The European Solar Spine

An open letter to Prime Minister Rob Jetten and the cabinet.

We are late. Not too late.

Artificial intelligence runs on electricity. Whoever has the cheapest, most abundant power builds the models, keeps the engineers, and sets the terms. Everyone else rents. Europe is energy-short — not by accident, but by a generation of choices. We can spend the next decade blaming those choices, or we can build our way out of them. This brief is about building.

The coalition agreement got the ambition right and the instrument wrong. Locking in 40 GW of offshore wind through long-term Contracts for Difference — just as that market is failing across Europe — bets the house on the one clean technology whose costs are rising. There is a bolder move available, and a more characteristically Dutch one: don't just buy the electrons — build the industries that make them. This country has made that move before. The Netherlands still buys its chips — and every advanced chipmaker on Earth buys from ASML.

The vehicle is a European Solar Spine — a high-voltage corridor carrying cheap Iberian sun north. But the corridor is the means, not the point. The point is to build three sovereign industries at once: European solar manufacturing, a European HVDC grid and cable industry, and the clean generation behind them — with nuclear and small modular reactors as a co-equal pillar. The Netherlands does not need to host every factory. It needs to lead, finance, and anchor the demand — and let willing partners build the rest.

The stakes are not abstract. Arthur Mensch warned the French National Assembly that Europe has roughly two years to build its own AI infrastructure or accept becoming, in his words, a "vassal state." He is right. The one thing standing between Europe and that fate is electricity — and the will to produce, not merely consume.

We are late. Not too late.

The problem with betting the house on North Sea wind

Start with the technology the coalition wants to bet on. In October 2025 the Netherlands offered a 1 GW subsidy-free offshore-wind permit at Nederwiek I-A and received zero bids. The market had spoken, and the state's answer was to put the subsidies back: roughly €3.9bn is now earmarked for the IJmuiden Ver sites, and the 2026 round leans on public money again — the €948m drawn from the Climate Fund is only one slice of a support package worth around €2.5bn.

Be precise about how wide this goes. The clearest collapses are in Germany and Denmark, where subsidy-free tenders also drew no bids. The UK's January 2026 AR7 auction, by contrast, cleared a record ~8.4 GW — but only with a subsidy, which makes the same point from the other side: the volume shows up when the state pays, and vanishes when it doesn't. Belgium's auction has been repeatedly delayed. The pattern is consistent — across the North Sea, offshore wind no longer stands without public money.

Why does wind keep needing the subsidy? Because, unlike solar, it does not ride a falling cost curve. Solar is a manufactured semiconductor product; its price has fallen by roughly 99% since the 1970s and keeps dropping. An offshore wind farm is mostly not the turbine — the turbine is only about a third of capital cost. The rest is foundations, subsea cable, installation vessels, and decades of offshore maintenance: steel, ships, and salt water. No amount of AI makes a 200-metre crane on a North Sea vessel cheap. The US reference cost for fixed-bottom offshore wind is around $181/MWh; the Dutch IJmuiden Ver tender caps bids near €104/MWh — and that is a ceiling under a temporary subsidy, not a guaranteed strike price.

There is also an honest non-economic ledger, and we should state it plainly rather than overclaim: offshore turbines are visually intrusive even far from shore, they crowd busy shipping lanes, and construction noise and seabed works affect the marine environment — impacts that are real, partly mitigable, and worth weighing.

None of this means no wind. It means wind should be the winter hedge, not the whole bet. Lock the house to a rising-cost technology through 15-year contracts and you pay above-market rates well into the 2040s.

The real prize: three industries, not imported electrons

So if not 40 GW of wind, what? The instinct is to buy the cheap electrons — import Spanish sun and be done. That misses the prize. The real opportunity is not cheaper power; it is to build three industries at once, and to own them.

• European solar manufacturing — not assembling imported cells, but the hard upstream: polysilicon, ingots, wafers, cells.
• A European HVDC grid and cable industry — converters, cable, offshore installation. This is infrastructure that outlives any single energy source: the corridors that carry solar today carry nuclear, wind, and storage tomorrow.
• The generation itself — Iberian solar at the south end, and (below) nuclear and SMRs as a sovereign baseload industry.

You cannot import an industrial base. You can buy a shipment of panels; you cannot buy the capacity to make them, the engineers who run the lines, or the supply chain that compounds over decades. China understood this. It now controls the upstream solar chain — ~80% of polysilicon and ~95–98% of ingots and wafers — the steps where real sovereignty lives. Assembling modules in Europe while buying every wafer from one country is not sovereignty; it is a logo on a dependency.

A warning, honestly told. Cheap power alone does not rebuild an industry. NorSun ran a wafer plant in Norway on near-free hydropower and still closed it in December 2024, undercut by Chinese product sold below cost. The lesson is not "give up" — it is that building here takes three things together: cheap power, guaranteed demand, and trade defence. Miss any one and the factory dies. The Spine supplies the first; the rest of this brief supplies the other two.

And this is not foreign terrain for the Netherlands. The country that built the world's largest dredging fleets and the heaviest offshore-installation vessels already owns much of the engineering to lay this corridor and build at sea. We have done "build the industry, not just buy the product" before — with water, with wafers, with food. Solar, grid, and reactors are next in that line.

Two honest bets, never blurred

Every honest plan separates its safe bet from its bold one. This plan has both, and they must never be blurred — because a critic who conflates them can kill the whole thing by attacking only the riskiest piece.

The safe bet: importing the electrons. Building an HVDC corridor and buying Iberian solar is low-risk and incremental. Cross-border HVDC already works across Europe — NordLink, North Sea Link, Viking Link, and the France–Spain Biscay Gulf interconnector now under construction. This is proven engineering connecting allied EU members inside one market. It is not Desertec (more on that below). Even if nothing else in this brief happened, the corridor would stand on its own footing as ordinary interconnection — its returns resting on a contracted price floor rather than today's volatile spot market (see Appendix C).

The bold bet: rebuilding the industry. European-made wafers and cells, majority-EU ownership of the grid and cable supply chain, joint ventures on European terms — this is the high-risk, high-reward play. It can fail; NorSun did. It needs patient capital, guaranteed offtake, and trade defence to work.

Do both — and price each honestly. Fund the safe leg like infrastructure: long horizons, modest returns. Fund the bold leg like the strategic industrial bet it is: public risk capital alongside private money, eyes open to the failure rate. Pretending the bold leg is as safe as the corridor would be exactly the kind of overclaim that has sunk grand European projects before. Keeping the two distinct is what lets the safe leg proceed now while the bold leg is built deliberately.

How it stays legal — and Dutch-led

Two objections arrive together: isn't forcing technology transfer how China broke the rules? and won't this stall in Brussels for a decade? Both have answers.

China is the proof the strategy works, not the model to copy. By requiring joint ventures for market access, it built national champions and absorbed foreign technology — in autos, the US Trade Representative formally found forced technology transfer. (In high-speed rail the same playbook produced the champion CRRC, though — to be precise — there was no adjudicated WTO ruling that the rail JVs were illegal; that "violation" is analysts' judgement, not a decided case.) The point stands: the outcome is achievable. The method must be legal.

It can be — using rules Europe already wrote. Replace China's stick (a blunt local-content mandate) with a carrot: make a majority-EU-owned joint venture the only rational way to win something genuinely valuable. For a Dutch-led project that valuable thing is concrete and controllable — guaranteed long-term offtake from the Spine, IPCEI funding for gigafactories, and resilience-weighted procurement under the EU Net-Zero Industry Act, which lets auctions favour diversified supply where Europe is dangerously dependent on one country. With China at ~95% of the upstream PV chain, that condition is plainly met. The grid and cable industry is protected differently — through EU foreign-investment screening and the GATT security exception, a defensible basis for majority-EU ownership (lead with screening; the security clause is support, not a magic shield).

And it is Dutch-led, not an EU programme — which is the whole point on timing. The Netherlands convenes a coalition of the willing and uses Europe's instruments as tools, rather than waiting for twenty-seven governments to agree. Spain and France sell the sun; Germany and Poland join as manufacturers; the Netherlands finances and anchors the demand. Europe already has the champions to build the grid leg — Siemens Energy, Prysmian, Nexans (note: Hitachi Energy, often listed beside them, is Japanese-owned — a supplier, not a European champion). On PV, the rule is different. Don't chase commodity silicon — China has already won that race. Aim at the next generation instead: perovskite-on-silicon tandem cells, where Europe's labs still have real expertise. But be honest about the limit. That lead is not secure; China now holds the efficiency records. So treat tandems as a bonus, not the foundation of the plan.

The full legal toolkit — every instrument, threshold, and how it is triggered for the Spine — is in Appendix B. Use Europe's rules. Don't wait for Europe's permission.

Winter, and the case for nuclear in many countries

Solar has an obvious problem: it is dark at 6 p.m. on a January evening, exactly when Dutch homes turn up the heat. An honest plan answers that directly.

It answers with an and/and, not an either/or. South solar flowing north by day; nuclear baseload at night and in deep winter; offshore wind as the winter hedge (this is what wind is genuinely good for); and gas and storage firming the gaps, declining over time. The same HVDC corridor that carries Spanish sun by day can carry French nuclear after sunset — the cable does not switch off when the sun does. A solar-heavy system also needs stability, not just energy: inertia and grid-forming hardware (synchronous condensers, grid-forming inverters), funded inside the corridor's own capex — a residual risk we own squarely in Appendix F.

Nuclear here is not merely a hedge; it is the third sovereign industry. Not only France — several European countries should build and maintain reactors, including smaller, safer SMRs, as an industrial base in its own right: the same logic that drives solar and grid. Be realistic about time: a new large reactor in the Netherlands will not be running by 2035 — the late 2030s at the earliest, with SMRs later still. So nuclear's weight grows through the 2040s, and the scalable Spine carries the load until it arrives.

What does the mix look like? Sized to demand — the Netherlands is heading from about 109 TWh today toward roughly 200 TWh by the mid-2030s as transport, heat, industry, and AI all electrify.

A defensible central case: ~40 GW of domestic solar, ~12.5 GW of de-risked offshore wind, ~0.5–1 GW of nuclear ramping behind it (Borssele plus an early increment — the bulk of nuclear arrives in the 2040s), and a Solar Spine scaling from tens of gigawatts of Iberian solar upward — gas and storage firming the edges. It is a central case with an honest band, not a single hero number; the full derivation, source by source, is in Appendix A.

The jobs answer: what AI takes, engineering wins back

There is a fear under all of this, and it deserves a straight answer: if AI eats knowledge work, what are the jobs?

The answer is redirection, not despair. The work AI cannot offshore or automate wholesale is the work of building physical things — gigafactories, grids, reactors, cables, and the installation and maintenance of hard infrastructure. A re-industrialisation plan is, by construction, a re-employment plan. If we lose jobs to AI, we win them back by doing more engineering — and by fighting for the right to produce things in Europe.

This is the most Dutch argument in the brief. This is a country that turned back the sea and sold the world its dredging and marine-engineering industry (Boskalis, Van Oord, and the heaviest offshore-installation vessels on Earth); that turned chip-making into ASML, the sole maker of the machines behind every advanced processor; and that became the world's second-largest food exporter — €137.5bn in 2025 — from a small, cold delta, by building a high-tech agriculture and seed industry rather than simply farming harder. Nearly every European nation has a tradition like it. The plan is to rebuild and extend that tradition, not mourn it.

ASML is also the human face of the problem. Its planned new campus — up to 20,000 jobs around Eindhoven — ran headlong into grid congestion, a constraint serious enough that it took dedicated government action to keep the expansion on track. Europe's crown-jewel company, throttled by the same grid scarcity that throttles everyone else. That is the cost of standing still, made concrete.

The employment case is not hand-waving: the sourced job-creation ranges for a European PV, HVDC, and nuclear build-out — and the existing engineering clusters each draws on — are set out in Appendix E. The headline is simpler. The same plan that secures Europe's electrons rebuilds the high-value engineering base that gives the next generation something to do.

Who builds it, and who pays

Who builds it, who owns it, and who pays? This is where bold meets disciplined.

Roles. The Netherlands plays to its strength — not as host of every factory, but as financier and anchor, borrowing on its top-tier credit rating and guaranteeing the demand that makes the whole thing bankable. Spain and France host the solar and sell it north. Germany and Poland — and any willing others — take the manufacturing, and the jobs with it. No single country carries it alone; each joins where it is strongest.

The money — bold, and structured to stand on its own. The aim is a roughly 50/50 public-private vehicle that behaves like a business, not a subsidy. The anchor lender is the European Investment Bank, which already deploys around €100bn a year and put €11.6bn into grids and storage in 2025. The bulk equity is patient European capital — the big pension funds ABP and PFZW, EU-domiciled and built for thirty-year infrastructure. Invest-NL is the catalytic cornerstone: small (about €2.5bn), but first-loss money that crowds the larger investors in. Operators bring the knowledge — TenneT to run the corridor, RTE and Red Eléctrica at the ends, and TotalEnergies, a French EU-domiciled major (in effect "the EU's Shell"), on Iberian solar. Non-EU heavyweights with real firepower — Norway's ~$2.2tn fund, Shell, Equinor — are welcome as minority co-investors, but cannot hold the controlling stake: that must stay majority-European for the legal structure to hold.

Who benefits — and becomes an owner. The cheapest power on the continent is for the industries that need it most, and those industries should take equity as offtaker-shareholders, not merely customers: data centres and sovereign AI compute (Mistral the standard-bearer), Tata Steel's green-steel conversion, the chemical clusters, green-hydrogen electrolysers, and the PV gigafactories themselves. ASML and Mistral lend the project their weight as champions; the heavy industrial loads supply the guaranteed demand.

There is a partial answer here to the hardest question of all — can the Dutch grid even absorb this much power? The flexible loads (electrolysers, data centres, ingot furnaces) soak up surplus midday power and turn it into product instead of curtailment. Bring the demand to the power, not the power to the demand: AI training runs anywhere, on any schedule, so the compute goes where electrons would otherwise be thrown away — which is why it works even on a congested grid, where new generation cannot. But that is only a partial answer: landing 20–25 GW into an already-congested grid also demands coordinated TenneT reinforcement, and the corridor must be sequenced behind that build, not ahead of it — the single biggest execution risk, owned squarely in Appendix F. The full capital stack, ownership thresholds, and returns are worked through in Appendix C.

What to do — starting now

None of this requires tearing up the coalition agreement. It requires supplementing it — before the CfD ink dries. Here is how to start.

1. In the first 90 days: commission a joint TenneT–RTE–Red Eléctrica feasibility study for the corridor, and convene the willing — Spain, France, Germany, Poland — around a Dutch-led letter of intent. Cost: a few million euros to test a hundred-billion-euro question.
2. Preserve optionality on wind. Sign shorter, smaller contracts (10–15 GW, as a winter hedge) instead of locking 40 GW for fifteen years at today's elevated prices.
3. Launch the bridge now (2–5 years): rapid domestic and Iberian PV, storage, and "curtailment-to-compute" — siting AI compute where clean power would otherwise be wasted — at Dutch nodes such as Eemshaven and the Maasvlakte.
4. Build the corridor and the factories (5–10 years): majority-EU JV gigafactories via IPCEI; HVDC capacity scaling in phases; the manufacturing coalition standing up in Germany and Poland.
5. Let nuclear and SMRs arrive (10–25 years) as the sovereign baseload industry matures, with the Spine carrying the load until they do.

The phased roadmap — milestones, responsible parties, and decision gates — is in Appendix D.

The Netherlands is too small to generate all its own power. It is perfectly placed to lead the system that does — to finance it, anchor it, and build the industries behind it. That is not weakness. In the AI age, cheap and sovereign energy is the single greatest advantage a country can have.

We are late. Not too late.

Appendix A — The numbers, fully crunched

This appendix lays out the full energy-and-cost model behind the brief. Every figure is sourced or derived, and every derivation is shown. Where a number is our own extrapolation rather than a published projection, it is labelled [our extrapolation]. Where a figure would not survive a hostile expert without a caveat, the caveat is stated in the same breath. The honest bottom line is stated up front: at defensible capacity factors the Dutch system reaches roughly 167 TWh in the mid-2030s against a ~200 TWh central-case demand. The gap of ~33 TWh is exactly what the Solar Spine is for — it is the swing line, scalable from a modest start to ~150 TWh.

A.1 — The demand picture

The case for new firm clean supply rests on demand that is rising, not flat. The anchors:

Table A.1.

The ~200 TWh figure is ours, not PBL's. Published 2030 work clusters at 138–182 TWh. We extend to the mid-2030s by adding the loads that arrive after 2030 and are largely absent from the 2030 numbers: green-hydrogen electrolysis at scale, the ASML Eindhoven campus and comparable compute build-out, Tata Steel's green-steel conversion, and continued transport/heat electrification. Taking the upper-2030 range (~159–182 TWh) and layering these post-2030 loads lands near 200 TWh. Treat 200 TWh as a central case with a ±15 TWh band, not a forecast. The 238 TWh EU data-centre figure circulating in some decks is a high scenario, not the IEA/EC central case — we do not lean on it.

A.2 — The mid-2030s supply mix, every CF shown

The master identity is annual energy (TWh) = capacity (GW) × 8,760 h × capacity factor (CF) ÷ 1,000. The 8,760 is hours in a year; CF is the fraction of nameplate actually delivered. Every line below shows the arithmetic.

Table A.2.

Offshore-wind caveat (load-bearing). The 54.8 TWh result requires a ~50% CF, which is realistic only for a new, large-rotor, far-from-shore fleet. The 2023 actual Dutch fleet ran nearer ~38% (CLO wind capacity). The discipline: never pair the low end of the GW range with the high end of the CF. At 10 GW × 45% you get ~39 TWh; at 15 GW × 50% you get ~66 TWh. We carry 12.5 GW × 50% ≈ 55 TWh as the central line and flag the spread.

Nuclear caveat. Borssele (0.485 GW net) delivers ~3.8 TWh/yr at >90% availability (World Nuclear Association: Borssele). A first new Dutch reactor is not realistic by 2035 (Min. Hermans, Feb 2025), and EU new-build runs ~10–15 years decision-to-grid. So the mid-2030s nuclear line is honestly only ~4–8 TWh; the larger 10–15 TWh contribution belongs in the late-2030s/2040s.

A.3 — The Spine, decoupled: cable rating vs PV nameplate

A tempting but wrong shorthand is to write the Spine line as "32–40 GW → 80–100 TWh." That pairing implies a 28.5% capacity factor — physically impossible for Iberian PV, whose ceiling is ~24%. The fix is to state two separate numbers: the HVDC cable rating (a transmission spec) and the PV nameplate behind it (a generation spec). They are not the same quantity.

Step 1 — what does Iberian PV actually yield? Southern-Spain specific yield is ~1,700–2,000 kWh/kWp (PVGIS / JRC). Converting to CF: 1,900 kWh/kWp ÷ 8,760 h = ~21.7% CF. We use ~22% as the central planning CF (24% is the optimistic ceiling).

Step 2 — HVDC delivery losses. The route Iberia→Netherlands is ~2,400+ km. Line loss ~3–4% per 1,000 km plus ~1.8% at each converter station (HVDC losses, Wikipedia synthesis of cited figures):

• Line: 2,400 km × 3.5%/1,000 km = ~8.4%
• Converters: 2 × ~1.8% = ~3.6%
• Combined (multiplicative): 1 − (0.916 × 0.964) ≈ ~11.7%; with a higher-voltage/lower-loss line nearer ~9%.

We carry an end-to-end loss band of ~9–11%. Use 10% as central.

Step 3 — PV nameplate needed to deliver a target. Delivered TWh = nameplate × 8,760 × 0.22 × (1 − 0.10).

Table A.3.

Step 4 — HVDC cable rating is separate and smaller. A corridor does not need to be rated for full PV nameplate; PV peaks midday while the cable runs more hours via storage/firming. A realistic corridor is ~20–25 GW of cable rating behind which sits ~46–57 GW of PV for the 80–100 TWh tier, or ~19 GW of PV for the gap-filler start. State both numbers; never let one stand in for the other.

A.4 — The honest system sum (central case ± band)

Summing the defensible lines for the mid-2030s:

Table A.4.

The ~69 TWh "residual" is the balance of today's ~109 TWh system that is not domestic solar/wind/nuclear (conventional dispatchable, biomass, net imports) — it shrinks over time but does not vanish by the mid-2030s. The key honest statement: best-case lines can be stacked to ~195–205 TWh, but at defensible CFs the system lands ~167 TWh, a ~33 TWh shortfall. The Spine is the swing line that closes it, and it scales — from ~19 GW PV / ~33 TWh at the start to ~57 GW PV / ~100 TWh, and onward toward ~86 GW PV / ~150 TWh as demand and electrification climb. Present 200 TWh as central, with the Spine sized to the realised gap.

A.5 — Cost comparison (with caveats)

Table A.5.

Sources: NREL/LBNL Cost of Wind Energy Review 2024; Nederwiek I-A zero-bid & 2026 subsidised round. The honest framing: Iberian generation is far cheaper than North Sea wind, but delivered Spine cost must carry HVDC capex and ~10% losses, so the gap narrows. The Spine wins on a falling cost curve (solar is a manufactured semiconductor product) where offshore wind's costs are flat-to-rising; it does not win by a headline LCOE alone. We publish only an indicative band (~€45–60/MWh delivered), not a tendered price, until corridor capex is procured — an open item for the bankability study in Appendix C.

A.6 — Solar yield by region (why the south, not the north)

Table A.6.

Source: PVGIS / JRC; Andalusia GHI ~1,750–1,950 kWh/m². A panel in Seville produces roughly twice the energy of the same panel in Rotterdam — the entire physical rationale for generating in Iberia and transmitting north.

A.7 — Current installed capacity (the starting line)

Table A.7.

Spain added ~6.5 GW (2024) and ~9 GW (2025) of solar — the build rate that makes ~46–57 GW of additional corridor-dedicated PV a question of capital and offtake, not of feasibility.

A.8 — AI / data-centre demand (the reason this is urgent, sized honestly)

Table A.8.

The AI load is real and growing fast, but it is not, by itself, 200 TWh of new Dutch demand — it is one of several post-2030 loads (alongside electrolysis, green steel, and electrified transport/heat) that together push NL toward the ~200 TWh central case. We size the urgency on the combined load, not on data centres alone.

Appendix B — The legal & regulatory toolkit (A-to-Z)

This appendix is written so a ministry lawyer can trace every move. For each instrument we give: what it is, the exact legal mechanism, the thresholds and timelines, how the Spine triggers it as a Dutch-led project, and a primary citation. We also state, honestly, where an instrument is solid and where it is merely defensible.

A guiding principle runs through all of it. The Spine does not need Brussels to run a programme. It needs Brussels' statute book as a toolbox that a coalition of capable member states — the Netherlands as financier-anchor, Germany and Poland as manufacturers, Spain and France as solar hosts — can pick up and use. The decisive lever is not punishment. It is a guaranteed, multi-decade demand signal (the carrot), de-risked by EU funding rules and protected by EU economic-security law. Trade-defence instruments are the backstop, not the spearhead.

A note on the moving parts. As of June 2026 the FDI Screening Regulation is mid-revision (political agreement Dec 2025; Council provisional text Feb 2026) and a separate Industrial Accelerator Act is in train. Both are flagged below as "in flight" — the lawyer must check the consolidated text at the moment of use, not this appendix.

B.1 The primary carrot — guaranteed offtake (a long-term contract, not a subsidy)

What it is. A multi-decade, indexed offtake contract — a power-purchase agreement (PPA) or a state-backed Contract-for-Difference (CfD) — under which the coalition (anchored by Dutch sovereign credit and EU pension capital, see Appendix C) commits to buy a defined volume of Spine electricity, and of the modules / cells / cable made by the new European industry, at a known price for 15-25 years.

The mechanism it invokes. This is not an EU instrument at all — that is the point, and the reason it survives the "Dutch-led shrinks the carrot" critique. A long-term private or state-backed purchase commitment needs no Commission programme. It is the single thing a manufacturer needs to raise project finance: bankable, contracted demand. The graveyard of European PV (Solarworld, Meyer Burger's European lines, NorSun's closed Norwegian ingot/wafer plant — shut 18 December 2024 despite cheap hydropower, killed by Chinese under-pricing) failed for lack of exactly this. Cheap power alone did not save NorSun; guaranteed offtake might have.

Triggered for the Spine. The coalition issues long-term offtake tenders that bundle (a) Spine electricity and (b) European-made hardware, with the offtaker-shareholders (ASML, Tata Steel green-steel, chemical clusters, electrolyser operators) as anchor buyers. This is wholly within the coalition's gift. It does not depend on "access to the EU single market" — which a single member state cannot grant — so it is unaffected by going Dutch-led.

WTO survivability. A genuine commercial purchase contract is GATT-neutral. The risk only arises if the offtake is conditioned on local content (buy-European mandates), which would expose it to GATT Art. III (national treatment) and the Agreement on Subsidies and Countervailing Measures. The fix: keep the offtake open on its face and let the resilience-weighted procurement of §B.3 do the steering. Offtake = the carrot; resilience criteria = the legal steering wheel.

B.2 IPCEI + State aid — the funding vehicle (structured as a Dutch-led IPCEI)

What it is. An Important Project of Common European Interest (IPCEI) is the EU's mechanism for letting member states pour state aid into a strategic value chain that private capital alone will not bankroll, with the Commission blessing the aid as compatible with the internal market.

The mechanism invoked. State aid is in principle prohibited under Art. 107(1) TFEU. The IPCEI route uses the derogation in Art. 107(3)(b) TFEU ("aid to promote the execution of an important project of common European interest"), assessed under the Commission's 2021 IPCEI Communication. Approval is a Commission decision — there is no way around Brussels here, but it is an authorisation, not a programme Brussels runs.

Thresholds / structure / timelines.

• An IPCEI must be multi-member-state: it must involve more than one member state and its benefits must extend beyond the funding states.
• Aid covers R&D&I and first industrial deployment (not routine mass production).
• It requires Commission pre-notification and approval; approvals have historically taken 12-24 months.

Precedent (this is well-trodden). The hydrogen IPCEI Hy2Move was approved with up to €1.4 billion of public funding from seven member states (Estonia, France, Germany, Italy, the Netherlands, Slovakia, Spain), unlocking ~€3.3bn private — see Commission approves up to €1.4 billion for the fourth hydrogen IPCEI. The earlier hydrogen IPCEI ran to €5.4bn across fifteen member states (IP/22/4544). The batteries IPCEI ("Batteries II") mobilised €2.9bn public / €9bn private across 12 countries.

Triggered for the Spine. Structure the European PV-manufacturing and HVDC-cable build as a Dutch-led IPCEI with Germany and Poland as the manufacturing co-notifiers (and Spain/France on the generation leg). The Netherlands anchors the financing; the multi-state requirement is satisfied by the manufacturing partners — exactly the coalition the Spine already assumes. The Hy2Move precedent (NL already a participant) shows the pattern is live and the Netherlands is already inside it.

WTO survivability. State aid cleared as an IPCEI is internal-market-compatible under EU law, but it is still a subsidy in WTO terms. If it confers a benefit contingent on export or local content it is actionable. Keep aid tied to R&D&I and first deployment (as the framework requires) and avoid local-content conditions; this is the same discipline the existing hydrogen/battery IPCEIs operate under.

B.3 NZIA resilience criteria — the legal steering wheel for procurement

What it is. The Net-Zero Industry Act lets (and in defined cases requires) public authorities to weight renewable-energy auctions and other procurement on non-price "sustainability and resilience" criteria, not lowest price alone — the mechanism that lets a tender lawfully prefer a diversified, non-China-dependent supply chain.

The mechanism invoked. Regulation (EU) 2024/1735 (the NZIA), published in the Official Journal 28 June 2024, in force 29 June 2024 — Regulation (EU) 2024/1735, EUR-Lex. The detail sits in the implementing regulation on non-price criteria in renewable-energy auctions, adopted 23 May 2025 — see Adoption of Implementing Act on Non-Price Criteria in Renewable Energy Auctions, Covington and Herbert Smith Freehills Kramer analysis.

Thresholds / timelines.

Table B.1.

The critical honesty point. The resilience trigger is a Commission determination — it is not automatic. We must not claim it fires by itself. However, given China's ~95% share of upstream PV (polysilicon/ingots/wafers; ~98% of wafers in 2023), the >50% threshold for PV is essentially certain to be met once the Commission examines it. The accurate phrasing is: "the legal precondition is a Commission finding, and on the published market data that finding is all but inevitable for solar PV."

Triggered for the Spine. Two moves. (1) The coalition's own renewable-energy and grid-hardware auctions apply the resilience criterion to the PV-module and HVDC-component lots, lawfully favouring bids that source the finished product and a minimum number of main components from outside the dominant country. (2) The Netherlands and partners formally press the Commission to make the resilience determination for PV. A single member state can run NZIA-compliant auctions on its own volume — so this lever survives the Dutch-led downgrade; it does not need EU-wide unanimity.

WTO survivability. Non-price qualitative criteria applied transparently and non-discriminatorily are defensible under the WTO Government Procurement Agreement, which expressly permits non-price award criteria. The danger zone is again disguised local content. The resilience criterion is drafted around diversification away from a single source, not "buy-EU" — that framing is what keeps it GPA-survivable.

B.4 FDI Screening + GATT Art. XXI — protecting majority-EU ownership of the grid and cable

What it is. The two instruments that let the coalition keep the strategic backbone — the HVDC interconnector, converter stations and cable factories — under controlling European ownership, and screen out hostile foreign control.

Lead instrument: the FDI Screening Regulation. Regulation (EU) 2019/452 establishes a framework for screening foreign direct investment into the Union on grounds of security or public order, with critical energy infrastructure an explicitly listed sensitive sector — Regulation (EU) 2019/452, EUR-Lex.

In flight (verify at point of use): a revision is far advanced — Commission proposal 24 Jan 2024; political agreement 11 Dec 2025; Council provisional text published 10 Feb 2026 — making national screening mechanisms mandatory for all member states, setting a minimum mandatory-screening scope (incl. critical energy infrastructure), and — decisively — extending the regime to capture investments by EU-based entities that are themselves controlled by foreign investors, closing the subsidiary-circumvention gap. See Revision of the FDI Screening Regulation, European Parliament Legislative Train and Mayer Brown, FDI Screening Reform & the Industrial Accelerator Act.

Why screening leads, not Art. XXI. Screening is administrable national law: the Netherlands can review and block or condition a specific acquisition of a Spine asset on public-order grounds, case by case, with no need to win a trade dispute first. It is the workhorse. The "majority-EU ownership" design is enforced primarily here — and note that Shell, UK-incorporated since January 2022, counts as a foreign investor for this purpose and cannot be relied on toward the EU-control majority (it may co-invest as a minority/foreign party only).

Backstop instrument: GATT Art. XXI. The security exception in GATT Art. XXI(b) lets a member take action it considers necessary for the protection of its essential security interests. This is the treaty shield if a foreign state argues that excluding its investors from grid ownership breaches trade commitments.

The honest legal status — defensible, not airtight. Since the 2019 WTO panel in Russia — Measures Concerning Traffic in Transit (DS512), Art. XXI is justiciable: a panel will review whether the invoking state has plausibly articulated an essential security interest and whether the measure is plausibly connected to it, and applies a good-faith test against relabelling ordinary trade protection as "security" — Russia–Traffic in Transit, landmark ruling, HSF Kramer. A permanent peacetime ownership rule for grid assets is therefore defensible but contested — it is not a guaranteed win, because Art. XXI(b)(iii) is anchored in "war or other emergency in international relations." This is precisely why we lead with FDI screening (clean, administrable, uncontested) and hold Art. XXI as the fallback narrative, framing energy-grid control as a genuine essential-security interest (sabotage, coercion, single-point-of-failure dependence) rather than as industrial protection.

Triggered for the Spine. Cable, converters and interconnector SPVs are held by a majority-EU ownership structure; any non-EU acquisition is run through Dutch (and partner-state) FDI screening on critical-energy-infrastructure grounds; Art. XXI is reserved as the WTO-level justification should the ownership rule itself be challenged.

B.5 Foreign Subsidies Regulation — the backstop against subsidised bidders

What it is. The instrument that lets the EU neutralise the advantage a Chinese (or other) bidder gains from foreign state subsidies when it competes in the EU — including in the very procurement and M&A the Spine will run.

The mechanism invoked. Regulation (EU) 2022/2560 (the Foreign Subsidies Regulation), in force 12 Jan 2023, applied from 12 Jul 2023 — FSR legislation, European Commission. It is investigation-based and EU-administered (the Commission, not a national ministry, runs it).

Thresholds / timelines.

Table B.2.

This is live, not theoretical. The Commission opened its first ex officio procurement probe on 9 April 2024 into two Chinese wind-turbine makers (Envision and Mingyang) bidding for wind parks in Spain, Greece, France, Romania and Bulgaria; and, in a separate case, opened its first in-depth FSR wind investigation — into Goldwind — on 3 February 2026 — see The EU Foreign Subsidies Regulation explained, Norton Rose Fulbright.

Why it is a backstop, not the spear. The FSR is case-by-case and EU-run — a single member state cannot wield it to "exclude Chinese bidders" wholesale, and the earlier hope of using it as a blanket Dutch-led exclusion tool does not hold. It works as a deterrent and corrective once a distortive subsidised bid actually appears in a Spine tender, layered behind the resilience criteria (§B.3) that do the routine steering.

WTO survivability. The FSR is structurally designed to mirror EU state-aid discipline and is applied to subsidies that distort the internal market; it does not impose border tariffs. It is more WTO-robust than ad hoc anti-dumping duties precisely because it is conduct-based and individualised — though, being new, it is not yet stress-tested at the WTO.

B.6 What we deliberately do NOT use — CBAM

State this plainly so nobody reaches for the wrong tool. The Carbon Border Adjustment Mechanism (definitive regime from 2026) covers cement, iron and steel, aluminium, fertilisers, electricity and hydrogen — see Carbon Border Adjustment Mechanism, European Commission. Finished solar PV modules are not in scope. CBAM is therefore not a usable anti-China-PV lever and must be dropped from the toolkit. (It remains relevant to the green-steel offtaker, Tata Steel IJmuiden, but on the demand side, not as a PV trade defence.)

B.7 Precedent — China's playbook, framed honestly

The proof-of-concept. China built its automotive and high-speed-rail industries by making foreign-market access conditional on technology sharing. The USTR Section 301 Report (22 March 2018) found that China used joint-venture requirements, foreign-ownership/equity limitations and licensing to force or pressure technology transfer from foreign firms, with some JVs required to be majority Chinese-owned — Section 301 Report, USTR. In high-speed rail, foreign-JV-then-localise produced CRRC, which now holds roughly 90-95% of its home market.

The honest legal caveat (do not overclaim). The Section 301 finding of forced technology transfer is a US executive determination, not a WTO ruling. For high-speed rail in particular, there is no adjudicated WTO panel holding that China's JV regime violated WTO law — characterisations of "WTO violation" are analyst commentary, not law. So the precedent proves the strategy works commercially and politically; it does not prove it is WTO-clean.

What we take from it. The carrot — guaranteed, long-horizon demand — is what moved manufacturers into China and what can move them into Europe. The Spine replicates the commercial logic (offtake-for-localisation) while deliberately avoiding the unlawful mechanism: no coerced JVs, no mandated tech transfer, no majority-foreign-ownership decrees. We steer through resilience-weighted procurement (§B.3) and ownership screening (§B.4) — instruments that are EU statute, not Chinese-style fiat — so the Spine gets China's industrial outcome by lawful, WTO-defensible means.

B.8 One-page move-map for the ministry lawyer

Table B.3.

Appendix C — The capital, JV & ownership model

The Spine is not a subsidy. It is a balance sheet. The Netherlands' job in this coalition is not to host factories or to plant solar in the Veluwe — it is to do the one thing it does better than almost any state on earth: borrow cheaply, against a AAA rating, and lend that credit to assets that earn it back. What follows is the financing engine: where the money comes from, who carries which risk, what return it pays, and who ends up owning it. Every figure here is indicative — sized from real unit costs and real precedents, not from a finished bankability study — and is flagged as such. The point is to show the model holds arithmetically before a single euro moves, and to ground it in deals that already exist.

The discipline that makes it bankable is the same one that makes the rest of this brief work: guaranteed long-term offtake. The industries that need the power most — chip-makers, data centres, green steel, chemicals, electrolysers, and the PV gigafactories themselves — do not merely sign purchase contracts. They take equity, as offtaker-shareholders. Demand that would otherwise be a market risk becomes a contracted revenue line, and a contracted revenue line is what lets a pension fund hold the asset for thirty years.

C.1 What is being financed

The Spine is three balance sheets, not one, and they must be financed separately because their risk profiles differ by an order of magnitude. Conflating them is how Desertec died. We size the central case from the physical anchors derived in Appendix A.

Table C.1.

The corridor and the generation are the safe, incremental leg. The industries are the bold, high-reward leg. They share a coalition and a financing platform, but never a risk pool — a Chinese under-pricing shock to the gigafactory must not be able to default the cable.

C.2 The worked capital stack (indicative)

Unit-cost basis (derived, with sources)

• Iberian PV: ~€0.60/W installed (≈ USD 650/kW — in line with IRENA's 2024 global weighted-average utility-scale PV installed cost of ~USD 691/kW) → ~50 GW × €0.60/W = ~€30bn (sizes ~87 TWh net — between tiers; the ~80 TWh tier needs ~46 GW ≈ €28bn, the ~100 TWh tier ~57 GW ≈ €34bn). Source: IRENA, Renewable Power Generation Costs in 2024.
• HVDC corridor: Viking Link benchmark = 1.4 GW, ~765 km, ~€2bn (Viking Link, Wikipedia; cable ~€2–5m/km, converters ~€300–600m each, per NeoMarketData, 2025). A ~2,400 km, ~20 GW corridor is roughly 14× the Viking capacity over ~3× the route. Conservatively model ~€25–30bn for cable + converters + redundancy. Take ~€28bn.
• Sovereign industries: PV gigafactory cluster (poly→ingot→wafer→cell→module, ~10–20 GW/yr) ~€8–12bn; HVDC cable/converter manufacturing capacity ~€3–5bn; nuclear/SMR is financed on its own multi-decade track and is excluded from this stack's headline (flagged below). Take ~€14bn for the two industrial plays in scope.

Table C.2.

Sources of funds — the ~50/50 public–private split

The model targets ~50% public-backed / ~50% private, but "public" here means credit and catalytic equity, not grants. The state's contribution is mostly leverage on the AAA rating (cheap senior debt), not cash burned.

Table C.3.

Gearing is therefore roughly 45% debt / 55% equity at the programme level, but tranched: the corridor can carry far more debt (regulated, contracted) — think 65–70% — while the industrial play carries far less and leans on first-loss and IPCEI grant.

C.3 Indicative return and payback

All figures indicative. Returns are tranched to match risk, and benchmarked against real market hurdle rates: EU regulated transmission earns a ~5.5–6% regulated return (quasi-bond), while competitive renewable/infra equity funds target ~9–11% net IRR (market benchmarks, 2025).

Table C.4.

Worked illustration (corridor + generation, blended). Take the safe leg only: ~€58bn CAPEX (corridor + PV), delivering ~80–100 TWh/yr. The economics here are locational arbitrage, not a bet on Iberian spot prices: the corridor buys cheap southern midday power and delivers it into the structurally higher Dutch market, so the return depends on the south–north price spread persisting, not on the Iberian capture price. At an indicative net delivered value of ~€30–40/MWh — a contracted floor via a financial CfD over the interconnector, set against today's sub-€30 Iberian PPAs at one end and a higher Dutch delivered price at the other — ~90 TWh yields roughly €2.7–3.6bn/yr gross margin against ~€58bn deployed: a ~5–6% running cash yield before leverage, lifted into the ~8–10% equity range once ~45–65% cheap senior debt is layered in. A utility-grade return, not a venture return — which is what makes it ownable by ABP for thirty years.

The key sensitivity is the spread, and it is widening. Growing Iberian midday cannibalisation (record negative-price hours in 2025–26) is, counter-intuitively, a tailwind for an export corridor — it deepens the cheap southern trough the corridor monetises — even as it is a headwind for unhedged Iberian generation equity. The CfD floor is therefore an arbitrage hedge, not an out-of-market subsidy: the corridor earns the difference between two real prices. The risk to own honestly is the reverse case — if Northern European evening prices fall toward Iberian midday levels (mass storage, abundant wind), the spread compresses and the merchant tail thins; the floor then carries more of the return.

C.4 Majority-EU ownership and the security/FDI process

The ownership rule is not a slogan — as of 4 March 2026 it is a draft EU law. The Industrial Accelerator Act (IAA), proposed by the Commission, sets a 49% cap on foreign participation in entities undertaking manufacturing in designated strategic sectors (batteries, EVs, solar PV, critical raw materials), and requires a JV structure with EU participation, IP licensing, EU-based R&D, and workforce commitments where an investor's home country holds >40% of global capacity (Skadden; Dechert). China's ~95% PV-upstream share triggers it on its face. In parallel, the revised EU FDI Screening Regulation (political agreement 11 Dec 2025) makes screening of foreign investment in energy, transport and digital infrastructure mandatory across all member states, applying from ~H2 2027 (Consilium).

Thresholds applied to the Spine:

Table C.5.

The Shell correction — load-bearing. Shell plc reincorporated in the United Kingdom in January 2022 (name change from Royal Dutch Shell, tax residence and HQ moved from the Netherlands; Shell 6-K, SEC). For EU screening it is therefore a foreign (non-EU) investor. Shell can co-invest, but only inside the ≤49% foreign envelope, and its stake cannot count toward the ≥51% EU-controlling bloc. The same applies to Norway's sovereign fund and Equinor: welcome capital, minority status. Critically, the revised FDI rules also capture EU-incorporated entities that are themselves controlled by a non-EU parent — so a Dutch SPV owned by a foreign group does not launder the foreign stake. The EU bloc must be genuinely EU-controlled.

C.5 JV, IP and knowledge-transfer terms — grid vs PV, differentiated

The two industries are gated by different instruments, so their JV terms differ.

Table C.6.

This is the "carrot, not stick" logic in contract form: Europe does not forbid Chinese capital; it makes a majority-EU-owned, IP-licensing JV the only route to the Spine's guaranteed offtake and IPCEI-funded slots — the same outcome China obtained in autos and rail by forcing 50% JVs and tech transfer (USTR Section 301 findings), achieved here with rules Europe already wrote. Defensible — not literally appeal-proof.

C.6 Risk allocation by phase

Table C.7.

C.7 Beneficiaries as offtaker-shareholders — how demand de-risks the stack

The whole model rests on turning demand into equity. Each of these is a real, named load:

• ASML — image anchor and proof Europe can lead a global industry; a modest direct consumer (~0.26 TWh/yr; its RWE green PPA is ~257 GWh/yr from 2021, ASML), but its ~20,000-job Eindhoven campus ran into grid-congestion limits requiring dedicated government action — the human face of "we are grid-short."
• Mistral — Europe's compute-sovereignty champion; a genuine future offtaker as it builds out to ~1 GW of compute.
• Volume offtaker-shareholders (the real demand): data centres / sovereign AI compute (NL already ~5.1 TWh, ~5% of national power); Tata Steel IJmuiden green-steel conversion (the single biggest industrial load); chemical clusters (Chemelot, Dow Terneuzen); green-hydrogen electrolysers (Holland Hydrogen, NortH2, Port of Rotterdam); and the PV gigafactories themselves — power-hungry ingot/wafer furnaces that close the cheap-power → manufacturing loop.

The flexible loads among them — electrolysers, data centres, ingot furnaces — do double duty: as offtaker-shareholders they de-risk the revenue line, and as interruptible demand they absorb surplus midday solar into product instead of curtailment, which is also the cleanest answer to the grid-absorption question.

C.8 Real precedents — this is not unprecedented

The model is assembled from deals that already closed:

1. TenneT Germany equity financing (2025–26) — the closest precedent. TenneT secured up to €9.5bn from NBIM, APG and GIC for a ~46% stake, while Germany's KfW took 25.1% for ~€3.3bn, and TenneT launched a €35bn debt issuance programme (GIC; BMWE). This is exactly the structure: state-anchored grid company, EU pension (APG/ABP) as patient equity, sovereign fund as minority — at multi-billion scale, already done.
2. EIB grid lending — €11.6bn to grids & storage in 2025, financing the Bay of Biscay (France–Spain, ~2 GW, ~2028) interconnector among others (EIB, 2026; INELFE).
3. IPCEI co-funding — the Commission has approved hydrogen IPCEIs unlocking large private co-investment: Hy2Tech ~€5.4bn public → ~€8.8bn private; Hy2Use ~€5.2bn → ~€7bn private (EC, hydrogen IPCEI). A Dutch-led PV/HVDC IPCEI with Germany and Poland is the precedented vehicle for the industrial leg.
4. Pension-fund infrastructure mandates — APG/ABP and PFZW already hold large unlisted infrastructure; the TenneT deal proves the appetite is real, not hypothetical.
5. Anchor-offtake project finance — the standard renewables financing template (long-term PPA → bankable debt) is exactly what the offtaker-shareholder structure formalises and extends to equity.

Bottom line. The Netherlands borrows AAA-cheap and anchors; EU pensions take the controlling equity and the utility-grade return; EU operators run the assets; the power-hungry industries take equity and guarantee the demand; foreign capital is welcome up to 49% and no further. The corridor and generation are bankable today on contracted cash flow; the sovereign industries are the bet, ring-fenced and grant-supported so their risk can never reach the cable. The whole structure already exists in pieces — TenneT, the EIB, the hydrogen IPCEIs. The novelty is assembling them, at scale, under one Dutch-led platform, with demand as equity. The numbers hold — at indicative precision. The next step is a bankability study, not a leap of faith.

Appendix D — The A-to-Z roadmap ("pen to power")

This appendix turns the thesis into a sequenced plan a ministry official could open tomorrow and act on. It is organised in four honest horizons — 2, 5, 10 and 25 years from June 2026 — each with concrete milestones, a named owner, and a decision gate that must clear before the next phase starts. It assumes no heroics: HVDC submarine corridors take roughly 5-8 years from final investment decision to energisation, PV gigafactories commission in roughly 18-36 months, and a new Dutch nuclear reactor will not be on the grid by 2035 (World Nuclear News, 12 Feb 2025). The plan is built around those constraints, not against them.

A note on roles before the timeline. "Dutch-led" does not mean Dutch-built. The Netherlands is the financier and anchor offtaker — it borrows at AAA, underwrites demand, and operates the landing infrastructure through TenneT. Spain and France sell sun and host the southern converter stations (via Red Eléctrica/REE and RTE, the same two TSOs already building the Bay of Biscay link through their INELFE joint venture). Germany and Poland host the manufacturing (PV, cable, converter, SMR supply chain). Brussels supplies the tools — NZIA resilience auctions, IPCEI state-aid clearance, FDI screening, EIB capital — not the programme management.

Naming convention used below

Table D.1.

Horizon 0 — The first 90 days (what to do tomorrow)

Nothing on the critical path requires legislation in the first quarter. It requires a task force, a mandate letter, and three feasibility commissions so that the long-lead items (corridor routing, gigafactory siting, IPCEI notification) start their clocks now rather than after an election cycle.

Table D.2.

Decision gate to proceed past day 90: TenneT interim absorption finding is not a hard "no," and the coalition LoI is signed by at least Spain plus one manufacturing state.

Horizon 1 — ~2 years (mid-2026 → 2028): the bridge plan

No new corridor and no gigafactory can deliver electrons in two years. So the bridge phase buys time and de-risks demand using assets that can move fast: domestic PV, batteries, and — critically — curtailment-to-compute, co-locating flexible loads (AI/data centres, electrolysers) at the grid edges where surplus solar and offshore wind are already being thrown away.

Bridge milestones

Table D.3.

Decision gate at end of Horizon 1: TenneT absorption study green-lights ≥20 GW landing; IPCEI cleared or near-cleared; coalition financing term sheet committed → take the corridor to Final Investment Decision.

Honest caveat: the bridge phase delivers flexibility and demonstration, not bulk new clean TWh. Its job is to keep the lights on credibly while the long-lead assets are built, and to prove the offtake-and-finance machinery at small scale.

Horizon 2 — ~5 years (2028 → 2031): break ground

This is where the plan becomes physical: corridor FID, first gigafactory steel, and the start of the parallel nuclear/SMR industrial track (note: industry-building, not reactors-on-grid — that comes in Horizon 3-4).

Corridor

Table D.4.

Build realism: an Iberia-NL HVDC corridor is a 5-8 year build from FID, so a 2028-29 FID means energisation around 2034-37, not before. Anyone promising imported Iberian electrons by 2030 is selling a fiction.

PV manufacturing (Germany / Poland)

Table D.5.

Own the graveyard: NorSun closed its Norwegian ingot/wafer plant in December 2024 despite cheap hydropower, undercut by Chinese pricing — the cautionary tale this track must beat. The answer is not hope but three locked conditions: guaranteed long-term offtake (NZIA resilience auctions), capex (IPCEI), and trade defence (FDI screening + FSR). Without all three, the factory closes again.

Nuclear / SMR industrial track (parallel, multi-country)

The nuclear track in this horizon is about building an industry and a pipeline, not pouring reactor concrete. France (EDF, six firm EPR2 reactors with an option for up to eight more) is the anchor; the Netherlands stands up its new state-owned nuclear company and progresses siting (Borssele region; Eemshaven added as a candidate) and technology selection (EDF, Westinghouse in the running after KHNP withdrew, World Nuclear News).

Table D.6.

Industry-building milestones for this horizon: a multi-country SMR supply-chain consortium (forgings, pressure vessels, fuel), standardised licensing cooperation, and Dutch FID preparation. No new NL reactor reaches the grid in this window or the next.

Horizon 3 — ~10 years (2031 → 2036): electrons flow

Table D.7.

System honesty for the mid-2030s NL mix (~200 TWh central case, ± band): domestic solar ~35-38 TWh, offshore wind ~50-55 TWh (12.5 GW new far-shore fleet @ ~50% CF), nuclear only ~4-8 TWh this decade (Borssele plus at most one unit; the bulk of nuclear TWh is a 2040s contribution), firming gas declining. At defensible capacity factors the domestic lines sum to roughly 167 TWh — the Solar Spine is the swing line that carries the ~33 TWh gap, scalable toward ~86 GW PV / ~150 TWh as PV is added behind a fixed cable.

Horizon 4 — ~25 years (to ~2051): the industrial base stands

Table D.8.

Decision-gate summary

Table D.9.

The discipline of the plan is in the gates. Each one can stop the project before it wastes capital — which is exactly why Desertec (no firm offtake, no grid, transit-country risk) and Xlinks (rejected by the UK as "first-of-a-kind, high inherent risk," June 2025) failed, and why a gated, intra-EU, existing-TSO, guaranteed-offtake structure is a different animal.

Appendix E — Jobs & the engineering-base case

The jobs-and-engineering-pride argument makes a human case, so it must rest on human numbers. This appendix assembles the employment evidence behind the three sovereign industries the Spine builds — European PV manufacturing, an HVDC grid-and-cable industry, and nuclear/SMR — and sets it against the displacement of knowledge work that the brief itself warns is coming. The discipline of the Pride Principle applies with extra force here: jobs studies are the most advocacy-soaked numbers in the whole energy debate. Almost every figure below was published by an industry trade body that wants the policy. We use them, but we say so, and we lean on the conservative end of every range.

A note on what these numbers are and are not. "Jobs" figures come in several flavours that are routinely — and sometimes deliberately — blurred: direct (on the payroll of the industry), indirect (suppliers), induced (the barber who cuts the welder's hair), and job-years (one person employed for one year — so a ten-year build of 10,000 job-years is 1,000 steady jobs, not 10,000). Trade-body headlines tend to quote the largest available number. Where a source mixes categories we flag it. Treat every total as an order-of-magnitude planning input, not a forecast.

E.1 Sizing the three industries — the headline ranges

The table gives the published European totals for each industry, the body that published them, and our read on how much weight they bear. These are whole-industry numbers for Europe; the share attributable to a Dutch-led Spine coalition is a fraction of them (see E.2).

Table E.1.

Honest reading of the table. Three of the four headline numbers are trade-association figures, and the two largest — solar's "1 million" and nuclear's "1.1 million / 3 million" — carry the most advocacy weight. The nuclear figure in particular leans on a 3.2× economy-wide job multiplier and counts the entire existing fleet's supply chain; it is a contribution estimate, not a count of people who would lose their jobs without nuclear. The solar number was quietly downgraded between 2021 ("triple to 1.1m by 2030") and 2025 ("1 million from 2030 onwards, growth stalling") — a useful caution against believing any single-year projection. The HVDC/cable number is the most credible because it is bottom-up and plant-level, and also the smallest in absolute terms.

E.2 From European totals to the Spine's share — a worked, conservative estimate

The trade-body totals above are for all of Europe across all projects. The Spine coalition is one (large) programme. To avoid the classic advocacy error of claiming a whole industry's employment for one project, we build a bottom-up estimate from the Spine's own physical scope and published per-unit job intensities, then take the conservative end.

Inputs (all from the brief's verified anchors and the sources above):

• PV manufacturing: the loop the brief closes is European-made modules for the corridor and domestic build-out. A serious EU gigafactory cluster (ingots → wafers → cells → modules) sized to supply tens of GW/yr is in the low tens of thousands of direct manufacturing jobs at maturity — consistent with SolarPower Europe's ~91,000 figure being the EU-wide manufacturing ceiling, of which a Dutch-led DE/PL cluster would be a meaningful minority.
• HVDC cable & converters: at ~1,000 on-site jobs per large extruded-HVDC plant (Nexans Halden), a corridor needing a multi-GW cable supply plus converter stations implies on the order of a few thousand direct manufacturing jobs spread across DE/IT/FR champions, with the build/installation phase adding more job-years.
• Construction & installation (the bulk, but temporary): HVDC corridor laying over ~2,400+ km, converter stations at both ends, and tens of GW of Iberian PV fields. This is large but is measured in job-years, not permanent posts — it peaks during the 2028–2035 build and tapers to O&M.
• Nuclear/SMR: at the Deloitte/Foratom rule of thumb of ~10,000 jobs per GW (direct + indirect + induced), the brief's mid-2030s ~0.5–1 GW of NL nuclear online is ~5,000–10,000 jobs on that (generous) basis — but most of that arrives in the late-2030s/2040s as the fleet and an SMR supply chain mature, matching the brief's own corrected nuclear timeline.

Conservative Spine-attributable estimate (planning band, not a forecast):

Table E.2.

We deliberately publish the ~13,000–33,000 permanent direct jobs band rather than a single hero number, and we keep construction job-years separate so no critic can accuse us of the "10,000 job-years = 10,000 jobs" sleight of hand. Apply standard indirect/induced multipliers (the literature clusters around 1.5–2.5× for manufacturing; nuclear's own studies claim 3.2×, which we treat as an upper bound) and the wider economic footprint is larger — but the defensible, payroll-level claim is the band above.

E.3 Why these jobs are AI-resilient — and the displacement they answer

The brief's premise is that AI will hollow out knowledge work. The evidence base, stripped of hype:

Table E.3.

The point is not that AI is a net job-killer — the WEF's own central case is a net gain. The point is which jobs and which people. The exposure is concentrated in exactly the white-collar, screen-based work (legal research, financial modelling, drafting, junior analysis) that Europe spent two generations steering its graduates into — while the work that is hardest to automate or offshore is physical, sited, safety-critical engineering and manufacturing: laying a 525 kV submarine cable, commissioning a converter station, running an ingot furnace, maintaining a reactor. You cannot prompt-engineer a HVDC splice in the North Sea. The Spine is therefore not only an energy plan; it is a re-employment plan that moves value back toward work that has to be done by people, in place, in Europe. If we lose jobs to AI, we win them back by building.

A second, harder-edged point: WindEurope's own report warns of acute shortages before 2030 — 7,000 blade technicians, 6,500 field engineers, 5,000 pre-assembly technicians needed and not yet trained. The constraint on this build-out is not a lack of work for people; it is a lack of trained people for the work. That inverts the usual jobs anxiety and turns the appendix's argument into an operational mandate: the binding scarcity is skilled hands, so the policy lever is vocational training and apprenticeships, not job creation in the abstract.

E.4 The engineering base Europe already has — and the Dutch heritage to anchor it

The case is not "build a workforce from zero." Each industry plugs into an existing European cluster, and the Netherlands brings world-class anchors in precisely the disciplines the Spine needs.

Pan-European clusters the three industries draw on:

• PV manufacturing plugs into Europe's surviving non-Chinese upstream — Wacker (DE) polysilicon — plus process-engineering depth in Germany and a perovskite-tandem research base (Oxford PV, Fraunhofer). Honest caveat from the brief: this base is contested and shrinking (NorSun closed Dec 2024; Meyer Burger, Solarworld before it), so the jobs are a bet on trade defence + offtake + capex together, not a given.
• HVDC grid & cable has genuine European champions: Siemens Energy (DE), Prysmian (IT), Nexans (FR) — a real industrial base, expanding now (Nexans Halden doubled; Prysmian ~$2bn capex). This is the strongest employment story because the firms, factories and order books already exist.
• Nuclear/SMR draws on France's EDF/Framatome fleet skills and a continent-wide reactor supply chain; SMRs additionally open a new manufacturing line (factory-built modules) that suits Europe's precision-engineering strengths.

The Dutch engineering heritage that anchors the coalition:

Table E.4.

The dredging/marine cluster is the single most under-appreciated Dutch asset for this brief: the same companies and crews that build artificial islands and lay subsea infrastructure worldwide are precisely the firms that would install the Spine's offshore segments and converter platforms. The heritage argument is not sentiment — it is a transferable, already-employed workforce.

E.5 The honest bottom line

A Dutch-led Spine coalition can credibly claim, at the payroll level, on the order of 13,000–33,000 permanent direct engineering and manufacturing jobs at maturity across PV, HVDC/cable and nuclear/SMR, plus a multi-year construction wave measured in tens of thousands of job-years, plus a larger indirect/induced footprint we decline to inflate. These are jobs in work that AI cannot offshore or automate wholesale, drawing on engineering clusters Europe — and the Netherlands specifically — already possesses. We do not claim the trade bodies' whole-Europe millions for one project, and we flag that the solar and nuclear headline figures are advocacy-funded and have a track record of being revised down. The argument holds at the conservative end; that is the only end the Pride Principle lets us publish.

Appendix F — Precedent & risk register

Every mega-project of this shape has a graveyard behind it. We do not pretend otherwise. This appendix sets the Solar Spine against the two failures everyone will raise — Desertec and Xlinks — separates the failure causes we genuinely avoid from the ones we still carry, states the China industrial-policy precedent we are deliberately copying, and then answers the ten hardest objections in turn. Where a risk is real and unresolved, it is listed as a residual risk, not argued away.

F.1 Why this is not Desertec

Desertec was launched in 2009 with the ambition of raising ~€400bn and powering a meaningful share of Europe from concentrated-solar plants in the Sahara and the Middle East. The industrial consortium (Dii) shrank from 17 partners to three by the end of 2014, and the concept was widely written off by 2014–2015 (Desertec, Wikipedia; Dialogue Earth, "Desertec's plan ... burns out").

It failed for identifiable reasons, and the Spine is engineered against each:

Table F.1.

F.2 Why this is not Xlinks

In June 2025 the UK government declined to support the Xlinks Morocco–UK project — 11.5 GW of Moroccan solar/wind feeding a 3,600 MW, ~4,000 km subsea HVDC cable, then proposed as the world's longest. The reason given was explicit: it "does not clearly align ... with the Government's mission to build home-grown power" and was "a first-of-a-kind mega-project with a high level of inherent, cumulative risk — delivery, operational, and security" (UK Parliament written statement HCWS745, 26 Jun 2025; Xlinks Morocco–UK, Wikipedia).

The Spine differs on each count the UK cited — and we name the one it shares:

Table F.2.

Proven HVDC building blocks the Spine extends (each already in commercial operation or financed):

Table F.3.

The Bay of Biscay link is the decisive precedent: a France–Spain HVDC connection, EIB-financed, due 2028 — the first segment of exactly the corridor the Spine extends northward.

F.3 The precedent we are copying: China's industrial leverage

The Spine's industrial leg deliberately mirrors the strategy China used to build its high-speed-rail champion. China linked access to its domestic rail market to forced technology transfer via joint ventures: Alstom, Bombardier, Siemens and Kawasaki were required to transfer technology to Chinese partners as a condition of contracts. The result is CRRC at ~90–95% of China's domestic high-speed-rail market — and the world's largest maker of high-speed rolling stock (ITIF, "Heading Off Track", 2021).

Two honest caveats:

• ITIF characterises the JV practice as "in violation of WTO rules," but no adjudicated WTO panel ever ruled it a violation — this is analyst characterization, not a legal finding.
• The autos parallel (~50% JV requirement; USTR Section 301 forced-tech-transfer findings) is the better-documented case and should lead.

The Spine copies the logic (use guaranteed market access + offtake as the carrot to anchor a domestic industry) within EU-legal instruments (resilience-weighted procurement under the NZIA, IPCEI funding, FDI screening) rather than the method (coerced JVs).

F.4 Objection → rebuttal register

Each row states the strongest form of the objection, the rebuttal, and — critically — the residual risk we still own.

Table F.4.

F.5 The manufacturing graveyard (own it)

The "build our own PV industry this time" claim must carry its own dead. Recent European upstream-PV failures:

• NorSun (Norway) closed its ingot/wafer plant 18 Dec 2024 despite cheap hydropower, citing Chinese under-pricing — a cautionary tale, not proof EU manufacturing works.
• Meyer Burger and Solarworld before it failed on the same Chinese cost wall.

The lesson, stated in the brief: cheap Spine power lowers the energy-intensive ingot/wafer cost (Czochralski growth at ~1,400 °C), but cheap power alone is insufficient. The industry survives only with the full package — trade defence + guaranteed offtake + capex support — which is precisely what the re-industrialisation case + the NZIA resilience criteria are for. Anything less repeats NorSun.

F.6 Residual risk summary (the four we cannot fully retire)

1. Grid absorption (TenneT) — the binding near-term constraint; corridor must be sequenced to grid reinforcement (Objection 7).
2. Manufacturing economics — EU upstream PV dies without the full trade-defence + offtake + capex package; China exposure persists ~10 years (Objection 3, F.5).
3. Winter firm capacity — the Spine cannot cover Dunkelflaute; nuclear + firmed gas + storage must be funded in parallel or the security case fails (Objection 4).
4. System stability — low inertia is a demonstrated failure mode (Iberia, Apr 2025); grid-forming + synchronous condensers must be in the capex (Objection 8).

These four are not arguments against the Spine. They are the conditions under which it is honest. Build to them and the numbers hold; ignore any one and the plan is Desertec again.

Appendix G — Sources & references

Every load-bearing number, legal claim, and precedent in this brief and its appendices traces to a primary source listed below. Where a primary document (an EU regulation, a statistics-office dataset, a WTO panel report, a company or agency release) exists, it is cited directly; where we lean on authoritative trade or news reporting because no public primary document carries the specific datum, that is noted. URLs were verified in June 2026; nothing here is invented. Figures that are our own extrapolation rather than a sourced number (notably "NL ~200 TWh mid-2030s" and the system-mix capacity-factor maths) are labelled as such in Appendix A and are deliberately not attributed to a third party here.

A reader who wants to audit the brief can work through this list group by group. Each entry gives the claim it supports, the title, the publisher, and the date.

Demand, system numbers & capacity factors

Table G.1.

Wind: tenders, auctions & cost

Table G.2.

NREL US fixed-bottom offshore reference LCOE (~$181/MWh) and the turbine ≈⅓-of-CAPEX share: NREL/LBNL Cost of Wind Energy Review 2024 (NREL, 2025; also cited in Appendix A.5).

Supply chain & PV manufacturing

Table G.3.

Legal instruments

Table G.4.

Note on B.7 (high-speed rail): forced JVs and CRRC's ~90–95% domestic share are real, but there is no adjudicated WTO ruling — treated as analyst characterization, so no WTO citation is offered for that specific claim.

Capital & investors

Table G.5.

Grid champions (Siemens Energy, Prysmian, Nexans — and Hitachi Energy, which is Japanese-owned) are established industry facts; TenneT is the Dutch state-owned TSO.

Nuclear

Table G.6.

Borssele net capacity (~0.49 GW) and annual output (~3.8 TWh): Borssele — World Nuclear Association reactor database (WNA).

Precedent: HVDC corridors & failed mega-projects

Table G.7.

France–Spain Bay of Biscay link (~2 GW, in service ~2028): INELFE — Bay of Biscay project; EIB financing (also in F.2).

A note on method. Two kinds of figure are flagged rather than dressed up. First, our own extrapolations — the NL ~200 TWh mid-2030s demand figure and the system capacity-factor reconciliation — are labelled as such in Appendix A and are deliberately not attributed to a third party; they are the brief's own arithmetic, shown in full. Second, a small number of facts (Desertec's history, Viking Link's specifications) cite a high-quality aggregator where no single public primary document carries the specific datum; these are noted at the point of use. Every other load-bearing number traces to the primary source in the tables above.