial rooftops — the simultaneity coefficient remains the appropriate instrument, as geographic spread and orientation differences do produce genuine coincidence effects across multiple plants. The 0.85 feeder and 0.95 transformer coefficients already applied under BOE-A-2025-12396 address this partially. This is within CNMC’s existing authority under Article 33(11) of Ley 24/2013. For demand permits, the equivalent measure is requiring demonstrated project readiness and staged financial guarantees before full-capacity permits are granted — preventing speculative applications from occupying the register at maximum requested values from the first day. Project readiness should be defined objectively: signed power purchase agreement or offtake contract, environmental permits obtained, and land acquisition completed. DSOs must verify these conditions before issuing the full permit; interim permits can be issued at 50% of requested capacity pending completion of all conditions. A 50% interim permit occupies 50% of the node’s capacity register — this is intentional. It reduces the speculative footprint by half compared to a full permit, halves the permit’s value as a tradeable SPV asset, and converts to full capacity only upon verified project readiness; it does not eliminate Population B pressure but meaningfully reduces it while preserving a legitimate project’s place in the queue. To prevent gaming, developers may hold only one interim permit per node; a second application at the same node is inadmissible until the first has converted to a full permit or been formally surrendered.
Instrument 2 — Mandatory flexible access as default at saturated nodes
Circular 1/2024 permits flexible connections. Article 32 of EU Electricity Directive 2019/944 requires Spain to establish a framework enabling DSOs to procure flexibility services. Current adoption is below 10% because the instrument is voluntary. Three changes make it work:
- Default, not optional: all new connections at nodes showing administrative saturation above 80% are granted as flexible access by default. Firm access becomes the exception requiring justification, not the rule.
- Separate register category: flexible connections must occupy a separate category in the DSO allocation register — not counted against the firm-capacity total. If counted as firm, the mechanism fails immediately. If genuinely separate, it is additive capacity that does not displace existing allocations. DSOs must report flexible and firm connections as separate register categories in their monthly capacity maps. CNMC publishes this data publicly. Miscounting becomes visible as a data inconsistency between reported flexible access grants and register category totals — creating transparency pressure rather than relying on financial penalties that DSOs would litigate for years.
- Compensation for curtailment: users accepting flexible access during P1 and upper P2 hours must receive compensation at the avoided cost rate. Without compensation, financiers will not accept the revenue risk. Belgium and France both apply sourcing-cost compensation for curtailed flexible connections. Spain has the regulatory authority to set this rate without new legislation. As a starting point, the avoided cost rate should be set at the real-time wholesale price minus the DSO’s avoided transmission cost — estimated at approximately €5–15/MWh based on recent Spanish wholesale market spreads, updated monthly to reflect market conditions.
Flexible access works for the majority of applicants currently blocked — manufacturing on shifts, logistics, commercial buildings, EV charging networks, most renewable generation. It does not work for 24/7 continuous-process industry and data centres requiring uninterrupted firm power. For those users, infrastructure investment at thermally constrained nodes remains the only answer. The two instruments together serve the full spectrum of blocked applicants.
What success looks like
The report does not make a confident forecast, but the order-of-magnitude implication of the diagnosis is worth stating. If mandatory permit expiry at the parameters proposed reduces Population B by 40–60% at the average distribution node — a conservative estimate given that the 2018–2021 bubble added permits equivalent to three times total system capacity at the time — the register load at those nodes falls to approximately 40–60% of rated capacity — well within the range where new firm connections can be granted. The 88% rejection rate is driven by near-total register saturation; nodes at 40–60% register load typically show acceptance rates above 50%. Combined with Population A measurement corrections already implementing, this suggests the weighted national rejection rate could fall toward 50–65% within the first expiry cycle. This is speculative — it depends on the rate of expiry enforcement, the pace of new Population B applications, and how quickly freed capacity is allocated. The monthly capacity maps will show the actual trajectory within twelve months of implementation. Treat this as a scenario, not a prediction.
5.3 Are There Other Solutions?
Two further categories of solution deserve acknowledgement. They are categorised as secondary not because they are less significant in their effects but because they address structural conditions that amplify the permit bubble problem rather than directly reducing existing Population B footprint. The two main instruments in Section 5.2 reduce what is already in the register. The solutions below shape what enters the register next, or change the underlying incentive that drives DSO behaviour. Both matter; the sequencing is logical.
Locational tariff differentiation. The CNMC has authority under Article 16 of Ley 24/2013 to apply different peaje rates at saturated versus unsaturated nodes. This would create a price signal steering new demand toward uncongested locations — reducing Population C pressure at the most contested nodes. Technically sound but complex to implement without creating perverse investment incentives. A medium-term complement, not an immediate fix.
DSO remuneration reform. Rewarding DSOs for connections enabled and capacity utilised — rather than assets in service — would remove the structural incentive to maximise booked capacity. The CNMC has authority to revise the remuneration methodology in the 2026–2031 regulatory period currently under consultation. The proposed rate increase (5.58% to 6.46%) is an opportunity to attach utilisation conditions. Correct long-term direction. Does not reduce the existing permit bubble and will take years to influence DSO behaviour.
Infrastructure and storage as complements, not substitutes
The €13.6 billion grid investment plan and battery storage deployment are both necessary — but they solve different problems and should not be confused with each other or with allocation reform.
Transmission cables solve the spatial mismatch: moving surplus solar generation from where it is produced (southern and central Spain) to where demand is concentrated (Madrid, Barcelona, Valencia). The Tagus-Júcar corridor reinforcement is correctly prioritised — it is a genuine thermal fix for a genuine thermal bottleneck. Per GW of transfer capacity, transmission infrastructure costs approximately €250–500M/GW with a 40–50 year asset life. It is the most cost-effective instrument for the spatial problem.
Battery storage solves the temporal mismatch: shifting midday surplus solar to the evening demand peak when prices recover. Co-located BESS behind an existing PV connection requires no new grid permit — it charges during curtailment hours at near-zero marginal cost and discharges into the evening price premium. At approximately €150–200/kWh all-in installed cost (falling rapidly), the economics are approaching viability without subsidy: the daily price spread between midday solar collapse and evening recovery reached an average of €94/MWh on peak days in 2025. Per GW of equivalent firm capacity, BESS costs approximately €800M–1B/GW with a 12–15 year effective life before replacement cycles begin. Cables are cheaper, more durable, and more structurally sound. BESS is a necessary complement for the temporal dimension that cables cannot solve.
Both require allocation reform to reach their full potential. The transmission corridor reinforcement delivers power to distribution nodes that are paper-full — without allocation reform, the new transmission capacity papers over into new Population B permits within one application cycle. Co-located BESS improves individual PV park economics but does not free any register capacity for the 88% of blocked applicants. The correct sequencing: implement allocation reform in parallel with infrastructure investment, deploy BESS as costs continue to fall.
The EU Lever: Article 32 Infringement Proceedings
Spain’s transposition of Article 32 of EU Electricity Directive 2019/944 — which requires member states to establish frameworks enabling DSOs to procure flexibility services from distributed resources — remains partial as of early 2026. This is not a minor omission. Article 32 is the legal foundation for mandatory flexible access. Without full transposition, Spain’s DSOs can continue treating flexible connections as optional.
The European Commission can initiate infringement proceedings against Spain for incomplete transposition without requiring Spanish Congressional ratification. This is the one lever that bypasses the domestic political blockage that killed RDL 7/2025 in 28 days. The Commission has used infringement proceedings against member states on energy directives before — including on unbundling, on access to transmission, and on renewable energy targets. The mechanism is established, available now, and underused in the Spanish context. The Commission typically issues a formal notice of non-compliance allowing two months for response, then proceeds to a reasoned opinion if Spain does not transpose; infringement cases on energy directives have typically reached the Court of Justice within 18–24 months, creating substantial political pressure for compliance before judgment is required.
Full Article 32 transposition, compelled by infringement pressure, would provide the legal basis for both mandatory flexible access as default and the compensation framework for curtailed users. It would not require Spain’s fragmented parliament to pass new legislation. Given the political dynamics documented in Section 1.3, this external lever may be more achievable in the near term than domestic reform.
5.4 Reading the Monthly Maps
Spain’s DSOs have been publishing monthly capacity maps since September 2025 for distribution and February 2026 for transmission. These maps are the primary observable data source for tracking whether allocation reform is working. But they must be read correctly — and the most important insight is that both of the two plausible outcomes confirm the paper-full diagnosis, not just one.
Outcome 1 — headroom is freed and then refilled: If nodes where Population A measurement reform has been fully implemented show initial capacity recovery followed by rapid refill from new Population B permit applications — observable as a surge in new applications exceeding the node’s historical six-month average by 30% or more within two months of corrections taking effect — the paper-full mechanism is confirmed directly. The register corrects and immediately self-replenishes. Infrastructure investment at those nodes would be wasted.
Outcome 2 — headroom appears sustained: If nodes show sustained apparent headroom after Population A corrections without visible Population B refill, this does not falsify the paper-full diagnosis. It confirms a different and equally important finding: the capacity maps are so contaminated by Population B noise that they cannot reliably distinguish between genuine thermal headroom and administratively freed space. A node showing 40% register occupation after Population A corrections may be genuinely available for new connections — or it may be occupied by Population B permits that have not yet triggered new applications. The maps cannot tell the difference. The €4.4 billion distribution investment plan is being allocated against this same contaminated signal. Both outcomes mean the investment is being directed blind.
The monthly maps should therefore be read not as a test of whether paper-full is real — that question is settled by the magnitude analysis in Section 1.2.2 — but as a diagnostic tool for understanding how the paper-full mechanism manifests at specific nodes and how quickly Population B refill operates in practice. This information is valuable for calibrating the urgency and sequencing of the two instruments proposed in Section 5.2.
A longer-horizon observation point exists: nodes where RD 997/2025 permit expiry is eventually enforced — earliest November 2030 — should show durable capacity recovery that persists across multiple application cycles rather than the transient correction from Population A reform alone. If expiry is enforced at the parameters proposed and Population B stock genuinely falls, the register at those nodes should stabilise at materially lower occupation levels. That is the meaningful test of whether the instruments work. This report should be updated when that data becomes available.
End of Report
Version 19.0 — March 2026
Data sources: REE system reports 2024 and 2025; REE demand capacity maps February 2026; CNMC Circular 1/2024; BOE-A-2025-12396 (distribution capacity specifications); BOE-A-2025-25253 (transmission capacity specifications); BOE-A-2025-14863 (demand access tender); RD 997/2025; Ley 24/2013 as amended through December 2025; CNMC Resolución RAP/DE/009/25 (tariff rates 2026); UNEF 2025 investment stalled estimate.