2,061 GW of power projects sit in US connection queues and transformers now take four years to deliver. The real constraint on electrification is the grid.
The Queue Is the Transition
Why the binding constraint moved from generation to connection
At the last count, 2,061 GW of generation and storage were actively waiting for permission to connect to the American grid — a 10% fall on the year before, achieved mostly through cancellations rather than through construction ✓ Established [2]. Worldwide, the International Energy Agency counts more than 2,500 GW of renewable, storage and large-load projects stalled in connection queues ✓ Established [4]. The scarce input in the energy transition is no longer the turbine or the panel. It is the wire, the transformer and the permit.
The composition of the American interconnection queue makes the point better than any argument about policy ambition. At the end of 2024 roughly 10,300 projects were actively seeking a grid connection in the United States, representing about 2,300 GW: 1,400 GW of generation and 890 GW of storage ✓ Established [1]. Solar accounted for 773 GW of that total, battery storage for 749 GW, and natural gas — the fastest-growing category — for 253 GW after an 86% jump in a single year ✓ Established [1]. Whatever else is true of the transition, it is not suffering from a shortage of developers willing to build. It is suffering from a shortage of places to plug in.
The queue then did something it had not done in at least a decade: it shrank. Active capacity fell 10% to 2,061 GW ✓ Established [2]. That figure is routinely misread as evidence that reform is working. The Berkeley Lab data say otherwise. Of all the capacity that requested interconnection between 2000 and 2020, only about 13% had reached commercial operation, while roughly 75% had been withdrawn ✓ Established [3]. A queue that empties through abandonment is not a queue that is clearing. It is a queue that is failing at a faster rate, which produces the same headline number and none of the electricity.
Duration is the more revealing statistic. For projects reaching commercial operation between 2000 and 2007, the median time from interconnection request to switch-on was under two years. For those built between 2018 and 2024, it exceeded four years, and the cohort that came online in 2024 had spent an average of 55 months in the process ✓ Established [3]. Nothing about a solar farm became four times harder to build in that period. What changed was the administrative and physical apparatus required to attach it to a network that was planned for a different century and a different load shape.
Until recently the queue was a generator problem. It is now two-sided. The queue of large loads seeking connection in Texas — overwhelmingly data centres and electrified industry — has nearly quadrupled in a single year ◈ Strong Evidence [30]. Grid-supplied demand from data centres was projected to rise 22% in 2025 and to almost triple by 2030, concentrated in a handful of regions rather than spread across the network ◈ Strong Evidence [26]. The same interconnection machinery that took four years to attach a wind farm is now being asked to attach gigawatt-scale consumers, and to do it in the eighteen-month cycles on which computing capital is deployed.
The money has been going to the wrong half of the system. Roughly 400 billion dollars a year is spent on electricity grids worldwide, and the IEA calculates that annual grid investment must rise by about half — to more than 600 billion dollars a year — by 2030 to match the generation already being financed and the climate targets already legislated ✓ Established [5]. Transmission specifically drew 140 billion dollars in 2023, and must exceed 200 billion a year by the mid-2030s, reaching 250 to 300 billion in scenarios that meet national emissions goals ✓ Established [6]. Generation investment has run ahead of the network for over a decade.
The 2,061 GW sitting in American queues ✓ Established [2] and the 2,500 GW stalled globally ✓ Established [4] are not forecasts or aspirations — they are filed, fee-paying applications. The constraint they describe is not the cost of electricity generation, which has fallen, but the cost and duration of network access, which has risen. Grid investment at roughly 400 billion dollars a year is running at about two-thirds of the level the IEA judges necessary by 2030 [5].
Part of the delay is self-reinforcing. Interconnection studies are run in clusters, and the cost of network upgrades is allocated among the projects in each cluster. When a project withdraws — as three in four eventually do [3] — the remaining projects inherit its share of the upgrade bill, which triggers further withdrawals and a restudy. The Federal Energy Regulatory Commission designed Order No. 2023 explicitly to break this cascade, replacing serial first come, first served studies with first ready, first served cluster studies backed by firm deadlines, financial deposits and site control requirements ✓ Established [24].
What the queue data establish is narrower than the rhetoric around them but more consequential. The rate limiter on electrification in the United States, Britain, Germany and increasingly India is not the price of a megawatt-hour. It is the serial dependency of permits, studies, network upgrades and physical equipment that stands between a financed project and an energised one. Each of those four links has lengthened independently over the past five years. The sections that follow take them in order: the permit, the transformer, the copper and the workforce — and then ask what the reforms have actually delivered.
Ten Years for a Power Line
The permitting stack that no single authority controls
Completing an environmental impact statement in the United States takes more than four years on average, and about a quarter of federal reviews under the National Environmental Policy Act take six years or longer ✓ Established [15]. The SunZia transmission project required roughly 17 years of permitting; a line serving the Pine Ridge Reservation took 20 ✓ Established [16]. Construction was never the slow part.
A long-distance transmission line is the most legally exposed object in the energy system. It crosses state lines, federal land, tribal land, private easements, watersheds and viewsheds, and it confers most of its benefit on people who do not live along it. Generation, by contrast, is a point object on a single parcel under a single jurisdiction. That asymmetry — diffuse benefit, concentrated cost, multiplied by the number of authorities with a veto — explains why permitting time rather than construction time dominates the schedule of every major line in the West.
The federal review numbers are unambiguous. An environmental impact statement now averages more than four years to complete, and roughly one in four such reviews runs to six years or beyond ✓ Established [15]. Those figures measure only the document. They exclude the state siting proceedings that run in parallel, the certificate of public convenience and necessity required in each state crossed, the eminent domain proceedings that follow a certificate, and the litigation that follows those. A line that touches five states can face five independent opportunities for a project-ending refusal.
The individual cases are worse than the averages. SunZia, a high-voltage line intended to move wind power from New Mexico to Arizona, spent roughly 17 years in permitting before construction; a line serving the Pine Ridge Reservation took two decades ✓ Established [16]. Neither delay was caused by engineering uncertainty. Both reflect a legal architecture in which any single objecting party can impose years of process cost on a project whose economics were calculated on a construction schedule of two to three years.
The asymmetry between generation permitting and transmission permitting has consequences developers act on. If attaching to an existing corridor takes four years and building a new corridor takes twelve, the rational response is to crowd every project onto the existing network and accept the congestion that follows. That is precisely the pattern the interconnection data show: 2,061 GW of applications concentrated on a network whose transmission investment is running at roughly 140 billion dollars a year against a mid-2030s requirement above 200 billion ✓ Established [2] ✓ Established [6].
Europe has attempted a legal fix rather than an administrative one. The European Grids Package, presented in December 2025, revises the trans-European energy networks regulation to introduce tacit approval: a project is deemed permitted if the responsible authority fails to respond within the statutory window, capped at two years for most projects and three for the most complex ✓ Established [23]. The Commission has estimated that 584 billion euros of grid investment is required before the end of the decade, and has proposed raising the Connecting Europe Facility for Energy from 5.84 billion euros for 2021-2027 to 29.91 billion for 2028-2034 ✓ Established [23].
Environmental impact statements average more than four years, with a quarter exceeding six [15]; individual interstate lines have required 17 and 20 years of review [16]. Against that, reconductoring an existing corridor with advanced conductors — which requires no new right of way — can be completed in months to a year, against 10 to 15 years for a new line ✓ Established [22]. The variable that dominates the schedule is legal exposure, not engineering difficulty.
The tacit-approval mechanism is a genuine structural innovation, but it addresses only the administrative clock. It does not shorten the ecological survey season, resolve a landowner refusal, or manufacture a transformer. Deadlines bind the authority that grants permission; they do not bind the physical world in which the line has to be built. That distinction — between the process constraint and the material constraint — is the one that most commentary on grid reform collapses, and it determines which interventions can plausibly deliver capacity before 2030.
A developer who wins every approval in record time then joins a second, invisible queue. Large power transformers take up to four years to procure and high-voltage direct current cables can take more than five [6]. A permit granted in 2026 for a line requiring bespoke high-voltage equipment does not become electricity before 2031. Permitting reform that ignores the equipment queue simply moves the bottleneck one step downstream and declares victory.
The honest reading of the permitting evidence is that it is a necessary but insufficient reform. Cutting a seventeen-year approval to a five-year approval is a large gain in a system where the alternative is abandonment. It is not, on its own, a route to the network expansion the 2030 targets assume, because the moment the legal clock stops the industrial clock starts — and the industrial clock, as the next section shows, has been running slower every year since 2021.
The Four-Year Transformer
When the constraint stops being legal and becomes physical
Cables now take two to three years to procure and large power transformers up to four — almost double the lead times of 2021 — while high-voltage direct current cables can exceed five years ✓ Established [6]. In the United States, standard power transformers averaged about 128 weeks in 2026 and generator step-up units about 144 ✓ Established [8]. No permitting reform reaches this constraint.
The IEA surveyed the transmission supply chain and found a market that has stopped clearing. Cables take two to three years to procure. Large power transformers take up to four. High-voltage direct current cables — the components on which every offshore wind farm and every long-distance link depends — can take more than five years ✓ Established [6]. Compared with 2021, procurement times have roughly doubled. These are not congestion effects that resolve with a quarter of soft demand; they reflect order books filled years forward against manufacturing capacity that was sized for a replacement market, not a build-out.
The American numbers are more granular and no better. Standard power transformer lead times averaged about 128 weeks in 2026, and generator step-up units — the machines that connect a power plant to the transmission network — about 144 weeks ✓ Established [8]. A utility procurement officer placing a generator step-up order in mid-2026 is quoted delivery in 2029. That single fact reorders every schedule downstream of it: the interconnection agreement, the financing close, the power purchase agreement and the date on which a state expects the capacity to appear in its resource adequacy assessment.
The IEA supply chain assessment published in February 2025 found cable procurement at two to three years and large power transformers at up to four — approximately double the 2021 benchmark — with HVDC cables beyond five years ✓ Established [6]. Because a new cable manufacturing facility itself takes three to four years to build [6], a demand signal sent in 2026 adds physical capacity only towards the end of the decade.
Price has moved with delivery time, which is what a genuine capacity constraint looks like. Distribution transformer prices are up 78 to 95% since 2019, power transformers about 77% and generator step-up units about 45% ✓ Established [8]. The input side explains part of it: grain-oriented electrical steel, the specialised core material without which no transformer functions, has roughly doubled in price since January 2020 ✓ Established [9]. Cost inflation of this order does not merely raise the bill. It changes which network upgrades clear a regulator cost-benefit test, and therefore which projects are told their upgrade is uneconomic.
Demand explains the rest. Orders for generator step-up transformers rose 274% between 2019 and 2025, and demand for substation transformers rose 116% over the same period ✓ Established [9]. A manufacturing base that had spent two decades serving a slow replacement cycle was asked, within five years, to serve a construction cycle almost four times larger. Winding capacity and the skilled labour that operates it, not steel or copper in the abstract, became the specific point at which the industry stopped being able to say yes.
Not every analyst accepts that the transformer crisis is purely a supply failure. Trade analysis argues a meaningful share is self-inflicted: utilities specify bespoke units where standard designs would serve, order in stop-start cycles that prevent manufacturers from investing in capacity, and treat inventory as a balance sheet liability rather than a reliability asset ⚖ Contested [9]. If that reading is right, part of the four-year wait is a procurement culture, and can be shortened without a single new factory.
The manufacturing response is real but slow. New cable plants were not expected to come online before 2026, and building one takes three to four years from decision to production ✓ Established [6]. Manufacturers have been reluctant to add capacity against order books they suspect are inflated by developers placing speculative reservations — the same behaviour that fills interconnection queues with projects that never get built. The industry is caught in a symmetrical credibility problem: buyers do not believe delivery dates, and sellers do not believe order volumes.
The consequence for the queue is direct and underappreciated. A project can clear every study, sign an interconnection agreement, obtain every permit and still be unable to energise, because the specific transformer that connects it does not exist and cannot be bought inside four years. Queue statistics do not capture this. A project waiting for steel and a project waiting for a study look identical in the data, which is one reason reform measured in queue throughput can look successful while delivered capacity does not move.
Copper, Steel and Skilled Hands
The inputs that cannot be legislated into existence
Copper set an all-time high of 14,527.50 dollars a tonne on the London Metal Exchange in January 2026, driven by grid and data-centre demand ✓ Established [12]. The IEA still projects a copper supply gap of about 25% by 2035, narrowed from 30% as more mines advance but not closed ✓ Established [10]. Around 8 million people work on grids worldwide, and the workforce must grow by 1.5 million by 2030 ◈ Strong Evidence [6].
Copper is the material through which the entire electrification argument has to pass. Transmission lines, distribution networks, transformers, motors and the internal wiring of every data centre all consume it, and no substitute matches it across all those uses. The IEA judges that the projected copper supply gap by 2035 has narrowed from around 30% to around 25% as more projects advance, but that deficits will persist through 2035 because ore grades are falling, capital costs are rising and discoveries have slowed ✓ Established [10]. A quarter of projected demand has no identified mine attached to it.
The price has responded the way a market prices a decade-long shortage. Copper reached an all-time high of 14,527.50 dollars a tonne on 29 January 2026 and was still trading near 14,455 dollars in August ✓ Established [12]. Goldman Sachs Research expects prices to ease somewhat from those records as supply responds, while treating the underlying electrification demand as structural rather than cyclical ◈ Strong Evidence [11]. For grid planners the distinction matters less than the level: every network upgrade budgeted at 2019 metal prices now needs reapproval, and every reapproval is another year.
Labour is the constraint that receives least attention and binds soonest. About 8 million people work building, maintaining and operating electricity grids worldwide, and the IEA estimates the workforce must grow by 1.5 million by 2030 under existing policy settings — an increase of nearly one fifth in five years ◈ Strong Evidence [6]. Linemen, cable jointers, high-voltage commissioning engineers and transformer winders are trained over years, not quarters. No permitting statute, tariff or subsidy shortens an apprenticeship, and the countries competing for these workers are competing with each other.
The skills constraint sits inside the equipment constraint. The transformer backlog is frequently described as a steel problem, but the specific limiting step in most factories is coil winding — a manual, certification-heavy operation that cannot be scaled by buying a machine. That is why demand growth of 274% for generator step-up units between 2019 and 2025 translated into lead times rather than output ✓ Established [9]. Capacity expansion in this industry means hiring and training before it means capital expenditure, which is why announced investments do not show up as deliveries for three to four years.
We must invest in grids today or face gridlock tomorrow.
— Fatih Birol, Executive Director, International Energy Agency, October 2023The scale of the physical task is easy to lose in the financial framing. The IEA calculated that 80 million kilometres of power lines must be added or replaced globally by 2040 for governments to meet their stated climate and energy commitments — roughly the length of the existing global network, rebuilt inside two decades ✓ Established [7]. That is the demand signal against which four-year transformer waits, a 25% copper gap and a 1.5 million shortfall in grid workers have to be read. Each constraint is individually manageable. Simultaneously, they define a build rate.
There is one credible way to buy capacity without buying proportional quantities of copper, permits and time. Replacing conventional conductors with advanced composite-core designs inside existing rights of way could roughly double American transmission capacity by 2035, cut wholesale electricity costs by 3 to 4%, and save about 85 billion dollars in system costs by 2035 and 180 billion by 2050 ◈ Strong Evidence [22]. Crucially, reconductoring is measured in months to a year rather than the 10 to 15 years a new line requires ✓ Established [22] — it substitutes materials science for legal process.
Two Years in China, a Decade in the West
What planning a grid into existence actually buys
Approval to energisation in roughly two years is the norm for Chinese ultra-high-voltage projects, not the exception ◈ Strong Evidence [13]. The country expanded its ultra-high-voltage direct current network from about 28,000 km to more than 40,000 km in five years and now operates 45 such projects ◈ Strong Evidence [13]. State Grid Corporation of China has committed 4 trillion yuan — about 580 billion dollars — for 2026 to 2030 ✓ Established [14].
The comparison that matters is not ideological but procedural. During the five years to 2025, China grew its ultra-high-voltage direct current network from roughly 28,000 km to more than 40,000 km, and by late 2025 was operating 45 ultra-high-voltage projects — 19 alternating current lines and at least 23 direct current links ◈ Strong Evidence [13]. The Changji-Guquan link alone runs 3,293 km at 1,100 kV and carries 12,000 MW ◈ Strong Evidence [13]. No Western jurisdiction has built anything at that voltage, length or pace in the same period.
The financial commitment behind the next phase is larger still. State Grid announced in January 2026 that it will invest 4 trillion yuan — approximately 580 billion dollars — across the 2026 to 2030 planning period, a 40% increase on the preceding five years, and will commission 15 new ultra-high-voltage lines ✓ Established [14]. For scale, that single utility commitment approaches the 584 billion euros the European Commission estimates the entire European Union needs to invest in grids before the end of the decade ✓ Established [23].
The speed is a governance artefact, not an engineering secret. Chinese transmission projects move from approval to energisation in roughly two years because land acquisition, route selection, generation siting and industrial demand are decided by the same planning apparatus, on the same timetable ◈ Strong Evidence [13]. When a data centre cluster or industrial zone is designated, the grid to serve it is planned and built in parallel rather than applied for afterwards. The trade-off is explicit: the process that produces the speed is the same one that removes the objection rights Western permitting exists to protect.
The rest of the world is converging on the same diagnosis from different starting points. India has identified a transmission investment opportunity above 9.15 lakh crore rupees by 2032, with inter-regional transfer capacity planned to rise from 119 GW to 168 GW ✓ Established [29], and analysts describe transmission rather than generation as the invisible barrier to its clean energy growth ◈ Strong Evidence [28]. Brazil, which runs its expansion through concession auctions, will tender close to 4,500 km of line and more than 25 billion reais of investment across two 2026 auctions ✓ Established [32].
Projects do not wait in line; they are planned into existence.
— ChinaTalk, analysis of China transmission policy, 2026It would be a mistake to read the Chinese record as proof that authoritarian planning is efficient. The same system has produced substantial stranded assets and curtailment in its western provinces, and the two-year clock reflects a legal environment in which affected parties have no practical means of objection. What the record does prove is narrower and harder to dismiss: the ten-to-twenty-year Western timeline is a political and legal choice, not a physical property of high-voltage infrastructure. Lines can be built in two years. The question is what is surrendered to build them that way.
The instructive comparison is therefore between Western jurisdictions rather than across systems. Britain cut its connections queue from more than 700 GW to 381.5 GW in a single administrative decision ✓ Established [17]. The European Union proposed statutory deadlines with automatic approval on expiry ✓ Established [23]. Texas connects generation faster than any other American market by declining to guarantee it firm network capacity ✓ Established [20]. Each of these is a choice about who bears risk — the developer, the consumer or the system — and none of them requires a different political system.
What Congestion Costs
Curtailment, redispatch and the bill that reaches consumers
German redispatch and congestion management cost roughly 2.7 billion euros in 2025, with grid management costs approaching 2.2 billion in the first three quarters alone ✓ Established [18]. Those costs are recovered through network charges and land on end consumers. Congestion is not a theoretical inefficiency — it is an invoice that arrives every year the network stays smaller than the generation attached to it.
When generation is connected faster than the network that moves it, the system pays to keep the two in balance. Germany, which built wind capacity in the north and industrial demand in the south, spends the difference on redispatch: paying generators to reduce output on the constrained side and other generators to increase it on the other. That bill reached roughly 2.7 billion euros in 2025, with grid management costs approaching 2.2 billion in the first three quarters according to the federal network agency ✓ Established [18].
The direction of travel in 2025 was mildly encouraging and structurally unremarkable. Curtailment compensation payments fell 22% year-on-year ✓ Established [18], reflecting grid reinforcement, milder congestion conditions and changes in market behaviour. But the underlying arithmetic is unchanged: the payments exist because renewable output cannot reach the load, and they are recovered through network charges paid by households and industry. The public debate treats grid investment as a cost and congestion as an act of nature. The accounts show both as line items in the same budget.
Congestion management in a single European country cost about 2.7 billion euros in one year [18]. That is money spent not to build anything, but to compensate for what was not built. Every year the network stays undersized, consumers pay a recurring charge that buys no asset, no capacity and no reliability. The political framing of grid investment as an expensive option is inverted: the expensive option is the one currently in force.
Texas illustrates the same trade-off from the opposite direction. Under the connect-and-manage model, generators are attached to the network quickly and accept the risk that their output will be curtailed when the network cannot carry it. The result is the fastest interconnection process in the United States — 14.2 GW brought online across 2021 and 2022 against 5.6 GW in the largest eastern market over the same period ✓ Established [20] — and a structurally higher level of curtailment, because proactive regional transmission planning was the thing traded away.
The completion statistics quantify the difference. Average interconnection waits run about 20 months in Texas against about 40 months in the largest eastern market, and of projects that entered screening by 2020, 40% in Texas reached an interconnection agreement or operation against 24% in the east ◈ Strong Evidence [21]. Neither model produces an adequate network. One produces access with congestion risk transferred to generators; the other produces firm access for a smaller number of projects and defers the rest.
Congestion is the routine cost. The tail risk is operational. On 28 April 2025 the Iberian Peninsula lost synchronisation with the continental European system at 12:33:19, and the Spanish and Portuguese grids had collapsed by 12:33:24 ✓ Established [19]. The expert panel convened by the European transmission operators identified insufficient voltage and reactive power control as the probable main cause, explicitly ruling out low inertia and inter-area oscillations as root causes, and issued 23 recommendations, 13 of them tied directly to those causes ✓ Established [19].
The interpretation of that event became a proxy for the wider argument. Industry analysis emphasised that the findings point to grid resilience, voltage control and operational coordination rather than to renewable penetration as such ⚖ Contested [31]. The distinction is not cosmetic. If the diagnosis is renewables, the remedy is to slow deployment; if the diagnosis is grid capability, the remedy is to spend on reactive power control, system services and network capacity. The formal finding supports the second reading, and the second reading costs money the first does not.
The Reform Scoreboard
Which interventions have actually moved capacity
The reforms of the past three years — cluster studies in the United States, a rebuilt queue in Britain, tacit approval in the European Union — share a common design: raise the cost of entering the queue and impose deadlines on the study process ✓ Established [24] ✓ Established [17] ✓ Established [23]. All of them ration access more efficiently. None of them adds a transformer, a kilometre of line or a trained cable jointer.
Order No. 2023, issued in July 2023, was the largest change to American interconnection rules in two decades. It replaced serial first come, first served studies with first ready, first served cluster studies, imposed firm study deadlines backed by financial penalties on transmission providers, and required developers to demonstrate commercial readiness and site control before entering ✓ Established [24]. The theory is sound: if 72% of projects entering queues between 1999 and 2018 were later withdrawn, largely because entry was cheap ✓ Established [25], then pricing entry properly should leave a queue that means something.
Britain applied the same logic more aggressively and produced the clearest natural experiment available. Ofgem approved the TMO4+ methodology in April 2025, converting the queue from first come, first served to first ready, first connected, and in December 2025 the system operator published a reformed queue of 381.5 GW — 283 GW of generation and storage plus 99 GW of transmission-connected demand — down from more than 700 GW ✓ Established [17]. Around 153 GW of battery storage projects were removed outright ✓ Established [17].
| Constraint | Severity | Assessment |
|---|---|---|
| Equipment lead times | Transformers at up to four years and HVDC cables beyond five, with new factories themselves taking three to four years to build [6] [8]. No policy instrument shortens this before 2029. | |
| Permitting and litigation | Environmental reviews averaging over four years, individual lines at 17 to 20 years [15] [16]. Reform is under way but applies mainly to projects not yet started. | |
| Copper and raw materials | A projected 25% supply gap by 2035 and record prices [10] [12]. Substitution and reconductoring mitigate but do not eliminate the exposure. | |
| Skilled workforce | A workforce of 8 million that must grow by 1.5 million by 2030 [6]. Training cycles run years and every jurisdiction is recruiting from the same pool. | |
| Queue reform without build-out | Faster access to an unchanged network raises curtailment rather than delivered energy [20] [21]. The risk is declaring victory on a throughput metric. |
The European approach attacks the administrative clock instead of the entry price. The Grids Package presented in December 2025 introduces tacit approval — a project is deemed permitted if the authority does not respond within two years, or three for the most complex cases — and proposes lifting the Connecting Europe Facility for Energy from 5.84 billion euros across 2021-2027 to 29.91 billion across 2028-2034 ✓ Established [23]. It is the most direct attempt yet to convert political urgency about grids into an enforceable deadline on the bodies that grant permission.
Texas offers the counterfactual for what pure access reform achieves. Connect-and-manage delivered 14.2 GW across 2021 and 2022 against 5.6 GW in the largest eastern market ✓ Established [20], with waits of roughly 20 months against 40 and completion rates of 40% against 24% for the 2020 cohort ◈ Strong Evidence [21]. The system connects projects faster than anywhere else in the country and curtails them more, because it did not pair fast access with proactive regional transmission planning. Speed of access and adequacy of network are separate variables, and only one of them was optimised.
Britain removed roughly half its queue in a single methodology change [17] and American queue volume fell 10% mainly through withdrawals [2]. Both outcomes improve the signal-to-noise ratio of the pipeline; neither adds a megawatt of transfer capability. With median connection times still above four years [3] and transformer lead times at about 128 weeks [8], the queue is now a more honest measure of a constraint that has not itself moved.
The most efficient intervention on the table is also the least discussed. Replacing conventional conductors with advanced composite-core designs in existing corridors could roughly double American transmission capacity by 2035, reduce wholesale electricity costs by 3 to 4% and save around 85 billion dollars in system costs by 2035 ◈ Strong Evidence [22]. It requires no new right of way, no new litigation and no new corridor, and pilot work suggests it can be delivered in months to a year against 10 to 15 years for a greenfield line ✓ Established [22].
Scored honestly, the reform record is one of process improvements against physical constraints that have not moved. Cluster studies, entry fees and statutory deadlines make the queue a better instrument. Reconductoring and grid-enhancing technologies actually raise transfer capability. Copper supply, transformer manufacturing and the workforce respond only to sustained investment over multiple years, and no jurisdiction has yet committed to that at the scale its own targets imply — except one, and it committed 4 trillion yuan ✓ Established [14].
What the Evidence Actually Supports
A process problem sitting on top of a materials problem
The strong claim — that the grid is the binding constraint on electrification — survives the evidence. The weaker claim behind most policy — that the constraint is chiefly administrative, and that faster studies and shorter permits will resolve it — does not ◈ Strong Evidence [3] ✓ Established [6]. Two constraints are stacked, and only the upper one has been addressed.
The first constraint is procedural. Interconnection studies, network upgrade cost allocation, environmental review and state siting together account for the difference between a two-year connection in 2005 and a 55-month connection in 2024 ✓ Established [3]. This constraint is genuinely tractable: it responds to cluster studies, entry deposits, statutory deadlines and tacit approval, and the American, British and European reforms of the past three years are all aimed at it ✓ Established [24] ✓ Established [17] ✓ Established [23].
The second constraint is physical, and it was not created by regulation. Transformers at up to four years, HVDC cables beyond five, cable factories that take three to four years to build, a projected 25% copper supply gap by 2035 and a grid workforce that must expand by 1.5 million people all sit downstream of every permit ✓ Established [6] ✓ Established [10]. No deadline binds them. They respond only to sustained, credible, multi-year orders — the exact commitment that stop-start policy and speculative queue behaviour have made hard to signal.
The Queue Is a Paperwork Problem
Some 72% of projects entering US queues between 1999 and 2018 were later withdrawn, largely because entering was cheap [25].
Britain cut its queue from more than 700 GW to 381.5 GW through a single methodology change [17].
Connect-and-manage delivered 14.2 GW in Texas across two years against 5.6 GW in the largest eastern market [20].
Tacit approval deems a project permitted if the authority fails to decide within two or three years [23].
Advanced conductors could roughly double US transmission capacity without a single new right of way [22].
The Queue Is a Steel and Copper Problem
Transformers take up to four years and HVDC cables more than five, with lead times roughly doubled since 2021 [6].
Distribution transformers are up 78-95% since 2019 and electrical steel has roughly doubled since January 2020 [8] [9].
A copper supply gap of about 25% is still projected for 2035 despite record prices and new projects [10] [12].
The global grid workforce of 8 million must grow by 1.5 million by 2030, and training runs in years [6].
Connecting generation to an unchanged network raises curtailment; German congestion management cost 2.7 billion euros in 2025 [18].
The evidence supports three conclusions with reasonable confidence. Queue reform improves the quality of the pipeline rather than its output: Britain halved its queue and America shed 10% of its own without adding transfer capability ✓ Established [17] ✓ Established [2]. Permitting reform is necessary but slow-acting, because it applies mainly to projects not yet begun ✓ Established [15]. And the fastest available capacity gain is reconductoring, which sidesteps both the legal and the corridor constraint while remaining exposed to the same conductor supply chain ◈ Strong Evidence [22].
Three claims do not survive. That the queue measures a real pipeline: three quarters of queued capacity is eventually withdrawn ✓ Established [3], and announced demand is also softer than filings imply — the American energy statistics agency cut its 2026 generation forecast to 4,327 billion kWh after large loads materialised more slowly than expected ✓ Established [27]. That the transformer shortage is purely exogenous: part of it is procurement behaviour ⚖ Contested [9]. And that renewable penetration caused the Iberian collapse: the formal finding is voltage control ✓ Established [19].
Electrification is not being held back by the cost of clean electricity. It is being held back by a stack of serial dependencies — permit, study, upgrade, transformer, conductor, crew — in which each link has lengthened independently since 2021. Reform has addressed the first two links, which are the cheapest to fix and the most visible. The remaining links respond to sustained industrial commitment measured in years, and they are the ones that determine whether the 2030 targets are met.
That reframing has a policy consequence. If the constraint were administrative, the correct instrument would be legal reform, and results would appear within a political cycle. If it is industrial, the correct instrument is a credible multi-year order book — long-term procurement frameworks, standardised transformer specifications, joint purchasing across utilities and training pipelines committed independently of election timetables. The IEA supply chain analysis makes exactly this recommendation, and its logic is the same one that made the Chinese build rate possible ◈ Strong Evidence [6] ◈ Strong Evidence [13].
The honest summary is this. The grid is the bottleneck, and the bottleneck has two components with different clock speeds. Britain, Brussels and Washington have spent three years fixing the fast component and have genuine results to show for it. The slow component — copper, electrical steel, winding capacity, cable factories, linemen — was never a matter of permission and will not be resolved by granting it faster. Until the second clock is set, every reform of the first one simply moves projects more efficiently towards the point where they wait.