Rewiring Europe: €800B Grid Upgrades by 2035

The grid spend is real, but most contractors will feel it as a conversion problem, not a lack of budget.

The European Commission cites €584bn of grid investment need by 2030. ENTSO-E's TYNDP 2024 adds €404bn of transmission investment need by 2050. Goldman Sachs Research puts transmission and distribution CAPEX alone at €1.2–1.4 trillion over 2026–2035, roughly double the prior decade. Put those horizons together and the 2026 to 2035 pipeline lands around €750bn to €850bn. The work will not flow evenly. It will arrive in starts and stops, tied to consents, design freeze and factory slots.

"Announced" is not the same as "notice to proceed", and that gap drives margin loss.

By late 2025, less than half of the decade's grid investment had reached financial close, according to a 2026 industry recruitment analysis. TenneT reports that roughly 60% of its Dutch onshore projects are already running 2.5 years behind schedule, mainly due to permitting and land acquisition. That shows up as tender churn, late scope changes, and clients asking contractors to hold key people without a buildable package. Treat early-stage awards as an option, not workload. Price the pre-construction drag and protect your resource plan.

Equipment lead times now set the programme, not the civils.

Large power transformer lead times have stretched to 36–48 months in most European markets, against 12–18 months before 2020. HV cable lead times sit at 2 to 3 years. RTE's response illustrates what this means in practice: in late 2024 France's TSO contracted Prysmian, Nexans, NKT, and Hellenic Cables for nearly €1bn of cable supply, reserving almost all remaining French production capacity through 2028. Contractors that wait for full approvals before ordering end up carrying resequencing cost and idle-gang overhead. Contractors that order early without clear client gates carry commercial risk. Neither outcome is priced into a standard tender.

Germany and the Netherlands set the pace for the whole European supply chain. Amprion reports €4.0bn invested in 2024 and a €36.4bn programme through 2029, with 1,300 km under construction at the same time. TenneT reports €14.8bn invested in 2025 across its Dutch and German grids, with a 2025–2029 forward plan of €108bn split €43bn for the Netherlands and €65bn for Germany. If you work in cables, substations, civils, or HV commissioning, these programmes pull factory slots and specialist crews forward. That pushes risk into other countries through lead times and price movement.

Germany dominates absolute spend. IDDRI estimates Germany's total grid investment requirement at €210bn by 2037, rising to €250bn by 2045. Four HVDC corridors account for most of that: SuedLink, SuedOstLink, A-Nord, and Ultranet. Combined corridor spend exceeds €50bn. Amprion's 2029 programme alone covers 1,300 km under concurrent construction. Offshore cluster substations for LanWin and BalWin add the largest single equipment category. Germany's TSOs plan a shift to 2 GW-class HVDC offshore converter platforms, with first commissioning from 2029 and 21 platforms intended by 2042. There is currently no converter platform production in Germany, and the only European shipyard fabricating foundation jackets for these structures is Dragados in Cadiz.

The UK programme is smaller in absolute terms but more compressed in time. National Grid's five-year plan commits £60bn to end-March 2029, nearly double the prior five years. Seventeen major 400 kV reinforcements sit under the ASTI framework. Eastern Green Links 1 to 4 and Sea Link account for the majority of HVDC civil and cable spend. National Grid's Great Grid Partnership commits £9bn across nine ASTI projects, with named contractors including Laing O'Rourke, Morgan Sindall Infrastructure, Morrison Energy Services, Murphy, and Omexom Taylor Woodrow. WSP describes the programme as up to 12 years of contracted work, with the activity peak around 2027 to 2028. That concentration of demand in a short window is exactly where supply chain strain shows up first.

France's RTE published a €100bn Strategic Plan through 2040, with roughly €75bn to €85bn falling inside this decade. The split matters for contractors: €24bn goes to renewing 23,500 km of existing line and 85,000 towers, €53bn to connecting new generation and large loads, and €16.5bn to reinforcing five priority corridors by upgrading existing routes rather than building new ones. RTE has already pre-committed nearly €1bn in underground cable supply to Nexans, Prysmian, NKT-SolidAl, and Hellenic Cables, covering 5,200 km from 90 kV to 400 kV through 2028. That deal reserves almost all remaining French production capacity through 2028. If your cable procurement sits outside that framework, you are competing for a very thin slice of European manufacturing.

The Netherlands carries a disproportionate load relative to its size. TenneT NL's 2025–2029 plan is €43bn, with IJmuiden Ver and Nederwiek offshore wind connections each requiring 2 GW HVDC converter stations as the dominant cost centres. Around 60% of Dutch onshore projects are running 2.5 years behind schedule due to permitting delays, and building a high-voltage substation takes 10 to 12 years end-to-end, with only around 30% of that being actual construction. A 2025 government acceleration package claims to cut lead times by up to 50%, with full impact expected from 2028. Contractors pricing Dutch work now should stress-test that assumption hard.

The Nordics carry €45bn to €55bn of programme across Statnett, Svenska Kraftnät, Fingrid, and Energinet. The Nordic Grid Development Perspective targets 420 kV AC backbone upgrades serving green steel and battery manufacturing corridors in Norway and Sweden. The scale mismatch in Finland is striking: Fingrid is handling production enquiries exceeding 400 GW against a 24.5 GW national system. Denmark paused new connections in March 2026, with more than 60 GW of new consumption queued against a 7.3 GW peak demand. These are not programme risks in the traditional sense. They are signals that the permitting and connection queue infrastructure cannot absorb the stated demand.

Iberia and Italy are smaller in absolute spend but show faster ramp rates. Spain and Portugal together represent around €27bn of programme, anchored by Red Eléctrica's interconnector build and a Bay of Biscay subsea link to France that will raise cross-border exchange capacity from 2.8 GW to 5.0 GW. Italy's Terna multi-year plan sits at around €28bn, including ELMED, a 600 MW interconnector from Sicily to Tunisia. Both markets face permitting dynamics that differ from northern Europe: civil code complexity, archaeological constraints in Italy, and bilateral regulatory coordination in Iberia add material delivery risk that does not show up in the headline CAPEX figures.

Poland and wider CEE represent the highest growth rate, not the largest absolute spend. Poland's Energy Law amendment was signed in April 2026, giving the legislative framework for accelerated grid investment. EU co-funding through the Connecting Europe Facility backs cross-border links including the 83 km, 330 kV Panevėžys to Aizkraukle reconstruction in Lithuania, with construction planned to start shortly and complete by end-2028. Ember's analysis across 20 reporting EU countries identifies Austria, Bulgaria, Latvia, Poland, Portugal, Romania, and Slovakia as the markets where available grid capacity can accommodate less than 10% of renewables planned by 2030. That gap is where the next wave of CAPEX will land, once permitting reforms take hold.

Grid work gets triggered by peak load in the wrong place, not annual demand. Eurelectric projects a further ~2,000 TWh of extra electricity demand by 2050 on top of today's ~2,500 TWh, but the spend shows up first as congestion, N-1 security failures, and reinforcement in a handful of hot zones. The load-growth vectors driving this are not evenly distributed across the map, data centres, industrial electrification, EV charging, and offshore wind landfall each concentrate demand differently, and that mismatch between where power is generated and where it is consumed is what forces transmission CAPEX faster than generation CAPEX (Eurelectric, 2024 via Modern Power Systems, 2024-11).

Axes are scored 1–10 on a relative basis across the five demand vectors. Bubble size is indicative of the transmission-level reinforcement scope each vector typically triggers.

Illustrative split based on published TSO and DSO reinforcement programme data. Offshore wind landfall and data centre connections push the largest share of scope to transmission voltage, driving the most expensive substation and corridor works.

Data centres move "building power" into "grid power" faster than any other load type, and they do it at transmission voltage, not distribution. JLL Research puts a new 50 MW connection at around 8 years in London and around 10 years in Amsterdam (Archdesk citing JLL, 2026-04). That forces land acquisition, easements, substation civils, and long-lead HV procurement to start before you have a final building layout. Across the EU, the IEA tracked 1,650 GW of solar and wind in advanced development stages awaiting grid connections in 2024, with data centres adding a separate surge of industrial-class connection requests on top (opna.earth, 2026-02). In Denmark alone, more than 60 GW of new consumption is queued against a 2024 peak demand of just 7.3 GW, enough pressure that Energinet temporarily paused new connections in March 2026 (Ember, 2026-04).

Residential electrification hits distribution first, then pushes work upstream. One household adding an EV and a heat pump can roughly double annual electricity use, and it lands on the same evening peak when networks are already tight (Podero, 2026-04). Eurelectric's modelling puts ~250 million EVs and ~250 million heat pumps on European distribution grids by 2050, which E.ON's CEO has translated into a need for one new connection every seven seconds of every working day between now and 2030 (Modern Power Systems, 2024-11). That is why "small" local schemes keep turning into primary substation upgrades, protection changes, and new upstream capacity works. Across 20 reporting EU countries, Ember finds a 120 GW gap between planned renewable expansion by 2030 and available grid hosting capacity, and the distribution layer accounts for at least 16 GW of that shortfall from rooftop solar alone (Ember, 2026-04).

Offshore wind landfall is the single biggest driver of new transmission corridors. Germany's WindSeeG targets 30 GW offshore by 2030, 40 GW by 2035, and 70 GW by 2045, requiring 21 new 2 GW-class HVDC offshore converter platforms in the German North Sea and Baltic by 2042 (taylorwessing.com, 2025-06). The UK's target is 50 GW by 2030, with National Grid stating the Great Grid Upgrade must deliver five times more electricity infrastructure over the next six years than has been built in the past 30 (AECOM, 2024-09). Offshore wind power lands at coast, but load centres are inland. Every gigawatt of generation that lands at Eemshaven or Dogger Bank needs a corridor to Birmingham, Frankfurt, or Paris. That north-to-south, coast-to-load mismatch is the structural reason why transmission CAPEX is outpacing generation CAPEX across the decade.

Protect margin by shifting your tender questions forward. Ask for connection status, required voltage level, and the client's responsibility split for substation land, easements, and outages. Across Austria, Bulgaria, Latvia, the Netherlands, Poland, Portugal, Romania, and Slovakia, Ember finds available grid capacity can accommodate less than 10% of renewables planned by 2030, meaning the reinforcement scope attached to any connection in those markets is almost certainly larger than the initial brief suggests (Ember, 2026-04). Programmes that track procurement status against the construction programme weekly will spot slippage early enough to hold delivery teams and recover change before it turns into delay damages.

Substation dates are set by factory slots, not site output. A 2026 CWIEME supplier review puts large power transformer procurement in Europe at 48 to 60 months. Even smaller distribution units are running 12 to 24 months in the same 2026 supply chain reporting. If your tender assumes plant will land "during civils", you will carry delay risk you can't control.

Component shortages are why "expedite" rarely works. A 2026 global supply chain report tracks EHV oil-impregnated paper bushings at 12 to 18 months, up from 4 to 6 months pre-2020. On-load tap changers sit at 8 to 12 months, up from 3 to 4. One late bushing can hold a finished transformer in the yard, then your energisation window slips and your site team burns cost on standby and re-sequencing.

Transformer prices have risen 70 to 95% since 2019. The main driver is grain-oriented electrical steel (GOES), a specialised steel used in transformer cores. GOES prices have doubled since 2019. Copper is up 50% over the same period. TenneT and National Grid both report that total substation CAPEX has effectively doubled since 2019, not because scope grew, but because the same equipment now costs twice as much to buy. If your cost plan still uses 2021 or 2022 benchmarks, your margin is already gone before you break ground.

HVDC converter equipment is the longest lead item on any transmission project. Converter transformers and valve assemblies from Siemens Energy, Hitachi Energy, and GE Vernova are now running at 48 months or more from order to delivery. The bottleneck is the same GOES shortage, compounded by the engineering complexity of each unit. No two HVDC converter stations are identical, so there is no stock to draw from. Every unit is built to order, and the order books at all three major OEMs are full.

The TSOs running the largest programmes have already responded to this by changing how they buy. Their procurement structures now set the standard that EPCs and specialist contractors on these schemes are expected to work within. Three approaches are worth understanding in detail.

Slot reservation frameworks. TenneT's 2GW HVDC programme, running from 2023 to 2031, uses multi-year framework agreements with NKT and Prysmian to reserve cable and converter capacity years before individual project orders are placed. The framework commits volume. In return, the suppliers hold production slots. This is not a traditional call-off contract. It is a capacity purchase. If you are an EPC working under TenneT on this programme, your procurement strategy is already partially set by the framework. You need to understand which slots are reserved, which items sit outside the framework, and where your exposure starts.

Early construction funding. National Grid's ASTI (Accelerated Strategic Transmission Investment) framework allows up to 20% of project value to be spent on early procurement and enabling works before final planning consent is granted. This is a deliberate policy choice to accept pre-consent spend in exchange for shorter post-consent delivery timelines. For the EPC, this changes the risk profile. You are ordering long-lead equipment before the project is fully consented. The client carries that risk, not you, but only if the contract is written that way. Check the allocation of pre-consent procurement risk explicitly before signing.

Standardised design. TenneT's 2GW programme uses a standardised 525 kV HVDC blueprint across multiple converter stations. The same protection architecture, the same harmonic filter specification, the same civil envelope. This cuts engineering lead time by 12 to 18 months per station compared to a bespoke design. It also means the OEM can manufacture on a repeating template. For the EPC, the lesson is direct: push for design freeze early, and resist any client-driven scope change after equipment orders are placed. Every change after order entry costs more than it appears to, because redesign and retesting time is not priced into standard LD structures.

Indexation and liquidated damages (LDs) need to be structured to reflect current market conditions, not 2019 norms. For materials, tie price adjustment to the London Metal Exchange (LME) copper and aluminium indices, plus a dedicated index for electrical steel. A fixed-price contract for a 400 kV transformer ordered today exposes the supplier to commodity risk they cannot hedge. They will price that risk in. If you want a competitive number, share the risk through a transparent indexation formula. On LDs, the market cap is now 10% of contract value on most European T&D packages. Anything higher is pushing suppliers into insolvency risk, and an insolvent transformer manufacturer does not accelerate your programme. Force majeure carve-outs for systemic supply chain disruption are also standard in current OEM contracts. If your subcontract does not mirror this, you are holding liability the supply chain has already passed upstream.

HVDC nodes amplify the same risk. A 2025 HVDC market review puts converter stations at $120m to $300m per terminal, and SINTEF reports up to a 2x price spread between projects that look similar on paper. The driver is late technical change, not civils. If protection and control architecture, harmonic filters, or gas specs move after ordering, you can lose months in redesign and retesting with no extra output on site.

Archdesk teams see substation packages perform best when the interface register is treated like a cost item, not a document list. Put owners next to each OEM deliverable, set dates that match factory reality, and price change control around it. That is how you stop equipment drift turning into unpriced site time.

Offshore transmission slips are driven by physical queues, not design progress. A 2024 offshore cable supply review says tier-one HVDC cable makers are booked into the late 2020s. The same review flags fewer than 60 vessels worldwide that can lay, repair, or maintain subsea power cables. That is why dates move after "award" and why contractors get stuck holding teams on standby.

The scale of the problem between 2026 and 2035 is larger than most project teams have priced for. Annual subsea cable installation volumes are projected to reach 18,173 km by 2028. That is a 9.5-fold increase on 2020 volumes. The UK and Germany alone need over 18,000 km of new cable networks in this period. Demand at that scale overwhelms what the four tier-one manufacturers, Prysmian, Nexans, NKT, and Hellenic Cables, can physically produce. Between them, Prysmian holds roughly 18% of the market and Nexans 14%. Manufacturing capacity for 525 kV HVDC cable is effectively fully booked through 2028 and into 2029. Adding a new production line takes three to four years. No amount of commercial pressure closes that gap before the 2030 offshore wind targets arrive.

The shift to multi-terminal HVDC nodes is concentrating delivery risk further. TenneT's 2 GW programme alone will deploy 14 HVDC offshore systems, totalling 28 GW, by 2031. All are designed for multi-terminal readiness. Germany's NordOstLink is developing a 525 kV multi-terminal hub scheduled for 2032. Each of these nodes requires converter stations costing between €120m and €300m per terminal. That figure has risen sharply. Scarcity has pushed a 20 to 30% pricing premium onto core components across the board. Any EPC contractor pricing a converter station package today on 2023 cost data is starting the job in a loss position before ground is broken.

The vessel constraint is the part of the programme that cannot be bought out of trouble. Nine new cable laying vessels are expected to enter service by 2026, including the Nexans Electra. That sounds like relief. It is not enough. The gap between supply and demand for cable laying vessels peaks between 2028 and 2030, which is precisely when the bulk of North Sea and Baltic HVDC connections need to be pulled through. A vessel window missed in 2028 does not get recovered in the same season. The recovery plan is re-sequencing into 2029, which cascades into every onshore civils package, substation energisation, and grid connection milestone downstream.

Installation damage is the margin killer because it sends you back into the same vessel queue. An IMCA statistic cited in a 2024 offshore cable review links 9 out of 10 subsea cable insurance claims to installation damage. The commercial point is simple. A rushed weather window can turn one jointing issue into months of rework and liquidated damages arguments across multiple interfaces.

TSOs are responding with framework agreements and early capacity reservations. TenneT and other major TSOs are locking in manufacturing slots years ahead of financial close. For EPC contractors, this changes the commercial dynamic. If the TSO has already placed a framework with Prysmian or NKT, the contractor's room to manage supply chain risk independently shrinks. The contract risk does not disappear. It shifts. Programme liability for any mismatch between the reserved slot and the actual construction sequence lands with the EPC team.

Protect margin by forcing slot evidence into the tender baseline. Ask for two items before you commit to dates and prelims. Ask for a named factory slot for cable and converters. Ask for a named vessel window and the rules for weather and standby. Archdesk sees fewer disputes on grid packages where these are treated as contract data, not assumptions.

Read an EPC league table as a delivery-stretch test, not a size contest. The best quick screen is backlog divided by annual delivery run-rate. A firm can carry a big order book and still be high risk if the work is outage-led, brownfield, and tied to long-lead plant. Wood Mackenzie reported average lead times of 151 weeks for high-voltage circuit breakers in Q4 2023, and 127 weeks for power transformers, up from 78 weeks in Q1 2022. Those queues turn "capacity" into a planning problem, not a headcount problem.

The scarcest "EPC capacity" sits at the interfaces. HVDC converter stations and brownfield substation outages fail first because the programme is locked to factory slots, outages, and outage access rules. If your package depends on long-lead switchgear and transformers, the commercial risk moves upstream. Late approvals, late AFC drawings, or slow vendor data reviews land straight on your critical path.

The top five firms in this market have already absorbed the majority of framework positions. Bouygues Equans carries a disclosed backlog of €25.4bn across its energy and multi-technical businesses. Saipem sits at €31.5bn total, though a large share of that is subsea and offshore platform work. ACS Group's Cobra IS business runs at roughly €20bn, concentrated in Iberian T&D and offshore HVDC packages including the IJmuiden Ver Alpha cable link off the Dutch coast. Vinci Energies, operating its T&D arm as Omexom, is above €18bn. Balfour Beatty is the dominant UK-focused player at roughly £9bn total backlog, with a lead position on National Grid's ASTI programme and the Electricity Transmission Partnership (ETP) covering the North East of England.

Permitting time is lost in rework loops, not slow progress. The time drain comes from route changes, EIA re-scopes, land access resets, and appeals that send you back to earlier gates. Price those gates as paid tasks and keep your delivery team light until the gate is cleared.

The 8-12 year headline breaks down into three distinct blocks. Front-end permitting, covering route selection and Environmental Impact Assessment (EIA), consumes 3-5 years on its own. Procurement and Final Investment Decision (FID) add another 1-2 years. Execution, covering civil works, cable laying, and converter station installation, takes a further 4-5 years. That sequence is why SuedLink in Germany has stretched to 15 years: the decision to shift from overhead line to underground cable reset the EIA clock entirely. Viking Link, run by National Grid and Energinet, completed in roughly 10 years from first proposal in 2013 to energisation in December 2023. That is a good outcome for a project of this scale, and the programme decisions made years before construction started are the main reason it held.

The construction phase is not where the time goes, but it is where the most recoverable time sits. Two bottlenecks define the execution window. First, the civil-to-electrical interface at converter stations, where civil handover dates dictate when Siemens Energy or Hitachi Energy can start installing transformers and HVDC valves. Both manufacturers are currently quoting 3-4 year lead times for that equipment. Second, cable manufacturing slots. Miss your slot at NKT, Prysmian, or Nexans and you are looking at a 3-month vessel delay at minimum, often longer. Those are not planning failures. They are sequencing failures that happen when procurement starts after FID rather than before it.

There are five specific levers that take 12-24 months out of the schedule. Each one requires a commercial decision, not just a programme management decision.

• Early Contractor Involvement and frameworks. Transactional tenders start procurement after FID. Framework agreements do not. TenneT's 2GW offshore programme, set up with GE/McDermott and Hitachi/Petrofac, allowed both contractors to reserve manufacturing slots and start detailed engineering 12 months before FID was confirmed. That single move pulls roughly a year out of the post-FID programme without touching the permitting timeline.
• Standardised converter station designs. Bespoke engineering on every project adds 12-18 months to the design-to-build cycle. TenneT estimates its standardised 2GW land station design cuts that cycle by up to 18 months compared to a fully bespoke approach. The saving comes from reusing validated designs, not from cutting corners on engineering.
• Enabling works under preliminary agreements. Full project consent is not needed for site levelling, access roads, and horizontal directional drilling (HDD) at cable landfalls. Prysmian and Balfour Beatty ran enabling works on Viking Link under preliminary agreements before the main consent was issued. The result was that installation vessels could start laying cable immediately on permit receipt, rather than waiting months for site readiness.
• 4D BIM to de-conflict converter station design. A converter station carries more than 3,900 individual design requirements. Clash detection done on site adds 6-9 months to commissioning. WSP and TenneT have used 4D BIM (three-dimensional models integrated with the construction schedule) to find clashes during design, not during installation. That is recoverable time that most programmes still lose.
• Vertical integration of cable supply and installation. Contractors that control both cable manufacturing and installation vessels can align factory output with weather-dependent installation windows. NKT and Nexans both operate this model. A contractor without that integration is dependent on a vessel schedule that was built around someone else's priorities. Miss a 3-month installation window in the North Sea and you are waiting for the next one.

Connection approvals are a programme in their own right. Bundesnetzagentur figures cited in a 2025 grid connections analysis put large renewables at 24 to 36 months to secure a grid connection. That can run in parallel with planning, but it still drives design freezes and outage planning later. Build your schedule around "permission to freeze design" rather than "planning submitted".

Appeals create stop-start costs that your tender often doesn't cover. Standby labour, remobilisation, and re-sequencing hit overhead first, then they hit productivity. Put remobilisation rates, standby rules, and change control around access dates into the contract, even on framework call-offs. If you can't get that, keep critical resources pooled across jobs so you can swing crews without bleeding prelims.

Archdesk sees better outcomes when teams run one programme view that links consent gates, design freeze dates, and procurement releases. Treat permitting as a live commercial programme, not a waiting room between tender and construction.

Grid work is now lost at the back end, not at “site progress”. The packages keep moving until you hit protection and control, testing, and outage access. TSOs flagged workforce shortages in 2025, but the commercial hit for contractors is simple: you reach mechanical completion, then sit waiting for specialist people and an outage slot. That is where prelims burn and LD risk builds, even if civils output is fine.

Transformer and switchgear inflation hits margin through cashflow first. A 2025 transformer cost breakdown puts raw materials at 70% to 80% of manufacturing cost. That turns your risk into factory stage payments and exposure to index moves before you have earned value on site. Price protection is not a “nice to have” on these jobs. It is the difference between a solvent package and a working-capital trap.

Protect margin by treating commissioning as a booked resource, like a crane or a road closure. Name the constraint in your tender clarifications: outage access, protection and control labour, test windows, and client-ready documentation. Build your programme around “ready for energisation” gates, not “installed” gates. Archdesk teams track progress at asset level, bay by bay and circuit by circuit, so commercial and ops see the stall early enough to re-sequence labour before the delay turns into unrecovered prelims.