Superconducting Infrastructure

Superconducting infrastructure means the hardware built from superconducting tape and installed in a power system: cables, fault current limiters, magnetic energy storage, transformers and busbars, together with the cryogenic plant that keeps them cold. The promise is easy to state. A superconducting cable carries several times the current of a copper cable of the same diameter with no resistive loss, which means an urban corridor can be upgraded without digging a wider trench, and a distribution voltage can do a transmission voltage's work.

The boundary with its sibling slot. High temperature superconductors owns everything priced per kilo-ampere-metre — the tape, its current, its yield, and the still-open physics of higher-temperature families, including the fusion magnet programme that sets the material's price. This brief owns everything priced per installed metre or per delivered megawatt: what has actually been energised, whether it stayed energised, and whether it paid. The arbitration rule is the unit of account.

This brief is organised around a base rate, and the base rate is the finding. Most writing on superconducting infrastructure proceeds from capability to implication — the cable can do this, therefore the grid could look like that. This one proceeds from the installed record. Summing every alternating-current and direct-current project in the February 2026 peer-reviewed review gives on the order of ten kilometres of superconducting cable ever installed in or alongside the world's electricity grids, across twenty-five years, with no project in the review identified as permanent commercial infrastructure rather than demonstration. That number is small enough that the individual projects can be listed, and this brief lists them, because a field with a complete enumerable deployment record should be argued about from the record rather than from the promise.

Established The deployment record is roughly ten kilometres in twenty-five years, and every project is a demonstration. The controlling source is a peer-reviewed review published in February 2026, which states directly that superconducting cables “have not yet penetrated the commercial cable market and still some prejudices exist mainly about the cost and the reliability.” The arithmetic summing its tables is this brief's; the entries are the review's.

Established The longest superconducting cable ever built is direct current, is in St Petersburg, and is 2.4 km. Installed 2021 at 20 kV and 2.5 kA. The longest alternating-current installations are the 1.2 km Shanghai cable at 35 kV and about 133 MVA, in full-load operation since August 2023, and the 1.1 km Korean Shingal–Heungdeok link at 23 kV and 50 MVA, in commercial operation since July 2019. The highest-voltage AC field test was 1 km at 154 kV on Jeju in 2016.

Established The best-performing installation in the record was decommissioned and never replicated in its own city. AmpaCity in Essen — 1 km at 10 kV, about 2.4 kA and 40 MVA, commissioned 2014 — was planned as a two-year trial and extended to seven on the strength of its performance. The European transmission and distribution operators' own factsheet lists it as decommissioned in 2024. Essen did not build a second one. The February 2026 peer-reviewed review still describes it as operating and reports no superconducting cable degradation anywhere since 2001; this brief reports both readings and resolves neither.

Established The cost multiple is the constraint, and there is exactly one published head-to-head figure. The Korean Shingal system was projected at US$12 million against US$3 million for the conventional alternative — four times — with part of the gap recovered by avoiding a 60 MVA transformer. Every economic argument in this brief anchors on that number, because it is the only one of its kind in the public record.

Established And here is the finding that inverts the usual story: the cryogenic penalty is not the problem. AmpaCity's operators reported that the cost of the energy required to cool the cable was 15% lower than the cost of compensating losses in equivalent conventional 110 kV cables. Heat in-leak in current cryostats is down to 0.8 W per metre, with alternating-current losses well below 1 W/m at 2,600 A. The thermodynamics are fine. The capital cost and the organisational risk are what stop projects.

Established The operational objection that survives scrutiny is fault recovery, not steady-state efficiency. National-laboratory modelling of thermal recovery after an overcurrent fault gives the number that matters: a 1 km cable takes about 2.2 hours to recover with unlimited refrigeration, and 5.8 hours with a realistically sized 5 kW plant, because the fault heat load exceeds 13 kW against a 5 kW steady-state baseline. A fault does not destroy the cable; it removes it from service for a shift. And the refrigeration plant must therefore be sized for fault recovery rather than for steady state, which is a hidden capital multiplier that no efficiency comparison captures.

Established There has never been a grid-scale superconducting magnetic energy store, and the margin is four to five orders of magnitude. The frontier scale in 2026 is 10 megajoules. That is 2.78 kilowatt-hours. A single utility battery module of the class now installed by the thousand stores roughly 3.9 MWh, about 14,000 MJ. The largest superconducting storage demonstration in the world is therefore around 0.02% of one grid battery module.

Frontier The demand has left the grid. The best-capitalised superconducting transmission developer has shifted from overhead transmission to underground, and from utility transmission to data-centre power delivery, running a pilot in November 2024 that moved 3 MW through a single cable with a claimed order-of-magnitude reduction in cable size and weight. On current evidence the live subject of this brief in 2026 is fusion magnets and data-hall busbars, not transmission lines.

Established And the regulatory diagnosis is stated plainly in the review literature. Among the barriers listed alongside cost and reliability is the observation that “regulatory policies do not incentivize efficient transmission via loss reduction.” A utility that reduces its losses in most rate structures reduces its own revenue base. The technology's central benefit is one its buyers are not paid to want.

Established The genuine frontier here is institutional and financial, not technical, and the record supports that claim rather than merely asserting it. Cryostats work at 0.8 W/m. Terminations work. Splices work — the first in-grid superconducting cable splice was made at Albany, New York, in 2006. Refrigeration works. What has not happened is a purchase.

Frontier The one technical frontier that is real: fault-recovery-sized refrigeration. If a cable's plant must be sized for a 13 kW fault transient rather than a 5 kW steady state, the plant is roughly three times larger than the duty requires, and refrigeration is a substantial share of both capital and footprint. Reducing recovery time — through cable thermal design, staged cooling, or accepting a longer outage in exchange for a smaller plant — is a live engineering trade nobody has published an optimum for.

Frontier Data-centre power delivery is the frontier application, and it is frontier precisely because it changes the economics rather than the physics. A hyperscale data hall wants very high current at low voltage over short distances inside a constrained building envelope — which is the regime where superconducting busbars are least disadvantaged, because the cryogenic plant is centralised, the run is short, the load factor is near unity, and the customer is not a rate-regulated utility with no incentive to reduce losses. The November 2024 pilot moved 3 MW through a single cable. Whether this becomes a market or remains a pilot is the most consequential open question in this brief.

Frontier Fault current limiters are the least bad grid product, and their record is genuinely unresolved. The value proposition is the clearest in the sector: a superconducting limiter suppresses fault current so that existing switchgear need not be replaced and a substation upgrade can be deferred, which is a capital-avoidance argument rather than a loss-reduction argument, and capital avoidance is something utilities are paid for. The most recent verified deployment is a rail-traction limiter at a French substation, announced June 2024, self-regenerating in under five minutes without human intervention, described by its vendor as a world first in rail and scheduled for deployment in late 2025. Chinese and Russian installations at 66 kV, 110 kV and 220 kV are documented in literature this brief could not retrieve.

Frontier No worldwide count of installed superconducting fault current limiters could be established from any source consulted, and that is itself the finding. If these were a commercial product, an installed-base count would exist in a review or on a vendor page. None was found. This brief therefore declines to call fault current limiters a commercial success story, which is how they are usually described, and records the question as open and leaning negative.

Speculative Superconducting transformers are a frontier only in the weak sense that nobody is working on them. Essentially all the literature found dates from 2002 to 2015, and no grid-connected trial from 2024 to 2026 surfaced. This is recorded as a strong hypothesis rather than a verified absence, because it was not chased exhaustively.

Handwave Grid-scale superconducting magnetic energy storage. Not a frontier in any operational sense. The published diagnosis from the most recent design literature is that existing designs prioritise extreme performance metrics over magnet-design economy and scalability, and that the decisive problem is “low energy density and insufficient tape utilisation, severely constraining the economic viability of SMES magnets.” Four to five orders of magnitude separate the demonstrated scale from grid relevance, and no route across that gap has been proposed.

Established Capital cost, at roughly four times the conventional alternative. US$12 million against US$3 million on the only published head-to-head, partially offset by avoiding a transformer. An independent European assessment states the requirement bluntly: a price reduction by a factor of four is needed. Note that this is a factor-of-four problem in system cost, driven substantially by a tape price that has been flat for four years — which is why this brief's single dependency runs to the sibling slot.

Established Regulatory structure, which is the most underrated barrier in the subject. The February 2026 review states that regulatory policies do not incentivise efficient transmission via loss reduction. A superconducting cable's flagship benefit is lower losses; in most rate designs, losses are a pass-through cost rather than a shareholder cost, so eliminating them creates no return for the entity that must finance the cable. The technology is misaligned with the incentive structure of its only mass buyer, and no amount of engineering improvement fixes that.

Established Fault recovery time, and the plant oversizing it forces. Two point two hours to recover a 1 km cable with unlimited refrigeration; 5.8 hours with a 5 kW plant; fault heat load above 13 kW. A transmission asset that is out of service for a shift after a fault is a different reliability proposition from one that recloses in a second, and the refrigeration plant must be sized for the transient rather than the duty.

Established Consumables and their logistics, which is a live utility complaint rather than a theoretical one. A 2025 annual report filed with a state utility regulator in February 2026, covering a superconducting transmission project in New York, records that the lessons learned so far “relate to the volume of nitrogen required for both overhead- and underground-based solutions,” and that the developer has shifted from an overhead to an underground design. This is the closest thing in the public record to a contemporaneous utility statement of the consumables burden, and it appears in a regulatory filing rather than in marketing material.

Established Thin operational data, which compounds every other barrier. The grid-operator factsheet's own barrier list names limited long-term operational data alongside high material cost, lack of standardisation, mechanical sensitivity, complex system integration, and susceptibility to faults during quenches. With roughly ten kilometres installed worldwide and most of it decommissioned demonstrations, no utility can underwrite a superconducting cable against an actuarial record, because there is no actuarial record.

Established Black-start capability, a specific and rarely mentioned limitation. The review lists limited black-start capability among the barriers, because the auxiliary power required by the refrigeration plant must come from somewhere when the system is down. A transmission asset that cannot be brought up without external power is a constrained asset in exactly the scenario transmission planners worry most about.

Frontier Reliability history, which cuts both ways and should be reported both ways. The founding failure is well documented: the 2001 Detroit cooling system, in the review's phrasing, did not have the desired reliability and maintainability for continuous operation in a utility environment, and leaks in two cables meant they could never be energised. That single project shaped a generation of utility opinion. Against it, the same review reports no superconducting cable degradation anywhere since 2001, and AmpaCity's two-year trial was extended to seven on performance. The technology's reliability record after 2001 is good; its reputation was set in 2001.

Established And the recursive bottleneck the industry names itself. A utility publication describing its own installation called it explicitly a chicken-and-egg problem: price will not fall without volume, and volume will not come without a lower price. With no grid buyer at scale, nobody qualifies tape to utility duty cycles, no standards mature, and the learning that would lower costs does not occur.

Established This brief has exactly one technical dependency, and it is a price rather than a result. Everything else — cryostats, refrigeration, terminations, splices, protection — is engineered, demonstrated and reliable. What the field waits on is the cost and uniformity per kilo-ampere-metre of REBCO tape, which has been flat since 2022 and is set by a market this brief does not participate in. The sibling brief, high temperature superconductors, explains why: the superconductor is a few percent of the tape's cross-section, so production volume moves cost per metre and not cost per unit of current.

Established The second dependency is institutional and is the binding one. A buyer must exist whose incentives reward what this technology does. Under rate-of-return regulation, loss reduction is a pass-through rather than a return, so the technology's flagship benefit generates no shareholder value for its natural customer. Every project in the deployment record was financed as a demonstration, by a research programme, a national utility acting on policy grounds, or a vendor. None was financed as an ordinary capital investment expecting an ordinary return.

Established A third dependency, ordinary but real: liquid nitrogen logistics. A regulatory filing records nitrogen volume as the leading lesson learned on an in-progress project, and as the reason a developer shifted from overhead to underground. Nitrogen is cheap and made from air; delivering, storing and reliably circulating it along a right-of-way, for decades, without an outage, is not a solved logistics problem at transmission scale.

Frontier A fourth, which the record makes visible: standardisation. Named explicitly in the grid operators' barrier list. With every installation a bespoke demonstration, there are no type-tested designs, no standard terminations, no qualified duty cycles and no interoperable components. Standardisation follows volume, so this dependency is downstream of the institutional one.

Established What depends on this brief. Very high current density through a small cross-section, which dense computing loads and constrained urban corridors both want. Energy corridors, where this competes with high-voltage direct current and currently loses on cost and on operating record. And nothing else — which is worth saying, because a technology with a thousandfold performance margin and one enumerable list of dependants is a technology whose problem is demand.

Established The most informative experiment in this subject has already run, and it ran for ten years. AmpaCity was a genuine field trial: 1 km at distribution voltage in a working city network, planned for two years and extended to seven on performance, with a reported cooling energy cost 15% below the cost of compensating losses in equivalent conventional cable. Its result is unambiguous and is not the result the technology's advocates wanted: it worked, and it was not repeated. Essen did not build a second superconducting link. That outcome is more informative than any efficiency measurement, because it is a revealed preference by an operator with complete information.

Established The Shanghai cable is the longest-running full-load AC test and its cooling data is the missing piece. Operating at 35 kV and 2.2 kA over 1.2 km entirely within existing ducts, at full load since August 2023. A paper reporting two years of successful cooling-system operation exists and could not be retrieved for this brief; it would supply real cooling-plant power, nitrogen consumption and availability figures for the longest operating AC superconducting cable in the world. Its absence is the largest single gap in this brief and is recorded as such.

Frontier SuperLink is the experiment that would settle the transmission question, and it has not been built. The full project is 110 kV and 500 MW, with length quoted inconsistently at 12 km by the research consortium and up to 15 km by the manufacturer. What exists is a 120 to 150 metre demonstrator at a Munich substation, powered up in October 2024 with testing to complete in mid-2025. In April 2026 the manufacturer and the municipal utility signed a Letter of Intent — an agreement to negotiate toward a binding contract. A twelve-kilometre superconducting link at transmission voltage would multiply the world's installed base by roughly a factor of two on its own, which is the clearest possible statement of how thin that base is.

Frontier The data-centre pilot is the experiment most likely to change the field's direction. Three megawatts through a single superconducting cable in November 2024, with a claimed order-of-magnitude reduction in cable size and weight, funded partly by a hyperscaler's climate fund. If this scales, superconducting infrastructure will have found its first buyer with both the load factor and the incentive structure to want it, and it will not be a utility.

Frontier The rail fault-current-limiter deployment is a small experiment with a clean readout. Announced June 2024 for a French regional line, self-regenerating in under five minutes unattended, deployment scheduled late 2025. Rail traction is an attractive niche — concentrated fault duty, a single owner-operator, and no rate-of-return regulation dulling the incentive. Whether a second one follows is the readout.

Established A negative result recorded as a result. A full review of superconducting magnetic energy storage published in 2023 contains no installation list, no ratings and no costs. A national energy-storage database maintained by a government laboratory supports technology filtering and returned no rows for this technology to the query made for this brief. When two independent attempts to enumerate a technology's installed base both come back empty, the emptiness is the finding.

Established The deployment record, in full, because it is short enough to enumerate. Alternating-current projects, from the February 2026 review's tables:

Established Direct-current cables and busbars, which include the longest ever built:

Established What the engineering actually consists of. A superconducting cable is a former, the tape wound helically around it, electrical insulation, a shield layer of more tape carrying the return current, and a vacuum-insulated cryostat with liquid nitrogen circulating in and out. Terminations — where the cold, superconducting conductor meets the warm copper grid — are the hardest components, because they must carry full current across a temperature gradient of two hundred kelvin without either boiling the nitrogen or conducting a ruinous heat load inward. The engineering is mature; all four of these subsystems have worked in the field for two decades.

Established The thermal numbers are good and are the reason the cryogenics objection fails. Modern cryostat heat in-leak is 0.8 W/m, measured on a twelve-metre sample. Alternating-current losses are below 1 W/m at 2,600 A. A vendor calculation for a 1 km 110 kV comparison gives roughly 200 kW lost in two conventional cross-linked polyethylene cables each carrying 1.5 kA, against about 112 kW for one superconducting cable including thermal losses and cooling-machine power — better than a 40% reduction. That figure comes from an interested party and depends on the chosen copper baseline and utilisation, so it is quoted as a vendor calculation rather than a general result.

Established And the space argument is the one utilities actually respond to. The Shanghai installation runs entirely in existing ducts and reports roughly a 70% saving in duct space. AmpaCity's operator described replacing a substation the size of a gymnasium with equipment the size of a double garage. Where the constraint is urban right-of-way rather than energy loss, the value proposition changes character completely.

The sibling slot, high temperature superconductors, is adjacent in the strongest sense available in this corpus: the two briefs share a research pack and split it by unit of account. That brief's finding is that the material has been adequate for fifteen years and has not become cheaper per unit of current. This brief's finding is that almost nothing has been built from it. Read together they make a single argument about the distance between capability and deployment that neither makes alone, and the two findings are causally linked in both directions — flat tape prices deter buyers, and absent buyers remove the volume that would move tape prices.

Energy corridors are adjacent as the direct competitor. High-voltage direct current does most of what superconducting transmission promises, at lower cost, with a deployment record measured in tens of thousands of kilometres rather than tens. Any honest case for superconducting transmission has to be made against HVDC specifically, on right-of-way width or urban constraint rather than on losses.

Data-centre power delivery is adjacent as the emerging application and is where this brief's subject may actually end up. It differs from grid transmission in every dimension that matters here: short runs, centralised cooling, near-unity load factor, unregulated buyer, capital urgency.

Utility regulation is adjacent as the binding constraint, which is an unusual thing for a hardware brief to say and is the honest reading of the evidence. And commercial fusion is adjacent as the place the tape actually went — the largest structures ever built from this material are magnets, not infrastructure, which is why they are treated in the sibling brief.

Established No institution anywhere is a volume buyer of superconducting infrastructure, and that single fact organises everything else in this brief. Every project in the twenty-five-year record was financed as a demonstration — by a research programme, by a national utility acting on industrial-policy grounds, or by a vendor buying a reference installation. None was financed as an ordinary capital investment expecting an ordinary regulated return.

Established The regulatory diagnosis is explicit in the peer-reviewed literature and deserves to be quoted rather than paraphrased: regulatory policies do not incentivise efficient transmission via loss reduction. In most rate structures a network operator recovers losses as a pass-through and earns a return on capital deployed. A technology that reduces losses therefore transfers value to ratepayers while consuming the operator's capital budget, which is the opposite of the arrangement that drives investment. Fixing this requires rate reform, not engineering, and rate reform is nobody in this sector's competence.

Established The barrier lists compiled by the industry's own institutions are consistent and technical failures are not near the top. The European transmission and distribution operators' factsheet names high material cost, lack of standardisation, mechanical sensitivity, limited long-term operational data, complex system integration, and susceptibility to faults during quenches. The February 2026 review adds cost and reliability prejudice, dielectric and thermal limits in AC designs, limited black-start capability, and the regulatory incentive problem. Of roughly ten distinct barriers named across two independent institutional sources, most are commercial, informational or regulatory.

Established Standardisation is absent and cannot arrive first. Every installation is bespoke: no type-tested designs, no standard terminations, no qualified utility duty cycles, no interoperable components, no second sources. A utility procuring a superconducting cable is procuring a research project with a warranty, and its engineering staff must develop competence that has no other use in the organisation. Standards follow volume, so this cannot be solved ahead of the buyer problem.

Established Regulatory filings are the most reliable institutional source in this subject, and this brief leans on one deliberately. A 2025 annual report filed with a state commission in February 2026 records that the leading lesson from an in-progress superconducting transmission project concerns the volume of nitrogen required, and that the developer moved from an overhead to an underground design. Utilities tell regulators things they do not put in press releases, because the filing is compelled and the press release is not. Anyone researching this field should read the filings first.

Frontier The institution most likely to break the deadlock is not a utility. A hyperscale data-centre operator has the load factor, the capital, the urgency, the physical constraint, and — decisively — no rate regulation dulling its incentive to reduce losses and save space. The pivot of the sector's best-capitalised developer from utility transmission to data-centre power delivery, and a hyperscaler's climate fund financing a 3 MW pilot, are the clearest institutional signals in the recent record. If superconducting infrastructure gets an installed base this decade, this is where it comes from.

Frontier Rail is the other candidate and for structurally similar reasons. A single owner-operator, concentrated fault duty, constrained corridors, and no rate-of-return regulation. Both recent European deployments — a traction cable in Paris in 2024 and a fault current limiter on a regional line announced the same year — are rail. That is a small pattern in a small dataset, and it points the same way as the data-centre evidence: this technology's buyers, if it has any, will be operators who own their own constraint.

Established The most concrete ethical issue in this brief is the reliability of the evidence base, and it is severe enough to shape how the brief is written. Almost everything published about superconducting infrastructure comes from vendors, developers, utilities describing their own installations, or industry bodies. Every one of those parties has a stake in the perception that the technology works and is close to commercial. The two claims this brief flags most sharply — that Munich is building the world's longest superconducting cable, and that fault current limiters are a commercial success — both originate with interested parties and both fail against the primary record.

Established The concrete instance is worth naming because it is instructive rather than scandalous. A vendor describes a 200 metre installation as operational since 2021; the peer-reviewed review lists it among completed historical field tests. Both statements can be true simultaneously — hardware in place and energised, project concluded — and the difference between them is precisely the difference between a technology in service and a technology demonstrated. That gap is where most public understanding of this field lives. This brief cites the vendor source and states the disagreement rather than picking the flattering reading.

Established Public money and public accountability. A substantial share of these installations was publicly financed as demonstration projects intended to generate operating knowledge. Yet the single most valuable dataset in the field — two years of cooling-system operating data from the longest-running AC installation — sits behind a publisher's paywall and a robots policy, and could not be retrieved for this brief. Cooling-plant electrical power as a fraction of transmitted power could not be established for any operating installation. When demonstrations are publicly funded to produce knowledge, and the knowledge is not publicly reachable, the demonstration has not fully discharged its purpose.

Frontier An honest question about opportunity cost. Public funds spent on superconducting demonstrations are funds not spent on high-voltage direct current, on distribution automation, or on storage — all of which have better-established records per dollar. This brief does not conclude that the spending was wrong; demonstration programmes exist precisely to buy information about technologies that have not proven themselves, and information was bought. But the argument for continuing has to be made against alternatives with better records, and it is usually made against a counterfactual of doing nothing.

Established Where this brief's own uncertainty sits, stated as an obligation rather than a caveat. Four substantive questions could not be resolved from sources this brief could reach: whether the German flagship was truly decommissioned in 2024, whether the Chicago installation is in continuing revenue service, how many superconducting fault current limiters are installed worldwide, and what fraction of transmitted power any real installation spends on cooling. Each is a fact somebody knows. All four are stated in the brief as unresolved rather than filled with the most plausible estimate, which is the four-flag system's purpose and the reason the reader can trust the figures that are asserted.

Established The civilisational content of this brief is a case study in a general failure mode, and it happens to be unusually well documented. Here is a technology with roughly a thousandfold physical performance margin over its application, a mature engineering base, a favourable energy balance in operation, and twenty-five years of opportunity — which has produced about ten kilometres of installed cable worldwide, most of it decommissioned. Nothing about that outcome is explained by physics or by engineering competence. It is explained by capital cost, by rate design, and by the absence of a buyer whose incentives point the right way.

Established The rate-design point deserves stating as a general principle, because it generalises well beyond superconductors. When the entity that must finance an efficiency improvement is not the entity that captures its value, the improvement does not happen at any technology-readiness level. Loss reduction in most electricity rate structures is a pass-through cost rather than a shareholder return. A technology whose headline benefit is loss reduction is therefore misaligned with its only mass buyer, and no engineering progress corrects that misalignment. Anyone building a workback plan from a demonstrated capability to a deployed world has to model the buyer's incentive structure with the same seriousness as the physics, and it is the step most such plans skip.

Speculative The version of this technology that would matter civilisationally has not been attempted. Continental superconducting corridors would loosen a constraint that currently shapes which renewable resources are worth developing at all, by decoupling generation siting from load siting. The material supports it with enormous margin. The deployment record contains nothing pointing that way, and this brief marks the gap between what is physically supportable and what is institutionally underway as the honest content of the question rather than as a forecast.

Frontier The more likely civilisational path runs through computing, and it is a smaller and stranger story. If superconducting power delivery finds its first real market inside data centres, the technology will have been adopted not to decarbonise a grid but to fit more compute into a building. That is a much narrower outcome than the one this field has promised for forty years, and on current evidence it is the one actually in progress.

Established One more thing the record teaches, and it is a hopeful one. AmpaCity ran for a decade with no degradation, at a cooling energy cost below the loss-compensation cost of the cable it replaced. The pessimistic conclusion of this brief is about institutions, not about the hardware. If the buyer problem were solved, the technology would work — and that is a materially different situation from one where the technology needs a breakthrough.

Established 2000 to 2006, the first wave and its founding failure. Carrollton, Detroit, Copenhagen, Yokosuka, Albany. Detroit's cooling system failed to meet utility reliability requirements and two of its cables were never energised; Copenhagen was decommissioned on reliability. Albany produced the first in-grid use of second-generation tape and the first utility-grid splice. Utility opinion of this technology was substantially fixed in this period and has not been revisited.

Established 2012 to 2021, the second wave, longer and quieter. Yokohama, AmpaCity, Jeju at 154 kV, Shingal in commercial operation, Chicago, Shanghai, St Petersburg. This is where kilometre-class installations appear and where the reliability record is actually good. It is also where the record ends: no new kilometre-class grid cable has entered service anywhere since 2021 on the evidence consulted.

Established 2024 to 2026, contraction and redirection. AmpaCity decommissioned per the grid-operator factsheet. SuperLink still a 150 metre demonstrator with a Letter of Intent as of April 2026. The leading transmission developer shifted from overhead to underground and from transmission to data centres. Fault current limiters appearing in rail rather than transmission.

Frontier Late 2020s: SuperLink, or not. A Letter of Intent is not a contract, and no dates accompany it. If a 12 km 110 kV link is built it is the most significant event in this field's history; if the Letter of Intent lapses, that is nearly as informative.

Frontier Late 2020s: whether data-centre deployment moves from pilot to product. This has a shorter cycle than grid infrastructure, is not rate-regulated, and has a buyer with capital and urgency. It is the likeliest source of a genuine installed base within the decade.

Handwave Any date for grid-scale superconducting magnetic energy storage. Four to five orders of magnitude of scale-up with no proposed route. This brief declines to supply a date.

Speculative 2030s and beyond: superconducting transmission corridors at continental scale. Technically supportable by the material with enormous margin. Nothing in the deployment record, the regulatory structure or the industrial base suggests movement in that direction, and the honest statement is that the constraint is not on a timeline because nobody is working on it.

• Depends on Nothing on this map. This brief has exactly one technical dependency and it is a price, not a result: everything else — cryostats at 0.8 W/m, refrigeration, terminations, splices demonstrated in a utility grid since 2006, protection — is engineered and proven in the field. The price is recorded below.
• Requires (not on this map) Cost and uniformity per kilo-ampere-metre of REBCO tape, flat since 2022, with an independent European assessment putting the requirement at a factor-of-four reduction. A rate structure that lets a network operator earn on loss reduction, absent in the jurisdictions where these projects were built and named in the peer-reviewed literature as a barrier. Type-testing and standardisation, which cannot precede volume. Liquid nitrogen logistics at corridor scale, named in a 2026 regulatory filing as the leading lesson learned on an in-progress project. And an anchor buyer outside the regulated utility sector — on current evidence a data-centre operator rather than a transmission owner. Every one of these is industrial, financial or institutional. None is a discovery, and that is this topic's position on the map.
• Enables Very high current density through a small cross-section, which dense computing loads and constrained urban corridors both want — a 70% duct-space saving is reported on the longest operating AC installation. The relation is to an application class rather than to a specific brief, so no typed enabling edge is claimed.
• Adjacent High temperature superconductors, the sibling slot that sets this brief's only price; energy corridors, where this competes with high-voltage direct current and currently loses on cost and operating record; planetary-scale energy systems; and commercial fusion, whose magnets are by a wide margin the largest structures ever built from this tape and which sit in the sibling brief because they set the material's price rather than consume its infrastructure.

“Superconducting cables are held back by the cost and complexity of cryogenics.” Established This is the standard explanation and the evidence contradicts it. AmpaCity's cooling energy cost 15% less than compensating the losses of equivalent conventional 110 kV cable. Cryostat heat in-leak is 0.8 W/m; AC losses are below 1 W/m at 2,600 A. The binding constraints are capital cost at roughly four times conventional, a rate structure that does not reward loss reduction, and an operating record too thin to underwrite. Cryogenics is the intuitive objection and it is close to the weakest one.

“Munich is building the world's longest superconducting cable.” Handwave In the present tense, this is false. What exists is a 120 to 150 metre demonstrator powered up in October 2024. As of April 2026 the parties had signed a Letter of Intent — an agreement to negotiate — with no binding contract and no dates. The full project's length is quoted inconsistently, at 12 km by the research consortium and up to 15 km by the manufacturer, which is itself a signal about how settled the design is.

“AmpaCity proved the technology and it is still running.” Established It ran for roughly a decade against a two-year plan, which is a genuine success, and the European transmission and distribution operators' factsheet lists it as decommissioned in 2024. The February 2026 peer-reviewed review still describes it as operating. This brief reports both, resolves neither, and notes the fact neither source disputes: Essen did not build a second one.

“Fault current limiters are the commercial success story of applied superconductivity.” Frontier This is the sector's most repeated claim and no worldwide installed count could be verified from any source consulted for this brief — not from a review, not from a vendor's installed-base page. If the product were commercial, that count would exist. The most recent verified deployment is a single rail-traction limiter announced in June 2024 and described by its vendor as a world first in rail, which implies no prior rail installations. Treat the framing as unproven.

“Superconducting magnetic energy storage is an emerging grid storage technology.” Established It is not emerging and it is not grid storage. The frontier scale in 2026 is 10 MJ — 2.78 kWh, roughly 0.02% of one utility battery module. The historically significant deployment, six distributed units bought for a Wisconsin transmission loop in 1999 at $4 million against $6 to $15 million for the alternatives, was procured explicitly as a stopgap for voltage instability until a 345 kV line entered service in 2002. The line arrived and the stopgap's job ended. That is the technology's most successful commercial deployment.

“A quench or a fault destroys a superconducting cable.” Established It does not. It takes the cable out of service while the system re-cools: about 2.2 hours for a 1 km cable with unlimited refrigeration, 5.8 hours with a realistically sized 5 kW plant. The real consequence is subtler and worse for the economics — the refrigeration plant must be sized for a fault heat load above 13 kW rather than a steady-state 5 kW, which oversizes the plant by roughly threefold.

“The technology is unreliable.” Frontier The reputation dates to 2001, when the Detroit installation's cooling system failed to meet utility reliability requirements and leaks meant two cables were never energised. Since then the same review that documents that failure reports no superconducting cable degradation anywhere, and the flagship installation outlasted its planned life by a factor of three and a half. The reliability record after 2001 is good; the reputation was set in 2001 and has not been updated.

“This is about saving transmission losses.” Established On the record, it is about space. The longest operating AC installation reports roughly a 70% saving in duct space and runs entirely in existing ducts; the flagship German project's own operator described replacing a gymnasium-sized substation with equipment the size of a double garage. Where right-of-way is the binding constraint the value proposition is strong. Where losses are the argument, the buyer is a regulated utility that does not earn on reducing them.