Floating Offshore Wind

Floating offshore wind is the attempt to put a conventional offshore wind turbine on a moored, buoyant substructure so that it can be installed in water too deep for a fixed foundation — roughly beyond 60 metres, where the monopile and jacket become uneconomic. The turbine is the same machine. Everything below the tower flange is different, and so is the industrial system that builds, installs and repairs it.

Established The resource case is genuine and it is the reason the field exists. Most of the world’s good offshore wind sits over water deeper than fixed foundations can reach: the continental shelf off California, Japan, Korea, most of the Mediterranean, the deep Atlantic margin and the Celtic Sea. Those are also coastlines with dense demand and weak onshore siting options. A technology that unlocks them is not a marginal addition to the wind fleet; it is the difference between offshore wind being a North Sea phenomenon and a global one.

Three unit conventions, because this subject is unusually easy to misreport. First, capacity factor: a figure from a five-turbine pilot in an exceptional wind regime is not a fleet number, and a figure quoted gross of availability losses is not comparable with one quoted net. Second, cost: capital cost per megawatt installed, levelised cost of energy, and a contract-for-difference strike price are three different quantities, and only the third is an actual market observation — the first two are reconstructions. Third, scale: “the world’s largest floating wind farm’’ has meant 88 megawatts for years, which is one-twentieth of a routine fixed-bottom project.

Established Where this brief stops. Turbine aerodynamics and drivetrain engineering are shared with onshore and fixed-bottom wind and are treated only where floating changes them. Grid architecture beyond the export cable belongs to energy corridors; seabed law and marine environmental governance belong to ocean engineering; the quays, vessels and heavy-lift capacity that this technology competes for belong to future ports and shipping. This brief owns the substructure, the moorings, the dynamic cable, the maintenance strategy and the cost trajectory.

Established The claim this brief lands. The physics is settled and the pilots work. The open question is entirely industrial and financial: whether substructures can be fabricated serially in ports that do not currently exist at the required specification, whether moorings and dynamic cables achieve a twenty-five-year life that no installation has yet demonstrated for more than about eight years, and whether the cost of capital falls far enough to make the resulting electricity competitive. Two of those three are decided on quaysides and in credit committees, not in laboratories.

Established A note on sourcing. This brief was commissioned in September 2026 from the Institute’s research base. Reading-list entries without links are cited from the bibliographic record rather than re-fetched, and claims are dated no later than early 2026 unless carried by a linked source.

Established The installed base is tiny and that is the most important fact in the subject. Cumulative global floating offshore wind capacity was in the region of 250 to 300 megawatts at the end of 2024 — the exact figure depends on whether decommissioned demonstrators are counted — against roughly 80 gigawatts of fixed-bottom offshore wind. That is a ratio of about one to three hundred. Every cost projection, reliability estimate and industrial plan in the field is extrapolated from a fleet of a few dozen machines.

Established Hywind Scotland is the reference installation and its numbers are good. Five 6 megawatt turbines on Equinor spar substructures, 30 megawatts total, commissioned off Peterhead in October 2017. The operator has reported average capacity factors in the mid-fifties per cent across multi-year periods, with a twelve-month figure above 57 per cent and monthly peaks around 65 per cent (operator’s own figures). Those are among the highest capacity factors recorded by any wind installation anywhere. They reflect an exceptional wind regime and a very small sample, and they are the single strongest piece of evidence that the technical concept works.

Established Hywind Tampen is the largest and it was not built for the grid. Eleven turbines of about 8.6 megawatts each, 88 megawatts total, on concrete spar substructures, fully operational in August 2023 and supplying electricity directly to the Snorre and Gullfaks oil and gas platforms rather than to shore. Reported capital cost was in the region of NOK 7–8 billion, a substantial part of it met by the Norwegian NOx Fund and Enova. That implies a capital cost per megawatt several times fixed-bottom offshore wind, in a project whose economics were partly an emissions-abatement transfer rather than an electricity sale.

Established The semi-submersible pilots established the tow-out model. WindFloat Atlantic in Portugal, 25 megawatts on three Principle Power semi-submersibles, was completed in 2020; Kincardine off Aberdeen, about 50 megawatts on the same platform family, followed in 2021. Both were assembled at a quayside and towed out fully commissioned, which is the industrial promise of floating wind: no jack-up vessel, no offshore lift, work done in a shipyard rather than at sea. Both are also small enough that the fabrication throughput question was never posed.

Established Japan ran the honest counter-experiment and published the result. The Fukushima FORWARD demonstration, a national programme deploying 2, 5 and 7 megawatt floating turbines and a floating substation from 2013, was terminated and the units dismantled by 2021 after capacity factors far below design — the largest unit reportedly operating in the low single digits per cent — and a government determination that the project could not be made economic. It cost on the order of ¥60 billion. It is the field’s most complete negative result, and it is chiefly a result about system availability and maintenance access rather than about hydrodynamics.

Frontier The cost gap to fixed-bottom is large and contested in its size. The intergovernmental benchmark puts the global weighted-average levelised cost of offshore wind at about US$78 per megawatt-hour in 2025, a figure dominated by mature fixed-bottom projects including China. Credible published estimates for floating projects reaching financial close in the mid-2020s sit roughly in the US$150 to 250 per megawatt-hour band, and the spread between estimates is itself the finding: with a handful of projects, there is no market price, only models. The United States Department of Energy’s Floating Offshore Wind Shot set a target of US$45 per megawatt-hour by 2035 (programme’s own target), which is a policy ambition rather than a forecast.

Established The one real market price the field has is very high. The United Kingdom contract-for-difference mechanism awarded a floating-wind contract to the Green Volt project in its 2024 allocation round at a strike price near £140 per megawatt-hour in 2012 prices, which is substantially more in money of the day and roughly two and a half times the fixed-bottom offshore clearing price in the same round. That is an auction outcome, not a model, and it is the most defensible single number in the subject.

Established The 2023–2024 auction failures are bankability evidence and they were not about floating. The United Kingdom’s fifth allocation round in September 2023 received zero offshore wind bids because the administrative strike price was set below what developers would accept after input-cost inflation and rate rises. In the United States, several contracted fixed-bottom projects were renegotiated or cancelled across 2023 and 2024, with one developer taking a multi-billion-dollar impairment on two New Jersey projects. If a mature technology with a decade of delivery history could not hold its contracts through a rate cycle, the inference for a technology with no delivery history is not reassuring.

Frontier The service record on the parts that are new is short and not systematically published. The longest continuously operating floating array has been in the water since 2017. Moorings, chain and connector hardware, dynamic array and export cables, bend stiffeners and buoyancy modules are all specified for twenty-five to thirty years. That is a qualification gap of a factor of three or more in elapsed time, closed today by analysis and accelerated testing rather than by service experience. Individual incidents — units towed back to port for repair, cable interventions — are reported in the trade press; no pooled, component-level reliability dataset for the global floating fleet exists in public.

Established Subsea cable failure is already the dominant loss driver in fixed-bottom offshore wind. Insurers and operators have consistently attributed the large majority of offshore wind claim value to cable damage — installation defects, burial failures, bend-radius violations and third-party strike. Floating wind takes that component, which fails most often when static, and subjects it to continuous motion, curvature cycling and hang-off fatigue for its whole life. The dynamic cable is not one risk among several; on the evidence from the adjacent industry it is the risk.

Frontier The capital-cost structure is dominated by things that are not the turbine. Substructures for a 15 megawatt machine run to thousands of tonnes of steel or concrete each; moorings and anchors, dynamic cables and the port and vessel campaign account for much of the remainder. This is why the industry’s cost-reduction case rests on serial fabrication rather than on technology: the claimed path to competitive cost is a yard producing dozens of identical hulls a year with learning effects, and no yard anywhere has produced more than a handful.

Frontier Whether serial substructure fabrication achieves the assumed learning rate. Every published cost pathway assumes a manufacturing learning curve derived from other industries. Floating substructures are large, low-value-density steel or concrete objects whose transport cost rises steeply with distance, which constrains how much production can be concentrated in the cheapest yards. Whether learning beats that constraint is the central unresolved economic question and it will be answered by the first multi-hundred-unit order book, not before.

Frontier Whether dynamic cables reach design life. The qualification case rests on fatigue analysis, accelerated rig testing and read-across from offshore oil risers and umbilicals. That read-across is imperfect: wind array cables are smaller, more numerous, cheaper per unit and installed by contractors working to different margins. The open question is not whether a dynamic cable can survive but what the failure rate per cable-year is at fleet scale, and only a fleet answers it.

Frontier Whether major components can be exchanged without tow-to-port. The base case for replacing a gearbox or blade on a floating unit is to disconnect the moorings and cable and tow the whole machine to a quay — a multi-week operation requiring a weather window, tugs and a port that can take it. The alternative, offshore exchange from a motion-compensated floating crane vessel onto a moving platform, has been demonstrated only in limited form. The difference between these two is a large fraction of lifetime operating cost.

Speculative Whether floating substations arrive in time. Beyond a few tens of kilometres and a few hundred megawatts, arrays need a substation, and in deep water that substation must float. The pilots avoided the problem by being small or by exporting to an adjacent platform. A floating high-voltage substation is a large moored structure carrying transformers, switchgear and dynamic cable terminations in a moving environment, and as of early 2026 no commercial example was in service.

Established Ports, ahead of everything else. A floating wind assembly port needs deep water at the quay, quay bearing capacity measured in tens of tonnes per square metre, tens of hectares of laydown, no air-draft restriction between the quay and open sea, and heavy-lift or self-propelled transporter capability. Very few ports in Europe, and fewer in North America, meet all five. The upgrades are civil megaprojects with their own permitting and cost-overrun distributions, and they must be financed before the wind projects that would use them reach financial close.

Established Mooring and anchor-handling vessel capacity. Floating wind escapes the jack-up bottleneck that constrained fixed-bottom construction and inherits a different one: high-bollard-pull anchor handling tugs, which are a globally traded fleet in direct competition with offshore oil and gas. A large array requires dozens of anchors and hundreds of mooring line segments, all installed in weather windows.

Frontier Weather-window-limited intervention. The scheduling literature for offshore wind maintenance treats forecast horizons of up to about six weeks as the planning object, which is the honest statement that offshore work is weather-limited rather than equipment-limited. Floating units add a motion criterion for personnel transfer on top of the wave-height limit, narrowing the accessible window further, and every operating-cost model contains an assumed access rate that has not been measured at fleet scale.

Frontier Cost of capital, which for this technology is a technical parameter. Levelised cost for a capital-intensive plant is roughly proportional to the annuity factor, so a two-percentage-point difference in weighted average cost of capital moves floating wind’s levelised cost by something like a sixth. Lenders price unproven moorings and cables into that rate. The consequence is circular and is the field’s core trap: the cost falls when the technology is proven, and it is proven by projects that must be financed at the higher cost.

Established Port and vessel capacity that this brief does not own. The assembly quay, the heavy-lift capability and the vessel fleet are treated in future ports and shipping, and floating wind is one of several industries bidding for the same scarce waterfront. If those upgrades do not happen on schedule, no amount of turbine or mooring progress substitutes.

Established Transmission and market access. Deep-water sites are frequently far from load and from existing landfall points, which makes floating wind dependent on the expansion planning and interconnection machinery documented in energy corridors; that brief’s finding that nobody publishes finished project costs applies with equal force to the export systems these arrays need.

Frontier A marine environmental evidence base. Consenting in deep water requires characterisation of collision, displacement, entanglement and benthic effects that the monitoring literature has barely begun for moored arrays, which is the same evidentiary problem ocean engineering documents for other seabed industries.

Frontier The decisive result is an audit of hardware already in the water, and nobody has published it. Pool and publish the component-level service record of the entire global floating fleet: every mooring line, connector, chain segment, anchor, dynamic cable, bend stiffener and buoyancy module, with installation date, inspection history, failure mode where one occurred, repair duration and lost production. That dataset would convert the field’s central bankability question from an argument into a failure rate per component-year. It requires no new hardware, no new science and no new vessel. It requires operators to agree to disclose, which is why it does not exist.

Frontier The second experiment is a completed tow-to-port major-component exchange, fully costed. Disconnect a commercial floating unit, tow it in, replace a main component, tow it back and reconnect, publishing the duration, the weather downtime, the vessel spread and the total cost. Every operating expenditure model in the field contains this number as an assumption. One documented campaign would replace it with an observation.

Frontier The third is an outturn capital cost from a serial order. A commercial array of several hundred megawatts, reporting audited capital cost per megawatt and the fabrication throughput actually achieved at its yard, would test the learning-rate assumption on which every cost projection depends. The United Kingdom’s contracted floating project gives a date to watch toward the end of this decade; until it reports, the cost curve is a modelling convention.

Established Four substructure archetypes with different failure surfaces. Spars are deep, stable and simple, and need a sheltered deep-water site for upending and turbine installation, which geography supplies only rarely. Semi-submersibles have shallow draft and can be completed at a quay and towed, at the cost of more steel and more sensitivity to wave-induced motion. Tension-leg platforms have the least motion and the hardest installation, because the tendons must be pre-tensioned against anchors that take vertical load. Barges are cheap and motion-prone, mitigated by internal damping. The choice sets the port requirement, and the port requirement is usually the binding constraint.

Established Moorings fail in ways the industry already knows about from oil and gas. The named mechanisms are out-of-plane bending fatigue at the fairlead, chain wear and corrosion-fatigue in the splash zone and at the touchdown point, abrasion of synthetic lines against the seabed, marine growth adding mass and drag, and anchor creep in soft sediment. None is exotic. All of them were learned expensively on floating production units, and the analogous lesson for wind is that mooring integrity management is a continuous inspection programme, not a design activity that ends at installation.

Established The dynamic cable is the component with the least margin. It hangs in a controlled configuration — lazy wave, steep wave or free hanging — maintained by distributed buoyancy, and it must accommodate platform excursions of tens of metres while keeping curvature inside the conductor’s bend limit for decades. Its failure modes are fatigue at the hang-off, buoyancy module slippage, armour-wire birdcaging and water ingress at the terminations. Repair means locating the fault, recovering the cable and jointing at sea.

Frontier Turbine scaling works against the substructure. In fixed-bottom wind, larger turbines reduce cost per megawatt because foundation and installation costs are shared over more capacity. On a floater, the overturning moment the hull must resist scales with rotor thrust and hub height, so substructure steel mass rises steeply with turbine size. The optimal turbine for a floater is therefore not obviously the largest available, and the industry’s reflex toward ever-larger machines may be importing a fixed-bottom intuition into a different cost structure.

Frontier Inspection is moving toward robotics for the same reason as the rest of offshore. Climbing and crawling robots for blade and tower inspection exist, with demonstrated traverse speeds around 125 millimetres per second on compound curvature, and remotely operated vehicles do the subsea work. The honest framing from the maintenance literature is that the scheduling layer — how often an intervention is dispatched and in what grouping — usually matters more to cost than how fast the robot moves.

Established Ocean engineering owns the seabed and the law. Ocean engineering documents how thin the environmental evidence and institutional capacity are for deep-water industry generally, and its finding that the first commercial seafloor mine died on project finance rather than on physics or law is the same failure mode this brief identifies for floating wind.

Established Ports are the shared bottleneck. Future ports and shipping establishes that the binding constraint on maritime decarbonisation is additional zero-carbon electricity and the infrastructure to land it; floating wind is one of the few technologies that could supply it, and is simultaneously competing for the same quay space.

Frontier Very large floating structures share the hydrodynamics. The response and fatigue analysis literature developed for floating cities and offshore platforms is the same body of method, and its central lesson — that connection fatigue is the life-limiting problem in multi-module floating systems — transfers directly to mooring and cable terminations here.

Established Seabed leasing in deep water is a different instrument from shallow-water leasing. Fixed-bottom offshore wind grew inside regimes designed around defined shallow seabed parcels. Deep-water leasing has to allocate much larger areas with mooring spreads that sweep the seabed, in jurisdictions where the offshore energy legal machinery was written for oil and gas. The legal questions have largely been litigated already in that sector, which is an advantage; the political questions about who gets the lease and on what terms have not.

Frontier Auction design has been the decisive policy variable and it has been set badly twice. An administrative strike price set below developers’ costs produced an empty round in 2023; a ring-fenced floating allocation produced a single expensive contract in 2024. Neither outcome is a technology result. Both are evidence that the instrument governments use to buy this electricity is immature, and that a floating industry can be created or stalled by a parameter set in a spreadsheet.

Established Standards exist and are the quiet success of the field. Classification-society standards for floating wind turbine structures, position mooring and subsea power cables give designers, insurers and lenders a common basis, and their existence is why a project can be financed at all in the absence of service history. Their limitation is the same as their strength: they encode analysis and testing conventions, and a convention is only as good as the fleet data eventually used to calibrate it.

Frontier Fisheries displacement is the sharpest distributional conflict. A moored array with large spreads and dynamic cables is materially harder to co-locate with mobile fishing gear than a fixed-bottom array, and the exclusion is effectively permanent over the lease life. The affected industries are small, local and politically organised, and the benefits of the electricity are diffuse and national. That asymmetry, not the ecology, is what usually decides deep-water consenting fights.

Frontier Entanglement and collision risk are plausible and poorly measured. Mooring lines and dynamic cables create a taut, mid-water structure that did not previously exist in the water column, with a recognised theoretical risk of secondary entanglement when derelict gear accumulates on it. The monitoring programmes attached to the existing pilots are small, and the evidence base is thin enough that both proponents and opponents can currently assert what they prefer.

Established The subsidy question deserves to be asked plainly. The floating contracts awarded so far are priced at multiples of the alternatives, and the difference is paid by electricity consumers or taxpayers. That is a defensible industrial policy if the purpose is to buy down a cost curve and the buy-down is measured. It is an indefensible energy policy if the cost curve does not fall, and the only way to tell the two apart is the outturn cost reporting this brief asks for.

Frontier The upside is geographic rather than quantitative. Floating wind does not offer cheaper electricity than onshore wind or solar; it offers high-capacity-factor electricity to coastlines that have no other large domestic option. For Japan, Korea, California and much of southern Europe, that is the difference between importing energy and generating it, which is a strategic rather than an economic argument and should be assessed as one.

Speculative If serial fabrication works, the second-order effect is an offshore construction industry. Yards, quays, mooring supply chains and anchor-handling fleets built for floating wind are the same assets other deep-water industries need, including aquaculture, floating substations for interconnectors and eventually floating platforms for other purposes. The infrastructure is more general than the application, which is the usual way an industrial base arrives.

These horizons track the industrial system — ports, yards, fleets and cost of capital — because the engineering questions are closer to settled than the industrial ones.

• 10 yr: Frontier First commercial arrays of a few hundred megawatts each in Scotland, France, Norway, Korea and possibly California; global installed capacity plausibly in the low single-digit gigawatts; the first audited outturn costs and the first multi-year fleet reliability data published, settling most of the arguments in this brief.
• 25 yr: Speculative Tens of gigawatts if serial fabrication and port investment both deliver, with levelised cost converging toward the fixed-bottom range in the best wind regimes; floating substations in routine use; tow-to-port maintenance largely displaced by offshore exchange where vessel capability allows.
• 50 yr: Speculative Floating wind as the default form of offshore wind in deep-water markets, with the industry’s constraint having moved from fabrication to transmission and system integration; retirement and repowering of the first generation providing the first genuine end-of-life data.
• 100 / 250+ yr: Handwave Proposals for far-offshore arrays feeding hydrogen or ammonia production on floating platforms rest on cost assumptions for both technologies that nobody has demonstrated, and on a marine governance regime for the high seas that does not exist; the brief declines to project past the point where the constraints stop being measurable.

• Depends on the quay, vessel and heavy-lift capacity treated in future ports and shipping, the transmission expansion and interconnection machinery in energy corridors, and the deep-water consenting and environmental evidence base in ocean engineering; nothing on the map blocks the substructure itself.
• Requires (not on this map) fabrication yards turning out dozens of identical hulls a year rather than a handful; quays with the draft, bearing capacity and air-draft clearance to assemble and launch them; a measured failure rate per dynamic-cable-year from a real fleet rather than an analysis; a financing rate that does not penalise unproven moorings into uncompetitiveness; an auction design that clears a floating round without either emptying it or paying a multiple of the alternatives; and anchor-handling vessels available when the offshore oil market is bidding for the same hulls.
• Enables large-scale offshore wind in Japan, Korea, California, the Mediterranean and the deep Atlantic margin; high-capacity-factor supply to coastal load centres with no onshore siting room; and an offshore fabrication and mooring supply chain other deep-water industries can use.
• Adjacent to ocean engineering for seabed law and environmental evidence, floating cities for the hydroelastic and connection-fatigue methods, and ocean thermal energy conversion for the cautionary case of an ocean energy technology that never left demonstration.

Established “Floating wind has capacity factors above fifty per cent.” One five-turbine array in an exceptional Scottish wind regime has, on its operator’s reporting. That is a true statement about Hywind Scotland and not a fleet property. Capacity factor is set by the wind resource and the availability of the machine; floating access to deeper, windier water raises the first and the maintenance-access problem threatens the second. A fleet number will be lower than the best pilot, and nobody yet knows by how much.

Frontier “Floating is cheaper because it needs no jack-up vessel.” It is cheaper in that line item and more expensive in several others: substructure mass, moorings and anchors, dynamic cabling, and a port specification that barely exists. The quayside assembly advantage is real and it is the strongest structural argument in the field. It is not, on any current evidence, larger than the offsetting costs.

Handwave “Costs will follow the fixed-bottom curve.” Fixed-bottom offshore wind fell in cost over two decades through volume, turbine scaling, competitive auctions and a purpose-built vessel fleet — and then rose again in 2021–2023 on input costs and interest rates. Assuming the decline transfers to a technology with a different cost structure, a different scaling relationship and one three-hundredth of the installed base is an assertion, not an extrapolation. The brief flags it handwave because the assumption is usually stated without any mechanism attached.

Established “Fukushima proved floating wind does not work.” Fukushima FORWARD proved that a set of one-off prototypes, including an unusually large turbine on a novel platform, in a typhoon-exposed site with a difficult maintenance logistic, could not be made economic. The European spar and semi-submersible pilots using series turbines have run for years at high availability. The correct reading is that the demonstration failed and the concept did not, which is also the reading that makes the Japanese result useful rather than rhetorical.

Frontier “Bigger turbines will fix the economics.” On a floater the hull must carry the overturning moment, so substructure mass rises with turbine size in a way it does not for a monopile driven into the seabed. There is an optimum, it is probably not at the largest turbine on the market, and the industry has not published the analysis that would locate it.

Speculative “The deep-water resource is effectively unlimited, so the potential is enormous.” Technical resource is not deliverable capacity. The constraint is grid connection points, port throughput, mooring vessel availability and consenting, all of which are finite and local. A resource estimate that does not subtract these is a statement about wind, not about electricity.