In-Space Servicing and Depots

In-space servicing and depots is the logistics layer of orbit: the vehicles that inspect, move, repair and refuel spacecraft already flying, the storage that would hold propellant between deliveries, the fittings through which anything transfers, and the contracts that pay for all of it. Three neighbouring briefs each assume this layer and none owns it. Orbital Shipyards records the servicing vehicles as flown hardware and hands the fluid problem across its own seam; Deep Space Infrastructure carries depots in its proposed column and names spacecraft-to-spacecraft cryogenic transfer as never demonstrated; Space-Based Manufacturing assumes platforms that get tended and resupplied. This brief owns the joint question: who can be serviced, what can be transferred, how long it can be stored, through which interface, and under what contract.

Established The organising device is a grid with two axes: how prepared the client is, and how hard the fluid is. Prepared clients carry ports, plates or fixtures designed for the visit; unprepared clients were never meant to be touched again; uncooperative clients are dead or tumbling. Storable propellants — hydrazine and its relatives — sit still and keep; cryogens boil. Every operational success in the record sits in the same corner of that grid, and every cancellation and failure sits diagonally opposite.

Established The routine corner is prepared-client, storable-fluid, and it is nearly half a century old. Progress freighters have pumped storable propellant into stations through purpose-built ports since February 1978. The commercial life-extension business that actually completed a contract never transfers fluid at all: it docks and stays, substituting its own thrusters for the client’s empty tanks.

Established The frontier corner is unprepared clients and cryogenic fluids, and the record there is a cancellation, a hardware failure, and one partial milestone. NASA’s flagship attempt to refuel an unprepared satellite was cancelled in March 2024 with about $1,400 million spent. The one flight experiment in long-duration small-scale cryogen storage lost its cryocooler and vented its methane. The one cryogenic transfer ever performed in orbit moved liquid oxygen between two tanks inside a single vehicle.

Frontier The framing under test is that orbital logistics is becoming an operating layer rather than a sequence of one-off demonstrations. The evidence for is a delivered five-year servicing contract, a second vehicle in transit, and a funded cryogenic flight-test portfolio. The evidence against is that the sector’s own April 2026 assessment still describes bespoke, one-off contracts, and that the transfer every deep-space architecture requires has not been attempted between two spacecraft.

Established Storable-propellant refuelling is not an open problem; it is an operating practice with a 1978 start date. Progress 1 docked with Salyut 6 in January 1978 and transferred propellant into the station’s tanks that February — the first orbital refuelling — and the practice has run continuously since: Progress vehicles refuel the International Space Station’s Zvezda module through dedicated fluid connectors, and China’s Tianzhou freighters have refuelled Tiangong since 2021. The qualifier that carries all the weight: these are prepared ports on cooperative stations, plumbed for the visit before launch.

Established Between two free-flying spacecraft, the storable transfer has happened exactly once. DARPA’s Orbital Express, launched 8 March 2007, transferred hydrazine and swapped a battery unit between ASTRO, the servicer, and NEXTSat, a client built to be serviced, across a three-month campaign of increasing autonomy. Orbital Shipyards carries the fuller account and its sharpest finding: autonomous servicing was demonstrated in 2007 and the first operational commercial servicing docking came thirteen years later.

Established The unprepared-client transfer has been demonstrated at experiment scale, on a workbench bolted to the ISS. NASA’s Robotic Refueling Mission, delivered in 2011, used the station’s Dextre robot in January 2013 to cut lock wires, remove a triple-sealed cap, and pump ethanol through the fill-and-drain valve of a mock satellite that had been built, like real legacy satellites, never to be opened again. That experiment is the proof of concept behind every unprepared-refuelling proposal since, and it remains the high-water mark: no unprepared free-flying satellite has ever actually been refuelled.

Established The cryogenic version of that experiment failed, and the failure is the field’s single flight data point on long-duration small-scale cryogen storage. RRM3, launched in December 2018, stored liquid methane on the ISS for about four months; its cryocooler failed in April 2019, the methane was vented, and the planned cryogenic transfer demonstration never occurred. A one-component failure ended the experiment — which is itself a measurement of where the technology was.

Established Commercial servicing that avoids fluid entirely is the part of this subject that works and sells. Northrop Grumman’s MEV-1 docked with Intelsat 901 in February 2020 by capturing its apogee engine nozzle — a feature shared by roughly 80% of geosynchronous satellites by accident of common design, not by standard — delivered its full five contracted years, and undocked in the graveyard orbit on 9 April 2025, the first undocking between two commercial spacecraft in geosynchronous orbit. MEV-2 docked with Intelsat 10-02 in 2021 and its contract has been extended to 2029. The follow-on Mission Robotic Vehicle, carrying DARPA’s two-armed Robotic Servicing of Geosynchronous Satellites payload, launched on 21 July 2026 and is in roughly year-long transit to GEO, where it is to install Mission Extension Pods — propulsion jet-packs sold to GEO operators — rather than transfer any fluid. Every element of the commercial record extends life by adding thrusters or pods, never by opening a valve.

Established The negative exhibit is OSAM-1, and its numbers are the most instructive in the subject. The mission — Restore-L as first conceived — was to refuel Landsat 7, a satellite launched in 1999 with no provision for servicing, using robotic arms to cut through insulation and access the fill-and-drain valve. NASA’s own Independent Review Board, reporting 29 February 2024, recorded a baseline of $626–753 million for a 2020 launch, a November 2023 estimate of $2,380 million with about $1,400 million already spent and $980 million to go, and a 70%-confidence launch readiness of March 2028. NASA cancelled the project on 1 March 2024, citing “continued technical, cost, and schedule challenges, and a broader community evolution away from refueling unprepared spacecraft, which has led to a lack of a committed partner”; about 450 employees and contractors had worked on it. Orbital Shipyards dissects the board report; what belongs here is the board’s finding that the mission was “overly focused on delivering propellant to an aged LandSat-7 spacecraft… not perceived to be of value.” The cancellation was not a verdict on servicing. It was a priced verdict on retrofitting: roughly $2.4 billion projected to put propellant into one satellite that was never designed to receive it.

Established The cryogenic record has its own cancellation, a decade earlier. NASA formulated a dedicated Cryogenic Propellant Storage and Transfer flight demonstration in the early 2010s and cancelled it in the fiscal year 2014 budget cycle, redirecting the work to ground testing. The flight portfolio returned in October 2020 as four Tipping Point awards totalling about $256 million: $53.2 million to SpaceX for a large-scale liquid-oxygen transfer demonstration, $89.7 million to Lockheed Martin and $27 million to Eta Space for small cryogenic flight experiments, and $86.2 million to United Launch Alliance. The one flown result from that portfolio is the March 2024 internal transfer: liquid oxygen moved between the header and main tank of a single Starship on its third flight test, a milestone NASA subsequently confirmed complete after data review. Frontier The spacecraft-to-spacecraft version — two Starships docked and transferring — has been NASA’s stated next milestone since 2023 and has been repeatedly deferred; no outcome report was obtainable for this brief, and none of the small demonstrations had reported flight results in the material available either.

Established China’s programme is the other active line, and it reads through tracking data rather than announcements. Shijian-21 docked with a dead Beidou satellite in January 2022 and towed it roughly 3,000 km above the geostationary belt, an operation followed by independent tracking organisations. Frontier Shijian-25, launched in January 2025 with a stated purpose of on-orbit refuelling and life-extension verification, spent mid-2025 in rendezvous with Shijian-21; commercial space-surveillance firms reported an apparent docking in geosynchronous orbit. Whether propellant actually flowed is not publicly confirmed; if it did, China performed the first satellite-to-satellite refuelling in GEO while the American programme built for the same feat sat cancelled. The claim deserves exactly that conditional.

Established Depots: none has ever flown. The flight record consists of testbeds — a small water-transfer experiment on the ISS in 2019 and Orbit Fab’s Tanker-001 Tenzing, a propellant-tank testbed launched in June 2021 — and a design literature. The best-developed depot studies, by launch-vehicle-industry authors, claim boil-off rates well under 0.05% per day using passive thermal control on existing upper-stage structures. Those are design numbers with interest running in their favour, and there is no flight measurement to set against them.

Established Interfaces are where the record is most lopsided. For crewed docking a genuine multilateral standard exists: the International Docking System Standard, published by the ISS partners in 2010, androgynous by design, and in operational use since 2020 by Crew Dragon and later Starliner. For servicing and fluid transfer nothing equivalent has been adopted. NASA’s own capability survey names the key gap as “standardized, interoperable interfaces for mechanical, fluid, power, data”, and the flown servicing market rests instead on the engine-nozzle accident. One constellation operator, OneWeb, has fitted standardized magnetic docking plates to production satellites — the first case of preparedness purchased at fleet scale before any servicer existed to use it.

Frontier Does settled cryogenic transfer scale? The engineering decomposition is written down — settle the propellant with micro-acceleration, pre-chill the receiving tank, manage the vent — but transfer efficiency at hundreds-of-tonnes scale is unmeasured, and the losses determine how many tanker flights a lunar landing needs. Public figures for that number range from roughly eight to the high teens, and the spread is itself the finding: the efficiency data that would collapse the range does not exist.

Frontier Can storage numbers survive flight? Design studies claim under 0.05% per day boil-off; the only long-duration small-scale flight experiment ended with a failed cryocooler and vented methane. Whether months-long storage at depot scale lands near the design curves or an order of magnitude worse is the open measurement on which depot economics turns.

Frontier Will a servicing interface standardise by mandate, by market, or by accident? The candidates are live: the US Space Force has been reported to favour one refuelling fitting while also flying a rival’s on its own experiments; a commercial docking-plate design is on one constellation; the incumbent capture mechanism is a nozzle nobody designed as an interface. No adoption mechanism has won.

Frontier Is there demand beyond GEO life extension? The April 2026 sector assessment finds servicing still sold as “bespoke, one-off contracts… No one is putting in for a five-mission servicing contract to GEO.” Whether any buyer — commercial or government — converts servicing from missions into a subscription is the market question the next few years answer.

Frontier Will operators pay for preparedness now against service later? A docking plate costs grams and little money at build time; OSAM-1 priced the alternative at billions. The economically rational move is obvious and the adoption record outside one constellation is nearly empty, which makes the client-preparation rate a better leading indicator of this industry than any servicer announcement.

Speculative Where does the first real depot go? Low Earth orbit staging, cislunar aggregation and GEO fuel caches all have paper advocates; the placement question is unresolved because it is downstream of transfer and storage numbers nobody has measured. Until they exist, depot architecture comparisons are arithmetic on assumptions.

Established The first bottleneck is cryogenic fluid management in microgravity, and NASA’s own programme reporting names it as unfinished. The servicing programme’s fiscal-year 2024 account states that in-space fluid transfer technology shortfalls still exist and are actively being addressed, with development efforts remaining — language published while the flagship carrying that hardware was being cancelled. The full engineering decomposition exists in published guidelines; the flight column next to it is one internal transfer.

Frontier The second is storage. Every depot concept stands on boil-off rates that come from thermal models and ground tests. The single relevant flight experiment, RRM3, is a negative result: a cryocooler failure ended it in four months. Zero-boil-off storage has been demonstrated on the ground; no orbital tank has ever held cryogens for the durations a depot requires, at any scale.

Established The third is interfaces, and it is the gap the institutions themselves name. NASA’s capability survey and the 2022 National Strategy both identify standardised, interoperable mechanical, fluid, power and data interfaces as underdeveloped. The operating market’s workaround is the nozzle accident; the crewed-docking world shows a standard is achievable when agencies co-write one and procurement enforces it.

Established The fourth is demand. The 2022 National Strategy lists “lack of clear demand signals from government to stimulate private investment” among its three barriers; four years later the sector assessment still finds one-off bespoke contracts. Life extension has customers because dying GEO satellites have quantifiable revenue; refuelling and depots serve missions that mostly do not exist yet, which is the circularity every infrastructure business faces.

Established The fifth is the economics of the unprepared client, now priced by a real programme. RRM proved robotic access to sealed valves is feasible; OSAM-1 established what feasibility costs when engineered into a full mission: a $2,380 million projection to deliver propellant to one legacy satellite. The community response, recorded in NASA’s cancellation language, was not to try harder but to stop refuelling unprepared spacecraft — shifting the burden from heroic servicers to prepared clients.

Frontier The sixth is authorisation. Rendezvous, capture and fluid transfer between commercial spacecraft sit awkwardly in national licensing regimes built for launch and communications; the US mission-authorization question remains unsettled in the material available here, and an industry consortium’s voluntary practices are the closest thing to rules of the road.

Frontier The scientific dependency is two-phase cryogenic fluid behaviour in microgravity at transfer scale. Ground models of sloshing, stratification, chill-down and vent behaviour need flight anchoring; the March 2024 internal transfer is the first data point at large scale, and one point does not validate a model family. Radio-frequency mass gauging — knowing how much liquid a tank holds when the liquid will not sit still — is the named metrology gap for unsettled operations.

Frontier Cryocooler power and reliability set the ceiling on active storage. Zero-boil-off requires coolers running for years at depot heat loads; the one flight test at experiment scale ended in cryocooler failure, and hydrogen-class temperatures are harder than the methane that failed.

Established The rendezvous and robotics record this layer stands on is owned by Orbital Shipyards — the DART collision anatomy, the abandoned autonomous-capture demonstration, the flown commercial dockings — and nothing there blocks servicing of cooperative clients, which is precisely why the fluid and storage gaps stand out as the binding ones.

Established Launch cadence and price are inherited constraints. Depot arithmetic is done against roughly $1,000 per kilogram to orbit in the serious cost literature, and a tanker campaign is a cadence bet; Space-Based Manufacturing records cadence as its own binding constraint from the other direction.

Established Independent space surveillance is an unplanned dependency that the record shows functioning. The claims in this brief about Chinese GEO operations rest on commercial trackers, not participants; verification of who docked with whom, and arguably of whether fuel flowed, is becoming a service the servicing world cannot operate without.

Frontier The decisive test is spacecraft-to-spacecraft cryogenic propellant transfer, and it is the single result that would most change the assessment this brief makes. The scheduled version is the Starship-to-Starship demonstration under NASA’s Human Landing System programme: two vehicles docking in low Earth orbit and moving liquid oxygen and methane between them at architecture-relevant scale. NASA’s public planning has placed it within a year or two of this writing since 2023, and it has slipped repeatedly; no outcome report was obtainable for this brief. A success converts depots from design literature into engineering; a failure mode discovered at scale would re-price every architecture that assumes tanking, including the lunar landing NASA has contracted.

Established The milestone already banked bounds the problem from below. Liquid oxygen moved between two tanks inside a single Starship in March 2024, under a $53.2 million Tipping Point contract, with NASA confirming the milestone after data review. Internal transfer exercises the fluid physics but not the docking, the coupler, or the thermal interface between two vehicles — which is to say it retired the easiest third of the risk.

Frontier The storage experiment is well-posed, small, and overdue: fly a tank and measure boil-off for a year. The funded small demonstrations — the Lockheed Martin and Eta Space cryogenic flight experiments from the 2020 awards — are exactly this test, and none had reported flight results in the material available. A measured percent-per-day figure, held for months, against the under-0.05% design claims, is the number depot economics is waiting for.

Established The interface experiment costs less than any of the above and would matter as much: one client fitting used by two different operators’ vehicles. Orbital Shipyards names the docking version of this test; the fluid version — two tankers, one port — is what would turn a fitting into a standard, and the Space Force refuelling experiments flying rival fittings make the absence of a common one a live procurement fact rather than an abstraction.

Frontier A natural experiment is already running in geosynchronous orbit. If Shijian-25 transferred propellant to Shijian-21, the receiving satellite’s subsequent manoeuvre budget is observable by the commercial trackers that watched the docking; sustained Shijian-21 operations beyond its plausible residual propellant would be the confirmation the announcements have not provided. The verification regime for orbital refuelling may end up being telescopes rather than telemetry.

Established The market experiment has a date attached: pod installations and the first multi-mission contract. The Mission Robotic Vehicle’s arrival in GEO and its first Mission Extension Pod installations will test whether life extension scales beyond bespoke docking missions; the sector’s own criterion, from the April 2026 assessment, is whether anyone signs a five-mission servicing contract. Until one exists, the honest description of this market is a sequence of firsts rather than an industry.

Established The coupler is a qualification campaign before it is a product. Published NASA guidelines enumerate what a cryocoupler must survive — burst, lifecycle, leakage, thermal and vibration environments, with hydrogen embrittlement and oxygen-impact hazards assessed — and distinguish settled transfer, simplified by micro-acceleration from thrusters or venting, from unsettled transfer, where the primary risks are venting liquid overboard and losing liquid at the outlet, and where the receiving tank must be pre-chilled below a determined target. Every clause of that decomposition is a test with no flight heritage.

Frontier Storage engineering divides into passive and active, and only passive has any orbital pedigree. Multilayer insulation, sun-pointing attitudes, shading and vapour cooling are upper-stage inheritances; the sub-0.05%-per-day depot designs are built from them. Active zero-boil-off adds cryocoolers whose ground demonstrations have not been followed by flight, and whose one flight relative failed on RRM3.

Established The robotics toolchain for unprepared clients exists and is proven at experiment scale. The RRM tool suite — wire cutters, cap removal, valve adapters, all operated by Dextre — showed that a sealed satellite can be opened robotically; OSAM-1’s named technical risks, a late LiDAR and the servicing arm assembly among them, show what industrialising that toolchain into a free-flyer costs.

Established Capture mechanisms span a gradient of client preparation. The androgynous International Docking System Standard serves crewed ports; the nozzle-capture mechanism serves the unprepared-but-cooperative GEO fleet; magnetic plates serve constellations that opted in at build time; nothing flown serves a tumbling, uncooperative object, a gap Orbital Shipyards records as unbridged.

Frontier Depot structure is likeliest to be an adapted upper stage rather than a new vehicle class. The design literature reuses existing tanks and launch-vehicle structures precisely to keep the depot a thermal and plumbing problem instead of a development programme — a choice that reads as a lesson learned from OSAM-1’s cost history rather than a limitation.

Frontier Metering and custody transfer are unsolved commercial engineering. Selling propellant requires gauging it in microgravity to commercial tolerances; radio-frequency mass gauging is the named technique and its accuracy at custody-transfer standards is undemonstrated, which means the gas-station metaphor currently lacks the pump meter.

Established The seams with the three sibling briefs are deliberate and load-bearing. Orbital Shipyards owns assembly as a facility and the rendezvous, robotics and capture record, including the failure anatomies; this brief owns what flows through the fittings once a vehicle is captured. Deep Space Infrastructure owns where depots would plug into networks beyond Earth orbit and the capacity arithmetic of deep space; this brief owns whether a depot can exist at all. Space-Based Manufacturing owns making products in orbit and carries the cadence and buyer constraints this layer inherits.

Frontier Asteroid Mining is the proposed supply side of the depot economy — propellant sourced off-Earth is the only scenario in which depots stop being launch-fed buffer tanks and become markets — and the dependency runs in both directions, since mined volatiles have no customer without storage and transfer.

Established Spaceports owns the ground half of the cadence constraint: a tanker campaign of eight-to-high-teens flights per lunar landing is a launch-site throughput problem before it is a fluid one.

Established Two terrestrial adjacencies do real work in this subject. Aerial refuelling is the standardisation precedent — boom and probe-and-drogue fittings shared across allied fleets turned aircraft range from a design constant into an operational variable — and industrial gas handling supplies the cryogenic couplers, chill-down practice and custody-transfer metering that orbital versions must re-derive without gravity.

Established The subject has a White House-level strategy, and the strategy diagnoses the market rather than the technology. The 2022 In-Space Servicing, Assembly, and Manufacturing National Strategy, from an interagency working group, names three barriers: insufficient coordination, underdeveloped standards, and lack of clear government demand signals to stimulate private investment. Four years on, the sector assessment’s bespoke-contracts finding says the third barrier stands.

Established NASA’s institutional record is a flagship cancelled and a capability relocated. OSAM-1’s cancellation moved the centre of gravity of American fluid-transfer work from a Goddard servicing office to the Artemis programme’s lunar-lander contracts, where cryogenic transfer is a funded milestone rather than a mission of its own. The same agency that stopped refuelling an unprepared satellite is paying to refuel a prepared one, which is the community evolution its cancellation language described, enacted in budget.

Established DARPA is the sector’s repeat institutional entrepreneur. It flew Orbital Express in 2007, seeded the CONFERS consortium in 2018 — whose recommended design and operational practices are the nearest thing servicing has to rules of the road — and put the RSGS robotic payload on the vehicle now in transit to GEO. Frontier The US Space Force has made orbital refuelling an explicit mobility goal, with small experiments flying rival refuelling fittings and reporting that suggests a preferred interface without a published service-wide mandate; a procurement mandate, if one comes, would be the strongest standardisation force the sector has seen.

Frontier The international programmes are national debris and servicing efforts with different institutional shapes. Japan’s JAXA runs a phased commercial debris-removal partnership whose inspection phase flew; ESA’s ClearSpace-1 removal contract was restructured after its intended target was itself struck by debris in 2023 — an irony that doubles as evidence for the mission class; China’s Shijian series proceeds without published programme documents, legible only through tracking. Three governance styles, one capability, no shared rules.

Frontier The regulatory gap is authorisation. No major jurisdiction has a settled licensing pathway purpose-built for rendezvous, capture and fluid transfer between commercial spacecraft; the practical regime today is voluntary consortium practices plus the transparency imposed by independent tracking, and whether formal mission authorization arrives before or after the first commercial refuelling is an open race.

Established Every capability on this page is dual-use, and the record already reads both ways. A vehicle that can dock with and move a dead satellite can dock with and move a live one; the Shijian tow and the apparent Shijian refuelling were followed by surveillance satellites and debated in exactly those terms, and no technical line separates the servicer from the interceptor. The honest framing is that intent is inferred from behaviour and transparency, not from hardware.

Established Servicing is both debris remedy and debris risk. Failed proximity operations create debris — the collision anatomy lives in Orbital Shipyards — and fluid transfer adds venting and overpressure failure modes with no flight statistics at all. A depot is additionally a concentrated failure: one tank farm in a valuable orbit is a single event away from being a debris field with a fuel load.

Frontier The market serves value, not need. Life extension sells because GEO revenue is quantifiable; debris removal, the public-good half of the same capability, has no commercial buyer and is funded, where it is funded, by agencies. The contract record is a clean natural experiment in which halves of orbital stewardship the market prices, and the answer so far is: the half with a revenue line.

Frontier Verification is being privatised by default. The public’s knowledge of who docked with what in GEO now rests substantially on commercial tracking firms; that is a working transparency mechanism, and also an accountability question, since the observers are vendors with customers rather than treaty instruments.

Frontier The stakes are the difference between missions and operations. Every spacecraft ever flown beyond low orbit has been a single-tank vehicle: its useful life and reach fixed at launch. A working servicing and depot layer converts propellant from a design constant into an operational variable — the same conversion aerial refuelling performed for aircraft — and it is the unstated premise of every architecture in this category that involves staging, reuse, or infrastructure that persists.

Established The analogy’s limits deserve equal billing. Jet fuel does not boil away in the wing, tanker fleets were paid for by defence budgets rather than markets, and the fitting standardised because one procurement authority mandated it across a fleet. All three conditions are currently absent in orbit, and naming them is more useful than the metaphor.

Speculative If cryogenic transfer and storage both work, vehicle design decouples from launch mass, and the consequences compound. Staged architectures stop being paper; Deep Space Infrastructure’s proposed column becomes buildable; off-Earth propellant acquires a market. If either fails at scale, deep space remains a single-tank domain and the category above this brief shrinks accordingly. Handwave Claims that depots multiply payload to Mars by fixed factors are arithmetic on unmeasured transfer efficiencies and boil-off rates, and should be read as advocacy until the two numbers exist.

These horizons track the capability gradient this brief maps — from the prepared-and-storable corner that works today to the depot economy that does not exist — and they move on measured results, not announcements.

• 10 yr: Frontier The decade of the decisive tests: spacecraft-to-spacecraft cryogenic transfer flies or visibly fails; the small cryogenic demonstrations return the first months-long orbital boil-off measurements; Mission Extension Pods are installed or are not; the first multi-mission servicing contract is signed or the bespoke pattern holds; the Shijian refuelling is confirmed or fades.
• 25 yr: Speculative An operating depot serving more than one vehicle family, a cross-vendor fluid interface with at least two implementing operators, and serviceability fitted to a majority of new high-value satellites — each plausible on current trajectories, none guaranteed by them, and all conditional on the 10-year results landing well.
• 50 yr: Speculative Propellant sourced off-Earth entering depot supply chains, making the logistics layer a market rather than a government-fed buffer — contingent on Asteroid Mining-class extraction that has its own unproven chain.
• 100 / 250+ yr: Handwave A solar-system logistics network of depots and servicers is the standard far-future furniture; nothing in the measured record yet distinguishes it from decoration.

• Depends on Three seams rather than blockers. Orbital Shipyards supplies the rendezvous, capture and robotics record this layer operates with, including the failure anatomies that define its risk register; nothing there blocks servicing of cooperative clients. Deep Space Infrastructure holds the deep-space placements a depot would eventually serve and documents, from its side of the seam, that the shared fluid-transfer capability remained a named technology shortfall in the servicing programme’s own fiscal-year 2024 reporting. Space-Based Manufacturing carries the launch-cadence and buyer constraints this layer inherits wholesale.
• Requires (not on this map) Five requirements, each documented rather than inferred. Flight data on two-phase cryogenic fluid behaviour at transfer scale, of which exactly one point exists — the March 2024 internal Starship transfer. A qualified cryogenic coupler and a tanker fleet to carry it, currently a written qualification checklist with no flight heritage. A cross-vendor refuelling interface standard that operators actually adopt, the gap NASA’s capability survey names in so many words. Multi-mission servicing contracts to replace the bespoke one-off awards the April 2026 sector assessment describes. And a supply chain of prepared clients — new satellites launched with serviceable ports and plates as default fittings — because the cancelled flagship priced the alternative at $2.4 billion per unprepared satellite.
• Enables The proposed column of Deep Space Infrastructure — depots, staging, assembled apertures — and the departure stages any Orbital Shipyards facility would fuel; more immediately, it is the funded precondition of NASA’s contracted lunar landing architecture, which cannot reach the Moon without the transfer this brief records as undemonstrated between spacecraft.
• Adjacent Asteroid Mining as the speculative supply side of a propellant economy, and Spaceports as the ground half of tanker-campaign cadence.

Established “Orbital refuelling has never been done.” False by forty-eight years. Storable propellant has flowed into stations through prepared ports since February 1978, and one free-flyer-to-free-flyer hydrazine transfer was completed in 2007. What has never been done is refuelling an unprepared satellite, and cryogenic transfer between two spacecraft — the precise claims are the useful ones.

Established “Starship has already demonstrated orbital refuelling.” It has demonstrated one internal liquid-oxygen transfer between two tanks of a single vehicle, in March 2024, confirmed as a NASA milestone. Docking, couplers, thermal interfaces and vehicle-to-vehicle transfer remain untested, and the demonstration that would test them has slipped repeatedly.

Established “OSAM-1’s cancellation killed satellite servicing.” The commercial servicing business completed a five-year contract, undocked, and launched its next vehicle after the cancellation. The board that recommended cancellation found the mission aimed at a target the community did not value — propellant delivery to an aged, unprepared satellite. Orbital Shipyards carries the full dissection; the transferable lesson is about client preparation, not about servicing.

Frontier “Depots are solved engineering awaiting only funding.” The design literature is real and detailed, and that is the strongest version of the claim. Against it: no depot has flown, the boil-off numbers are unflown design claims from interested authors, the one long-duration cryogenic flight experiment failed, and the metering that commerce requires is undemonstrated. A complete decomposition with no flight column is a programme proposal, not a solved problem.

Frontier “China is refuelling satellites in GEO.” What the public record supports: a stated refuelling mission, a tracked rendezvous, and an apparent docking with Shijian-21 in mid-2025, reported by commercial surveillance firms. No independent confirmation that propellant flowed has been published in the material available here. The claim may well be true; it is currently an inference from proximity, and the satellite’s future manoeuvres are the test.

Established “A docking standard exists, so interfaces are solved.” The International Docking System Standard is real, multilateral and operational — for crewed ports. Servicing and refuelling have no adopted equivalent; the flown market rests on a nozzle shared by accident, and NASA’s own survey names the interface gap in exactly those terms.

Established “MEV proved the servicing market.” It proved one vendor, one client type, and contracts renewed by the pair already docked. The sector’s own April 2026 assessment describes bespoke one-off contracts and no multi-mission awards, and the national strategy’s demand-signal barrier reads unchanged four years on. A delivered contract is an existence proof, not a market.

Handwave “Depots multiply Mars payload five-fold” and kindred fixed-factor claims. The factor depends on transfer efficiency and boil-off rates that have never been measured in flight; the public disagreement over how many tanker flights one lunar landing needs — roughly eight to the high teens — is the current error bar, and any single confident multiplier is advocacy wearing arithmetic.