Deep Space Infrastructure

Deep space infrastructure is the connective tissue every other ambition in this category quietly assumes: the antennas that talk to spacecraft, the relays that extend their reach, the navigation services that tell them where they are, and the propellant depots that would let them go further than one tank allows. It is not a propulsion technology and it produces no headline. It is the thing that runs out.

Established The framing under test is that deep space needs, and can get, permanent infrastructure. The two halves have completely different evidential status, and this brief keeps them apart throughout. The needs half is now demonstrable with a specific, documented failure rather than a projection: a single uncrewed test flight degraded the operations of a flagship space telescope, and NASA says so in a public conference paper. The can get half is mostly plans.

Frontier The organising device for this page is a three-way sort — exists, funded, proposed — because the difference between those bins is the whole topic. Things that exist and operate: the Deep Space Network, a completed optical-communications demonstration, a completed near-Earth laser relay experiment, satellite navigation signals received at the Moon, and one internal propellant transfer inside a single vehicle. Things funded with hardware or a delivery date: antennas, mostly. Things proposed: relays at Mars, a lunar navigation constellation, depots, surface towers, and spacecraft-to-spacecraft cryogenic transfer.

Established The number that characterises the supply side better than any other is a delivery date. The Deep Space Network Aperture Enhancement Project began adding six 34-metre beam-waveguide antennas in the early 2010s. Two were operational in 2014 and 2016. As of NASA's 2024 account, DSS-23 at Goldstone had a planned delivery date of April 2026. That is roughly a decade per antenna programme, and it is the actual construction rate of deep-space infrastructure.

Established The network was already close to capacity before Artemis, and that assessment is JPL's own. A capacity analysis covering thirty years of projected demand describes the Deep Space Network as operating very close to capacity in the current decade, with asset contention as the immediate management problem; the following decade requires balancing human-rated operations against robotic missions; and the 2030s demand fundamentally new capability and capacity for human Mars exploration. Its recommendations are two cross-linked radio-frequency and optical areostationary relays at Mars, additional ground antennas and optical systems, and multiple-spacecraft-per-antenna techniques to reduce contention.

Established And then the incident, which is the spine of this brief. NASA's 2024 International Astronautical Congress paper states that supporting even a single crewed vehicle in cislunar space “pushes SCaN's present-day network assets to their limit”, and that asset contention during Artemis I led to major impacts on even flagship robotic missions like the James Webb Space Telescope. Established That is a US agency reporting, in a public conference paper, that one uncrewed test flight of one programme degraded the operations of the flagship observatory. Nothing else in this subject demonstrates infrastructure scarcity so concretely, and it converts the standard projection into an outturn.

Established The load is about to multiply rather than plateau. Artemis II's Orion subsystems are expected to generate about 300 gigabytes of data on orbit during a ten-day mission. Artemis III will have two crewed vehicles operating together in cislunar space — Orion and the SpaceX Human Landing System — both requiring Deep Space Network support. Established NASA's own network-evolution paper states that demands for communications and navigation services are projected to increase and outpace the current network capacity, and proposes commercial services, dynamic scheduling, load balancing and cloud infrastructure in response.

Established The deflating historical fact is that this is a recurring condition rather than a novel one. The network has been formally forecast to exceed capacity in 1979, with a loading-forecast tool built for the purpose; in 1999, with a warning that requirements would exceed the network's capacity in its present form; again in 1999, with a revitalisation programme for aging infrastructure and a rapidly increasing number of missions; in 2000, in a paper explicitly about mitigating the network loading crunch as missions rose from 25 in 1998 to a projected 34 by 2001; in 2006, projecting three times as many links, downlink rates two orders of magnitude higher and uplink rates four orders higher; and in 2018 and 2024 in the papers above. Frontier Each time it was managed through with scheduling, multiple-spacecraft-per-antenna techniques and incremental upgrades rather than by building what the forecasts asked for. The pattern of the last forty-five years is not that deep space gets infrastructure; it is that deep space gets rationing.

Established The antenna estate is what is actually funded, and it moves at the pace stated in section 1. The Aperture Enhancement Project added six 34-metre beam-waveguide antennas across Goldstone, Madrid and Canberra, with three 80-kilowatt uplink systems planned alongside; DSS-35 became operational in October 2014 and DSS-36 in October 2016, and DSS-23 had a planned delivery date of April 2026. Frontier The Deep Space Network Lunar Exploration Upgrades affect six antennas, two at each complex, with five of the six on track for Artemis III. And the Lunar Exploration Ground Sites programme adds three initial 18-metre sites — White Sands in the United States, South Africa with SANSA, and Australia with the Australian Space Agency and CSIRO — with a further procurement begun in the first quarter of 2023 for sites four to six carrying S-, X- and Ka-band services, and an aim of establishing the first three ahead of Gateway's launch to meet a 24-hour, seven-day coverage requirement.

Frontier The architectural point in that last programme is worth naming: NASA's answer to Deep Space Network contention is partly to build a second, separate network for the Moon rather than to grow the first. A dedicated direct-to-Earth capability for the cislunar region relieves the deep-space assets by removing the lunar traffic from them, which is a reasonable response and is not the same thing as adding deep-space capacity.

Established Optical communications produced the most quotable curve in this brief, and the curve is the point. The Deep Space Optical Communications demonstration launched on Psyche on 13 October 2023 and achieved first light on 14 November 2023 at 10 million miles; on 11 December 2023 at 19 million miles it downlinked at 267 megabits per second, sending the first ultra-high-definition video from deep space; on 8 April 2024 at 140 million miles it managed 25 megabits per second while integrated with Psyche's own communications system; on 24 June 2024 at 249 million miles, a Mars-comparable distance, it managed 8.3 megabits per second; on 29 July 2024 at 288 million miles it demonstrated uplink commanding and daytime tracking; and in September 2025, at 218 million miles, it completed its 65th and final pass. Established Rate falls roughly as the inverse square of distance: 267 megabits per second at 19 million miles becomes 8.3 at 249. Optical communications is transformative in the inner solar system and degrades hard with distance like everything else.

Established The rest of the optical record is mature and is largely somebody else's topic. The demonstration's design intent was downlink rates of 0.2 to more than 200 megabits per second at 0.1 to 2 astronomical units, with a flight laser transmitter delivering up to 4.5 watts at 1550 nanometres supporting more than 100 megabits per second. NASA completed a two-year Laser Communications Relay Demonstration experiment programme in the near-Earth regime, with an extended programme still reporting results in December 2025. Frontier For Artemis, the Orion Optical Communications System provides 80 megabits per second downlink with capability up to 260, and 20 megabits per second uplink — a sixfold increase over S-band radio. The link-budget physics, the photon-counting detectors and the coding belong to Deep Space Communications; what belongs here is the capacity these rates represent and who is competing for it.

Established And the flattest statement in this brief, from a NASA Marshall guidelines document of 2025. “The storage and transfer of cryogenic propellants in space has never been demonstrated to any scale, and no one has ever transferred cryogenic propellants between two spacecraft.” Established The one milestone is real and is smaller than it is usually reported to be: SpaceX completed a tank-to-tank transfer of liquid oxygen during a March 2024 Starship flight under a NASA Space Technology Mission Directorate Tipping Point contract — an internal transfer between tanks on one vehicle, not between two spacecraft. Established Both Artemis landers depend on the capability that has never been demonstrated: Starship's Human Landing System requires liquid oxygen and methane transfer, Blue Origin's Blue Moon Mk. II requires liquid oxygen and hydrogen transfer, and both concepts of operations require in-space cryogenic propellant transfer. NASA ranks it 21st of 187 technology shortfalls.

Established The most useful thing this section can do is the three-way sort, and then the open questions that sit inside each bin.

Established Exists and operates. The Deep Space Network. The Deep Space Optical Communications demonstration, completed after 65 passes with a documented rate-versus-distance curve. The Laser Communications Relay Demonstration, whose two-year experiment programme completed and whose extension was still returning results in December 2025. LuGRE, which received GPS and Galileo signals at the Moon and characterised the lunar satellite-navigation signal environment. Orion's optical system at 80 megabits per second. And SpaceX's internal liquid-oxygen tank-to-tank transfer of March 2024.

Frontier Funded, with hardware or a delivery date. The Aperture Enhancement Project's 34-metre antennas, with DSS-23 due April 2026. The Lunar Exploration Upgrades across six antennas, five of six on track for Artemis III. The three Lunar Exploration Ground Sites at White Sands, in South Africa and in Australia. Gateway's HALO module, which has arrived in the United States from Turin and had its initial power system activated, with launch no earlier than 2027 into a near-rectilinear halo orbit. The commercial lunar relay procurement, whose request for proposals went out in the first quarter of 2023. And a cryogenic fluid management technology roadmap identifying the gaps that must close to reach technology readiness level 6. Frontier Note what that column is: it is antennas, one module, a roadmap and a procurement.

Speculative Proposed. Two cross-linked radio-frequency and optical areostationary relays at Mars. A global lunar position, navigation and timing system, at pre-Phase A, covering the lunar globe and near-Moon space to 200 kilometres altitude, with the Lunar Augmented Navigation Service as its GNSS-like architecture and a standardised S-band Augmented Forward Signal. 50-metre lunar surface towers as wide-range communication and navigation relays. Propellant depots designed to hold cryogens at boil-off rates well under 0.05% per day with passive thermal control — none built. And spacecraft-to-spacecraft cryogenic transfer, which nobody has done.

Frontier Inside those bins, the live disagreements. Position one, established: the network is at capacity and Artemis will break it without new infrastructure. Three independent NASA and JPL papers say so, and one of them cites a specific degradation of a flagship mission. Frontier Position two: building a separate lunar network is the right response to deep-space contention. This is NASA's actual strategy — dedicated ground sites, commercial relays, a lunar interoperability specification — and it is a reasonable answer that leaves the deep-space capacity problem exactly where it was.

Speculative Position three: commercial providers can supply lunar relay as a service without burdening NASA's direct-to-Earth systems. The procurement is explicit that providers are responsible for returning data “without creating any further burdens on NASA's DTE systems”, which is a contractual acknowledgement of scarcity. One company has announced an effort to deploy a constellation of lunar communication relays. An announced effort is not a service.

Frontier Position four: optical communications solves the deep-space bandwidth problem. The demonstration's own curve says it solves the inner-system problem: 267 megabits per second at 19 million miles, 8.3 at 249. Frontier Position five: spacecraft-to-spacecraft cryogenic transfer is achievable on Artemis timescales. Two commercial lander architectures assume it, a NASA roadmap targets it, and it has never been demonstrated by anyone. Frontier Position six, and the interesting dissent: the right architecture avoids cryogenic depots entirely. An updated human Mars ascent vehicle concept removed cryogenic propellant concepts in favour of storables to minimise technology investment — at least one NASA architecture study routing around the depot dependency rather than through it, which is interest running against the consensus and is weighted up accordingly.

Frontier And position seven, which is about the topic itself: deep-space infrastructure is an integration and logistics problem rather than a physics one. That is true, and the fetched record sharpens it into something more uncomfortable: the integration problem is decade-scale per antenna programme, which means the binding constraint is institutional throughput rather than engineering difficulty.

Established The first bottleneck is antenna time, and it is being rationed now rather than prospectively. Asset contention is the immediate management problem in JPL's own capacity analysis, and the consequence has already been observed: Artemis I contention producing major impacts on flagship robotic missions. Established The mitigation techniques — scheduling, multiple spacecraft per antenna, load balancing — are how the network has absorbed every previous capacity crisis since 1979, and they are the reason the crises have been invisible until now.

Frontier The second bottleneck is construction rate, and it is the number this brief most wants a reader to take away. Six 34-metre antennas: two operational in 2014 and 2016, one with a planned delivery date of April 2026. Roughly a decade per antenna programme. Against that, Artemis III will fly two crewed vehicles simultaneously, both on the same network, and Artemis II alone is expected to generate about 300 gigabytes in ten days.

Established Third, distance, which no technology removes. Optical communications improves the inner-system link by a large factor and then degrades as the inverse square like everything else: 267 megabits per second at 19 million miles, 8.3 at 249. The bandwidth problem at Mars distance is not solved by the demonstration that is usually said to have solved it.

Established Fourth, and the one that gates every architecture rather than just the communications: cryogenic propellant transfer. It has never been demonstrated to any scale, and no one has ever done it between two spacecraft. Frontier The technical decomposition is well understood and that is precisely what makes the absence notable: settled transfer uses micro-acceleration from thrusters, settling motors or propulsive venting to position the propellant and is described as a simplified operation; unsettled transfer is harder, with the primary risks being venting liquid overboard and failing to keep liquid at the outlet port, and it requires pre-chilling the receiving tank below a determined target temperature. Cryocouplers must be qualified for burst, lifecycle, leakage, thermal and vibration environments, with hydrogen embrittlement and oxygen-impact hazards assessed, and radio-frequency mass gauging is the named technique for unsettled gauging. A capability with a complete engineering decomposition and no demonstration is a programme problem, not a science problem.

Frontier Fifth, boil-off, which is the depot version of the same constraint. A thermally optimised long-duration low-orbit cryogenic depot design achieves boil-off rates well under 0.05% per day using current passive thermal control and existing launch-vehicle structures, and settled transfer with low vehicle acceleration significantly simplifies the operation, with key fluid-management technologies already implemented on upper stages. Frontier Those figures come from authors in the launch-vehicle industry, which is interest running with the finding, and they are design-study numbers rather than flight results. Every depot in the fetched record is a design.

Speculative And sixth, servicing and assembly, which is where this brief's dependency lives. The on-orbit servicing programme's own reports through fiscal year 2024 state that in-space servicing, assembly and manufacturing fluid transfer technology shortfalls still exist and are actively being addressed, and describe unfinished development efforts remaining and technology gaps in key areas of in-space fluid transfer. Frontier The programmatic status of that mission is not stated in any source obtained for this rewrite and is not asserted here — but the technology-gap language is quotable as it stands, and it reads like a post-mortem.

Frontier This brief records one dependency, on FR-I-24, and the reason is the proposed column. Nothing in it — areostationary relays at Mars, a lunar navigation constellation, propellant depots, assembled large apertures — gets built at scale without orbital assembly and servicing. The seam between the two briefs is drawn explicitly: that brief owns the manufacturing process; this one owns the servicing and assembly infrastructure programme.

Frontier The dependency is not abstract. It has a document. The servicing programme's own fiscal-year 2024 reporting states that fluid-transfer technology shortfalls still exist and are actively being addressed, with unfinished development efforts remaining. The infrastructure this brief describes and the manufacturing capability it depends on are blocked at the same place, on the same fluid-transfer problem, in the same programme office.

Established A second dependency runs to launch and is a price rather than a result. Every element in the proposed column is priced per kilogram to its destination, and the cost anchor the serious literature uses is about $1,000 per kilogram at a $100 million launch baseline. No laboratory produces that number and no brief on this map owns it, which is why it appears here as context rather than as an edge.

Frontier And a set of dependencies this brief deliberately hands away. Deep Space Communications owns communications as engineering: optical link budgets, photon-counting Geiger-mode detector arrays, coding, delay-tolerant networking protocols and relay architectures. This brief owns communications only as capacity — how much exists, who is competing for it, and what happens when it runs out. The photon-efficiency figures and the detector arrays belong there; the Artemis I and JWST contention belongs here, because it is a scarcity fact rather than a link-budget one.

Established The decisive experiment in this brief has already run and it was not designed as one. Artemis I flew, the network could not serve everything at once, and flagship science was measurably degraded. That is a capacity experiment with a published result, and it is worth more than any modelling study of network loading. The follow-on is scheduled: Artemis III will fly two crewed vehicles in cislunar space simultaneously, both requiring the same network.

Established The optical experiment has also run to completion, and its result is a curve rather than a headline. Sixty-five passes between November 2023 and September 2025, from 10 million to 288 million miles, with peak downlink falling from 267 to 8.3 megabits per second as distance grew. Frontier The demonstration is over. Describing it as ongoing, as an earlier version of this page effectively did, misses that its final pass was in September 2025.

Established The experiment that has not been done is the one every architecture assumes. Spacecraft-to-spacecraft cryogenic propellant transfer, at any scale, by anyone. The one completed milestone was an internal liquid-oxygen tank-to-tank transfer within a single Starship in March 2024 under a NASA Tipping Point contract. Frontier The intermediate experiments are enumerated in NASA's own guidelines: settled transfer under micro-acceleration, unsettled transfer with pre-chilling of the receiving tank, cryocoupler qualification across burst, lifecycle, leakage, thermal and vibration environments, and radio-frequency mass gauging for unsettled conditions. Each is a well-posed test with no flight.

Frontier The navigation experiments are the most encouraging part of the record because one of them flew. LuGRE received GPS and Galileo signals at the Moon and characterised the lunar satellite-navigation signal environment — the single flown data point behind every cislunar position, navigation and timing concept. Speculative Everything above it is architecture: an interoperability specification described by its own governance paper as a lunar internet, a GNSS-like constellation broadcasting synchronised navigation signals with a standardised S-band augmented forward signal, and a global lunar system at pre-Phase A targeting coverage to 200 kilometres altitude. One receiver on one flight, and a specification stack above it.

Frontier And the commercial experiment, which is a procurement rather than a test. The lunar relay services request for proposals went out in the first quarter of 2023, with providers responsible for returning data without creating further burdens on NASA's direct-to-Earth systems, and one company has announced an effort to deploy a relay constellation. The experiment is whether a commercial market for cislunar relay clears at all, and its result is not in.

Established The engineering requirements here are unusually well specified because most of them are procurement documents rather than concepts. Ground: six 34-metre beam-waveguide antennas with three 80-kilowatt uplink systems; six further antennas upgraded for lunar exploration, two at each complex; three 18-metre lunar exploration ground sites with S-, X- and Ka-band transmit and receive, sited to give 24-hour, seven-day coverage. Delivery of one 34-metre antenna is a decade-scale undertaking, which is the requirement everyone underestimates.

Established Flight optical: a laser transmitter delivering up to 4.5 watts at 1550 nanometres supporting more than 100 megabits per second, against a design intent of 0.2 to more than 200 megabits per second across 0.1 to 2 astronomical units. On the crewed side, an optical system providing 80 megabits per second downlink with capability to 260 and 20 megabits per second uplink — a sixfold improvement over S-band radio. Frontier The ground segment for optical is the harder half and belongs to Deep Space Communications: photon-counting detector arrays and the receiving apertures to feed them.

Frontier Cryogenic fluid management is the requirement with the most detail and the least flight. Boil-off well under 0.05% per day for a long-duration low-orbit depot using passive thermal control and existing launch-vehicle structures. Cryocouplers actively or passively controlled, uni- or bi-directional, qualified for burst, lifecycle, leakage, thermal and vibration environments, with hydrogen embrittlement and oxygen-impact hazards assessed. Pre-chilling of a receiving tank below a determined target temperature for unsettled transfer. Radio-frequency mass gauging. Frontier NASA maintains a roadmap identifying the gaps that must close to reach technology readiness level 6, with model validation work for boil-off and transfer in microgravity ongoing. The requirement set is complete; the demonstration is not.

Speculative And the requirements for the proposed column, stated as what they would need rather than as what they are. Two cross-linked radio-frequency and optical relays in areostationary orbit at Mars. A lunar navigation constellation with an Earth control segment and lunar-surface monitoring, providing coverage across the lunar globe and to 200 kilometres altitude. 50-metre surface towers as wide-range relays, which is a lunar-surface construction problem rather than an orbital one and belongs to the surface-infrastructure briefs. Autonomous relay navigation using satellite signals, optical navigation and one-way Earth measurements.

Frontier The boundary that most needs drawing is with communications engineering, and it is drawn on capacity versus link. Deep Space Communications owns the technology itself: optical link budgets, photon-counting Geiger-mode avalanche photodiode arrays, coding, delay-tolerant networking, and relay architectures as engineering. This brief owns communications only as a scarce resource. The photon-counting detector physics belongs there; the Artemis I contention with the James Webb Space Telescope, the decade-per-antenna construction rate, and the rate-versus-distance curve belong here, because they are facts about scarcity rather than about signalling.

Frontier The dependency relationship with orbital industry is the other structural boundary. Space-Based Manufacturing owns the factory and the manufacturing process; this brief owns the roads, the refuelling and the relays that would let a factory operate beyond low orbit. Orbital Shipyards owns assembly as a facility. Frontier The servicing and assembly infrastructure programme sits on the seam and is claimed by this page, because its binding problem — in-space fluid transfer — is the same one that blocks the depots.

Established Three further adjacencies own things this brief stops short of. Lunar surface facilities, landing pads and habitats — including the 50-metre surface towers — belong to the surface-infrastructure briefs and to Spaceports; this page stops at orbit and at the ground stations on Earth. Advanced Nuclear Propulsion owns the vehicles that would use the depots. Frontier And Mega Telescopes owns the instrument whose degradation is this brief's central exhibit — a useful reminder that infrastructure scarcity is felt by science missions rather than by the infrastructure.

Speculative One adjacency is governance rather than engineering. Spectrum for a lunar network is being negotiated internationally, and the interoperability framework exists precisely so that government, international and commercial providers can operate on the same architecture. Space Law and Governance owns that; it is worth naming because the constraint on a lunar relay network may turn out to be an allocation rather than an antenna.

Established The institutional finding here is the most transferable thing on the page: a capacity crisis that is managed rather than solved becomes invisible, and stays invisible until it damages something famous. Six formal forecasts of network over-capacity across forty-five years, each absorbed by scheduling and incremental upgrade. What changed with Artemis I is not that the network became short of capacity but that the shortage produced an effect on a mission nobody could overlook.

Established The second institutional fact is the construction rate, and it is a fact about organisations rather than about engineering. A six-antenna programme announced in the early 2010s delivered two units in 2014 and 2016 and had a third due in April 2026. Nothing about a 34-metre beam-waveguide antenna is technically novel. The decade is procurement, siting, funding cycles and integration, which means the binding constraint on this subject is institutional throughput.

Frontier Third, the response to scarcity has been to build a second network rather than to grow the first, and to buy service rather than to own assets. Dedicated lunar ground sites, an interoperability specification, and a commercial relay procurement whose terms require providers to return data without further burdening NASA's direct-to-Earth systems. Speculative That contractual clause is the clearest institutional acknowledgement of scarcity in the entire record, and it transfers the capacity problem to a supplier rather than resolving it.

Frontier Fourth, Gateway is the one element with flight hardware in hand and a stated launch year, and its funding position should not be smoothed over. The HALO module has arrived in the United States from Turin, is described as the first long-term home for astronauts in lunar orbit, has had initial power system activation with final outfitting underway, and is to launch no earlier than 2027 into a near-rectilinear halo orbit with Canadian, European, Japanese and Emirati partners. Speculative NASA's own page carries a live notice that it is being updated to align with recent programme and national space policy announcements, and gives no funding figures. Hardware exists; budget stability is not asserted here because no source obtained for this rewrite establishes it.

Established The distributive question in this brief is unusually concrete: a scarce public asset was allocated, and science lost. Artemis I contention produced major impacts on flagship robotic missions including a space telescope. That is a resource-allocation decision with an identifiable loser, made inside a scheduling system rather than through any public process, and it will recur with two crewed vehicles on the network at Artemis III.

Frontier Second, the governance of a commercial layer. Moving lunar relay to commercial providers under a requirement that they not burden government assets transfers a capacity problem into a contract. Speculative If the commercial market does not clear — one announced effort is the current state of it — the capacity returns to the public network with the missions that were planned around it already committed. That is a real programme risk and it is not an engineering one.

Frontier Third, international dependency, which cuts both ways. The network's complexes are in the United States, Spain and Australia; the new lunar ground sites add South Africa and Australia through national agencies; Gateway's partners span four space agencies; and lunar spectrum is being coordinated internationally. Deep-space infrastructure is unavoidably multilateral because the Earth rotates, and that is a stabilising property of the field worth naming.

Established Fourth, and about this page's own claims. An earlier version described record data rates from tens of millions of kilometres and treated in-space refuelling as underway with early transfer demonstrations. The optical figures are now given as a full curve with its final pass in September 2025, and the refuelling claim is corrected: the one completed transfer was internal to a single vehicle, and NASA states plainly that no one has ever transferred cryogenic propellants between two spacecraft. Where a source says something this blunt about its own programme's readiness, repeating it exactly is the whole job.

Established The civilizational claim usually made here — that infrastructure is the difference between expeditions and a spacefaring civilisation — is true and is not what the evidence is about. What the evidence is about is that the infrastructure which exists is being consumed faster than it is being built, and that the build rate is roughly one antenna programme per decade.

Frontier The forty-five-year pattern is the deflating part and should be stated rather than avoided. Capacity has been formally forecast to run out in 1979, 1999, 2000, 2006, 2018 and 2024, and each time it was managed through with scheduling and incremental upgrades rather than by building what the forecasts asked for. Frontier The historical base rate is not that deep space gets infrastructure. It is that deep space gets rationing, and rationing works until two crewed vehicles and a flagship observatory want the same antenna.

Speculative The one genuinely new thing in the record is the shift from public asset to purchased service. A dedicated lunar network, commercial relay procurement and an interoperability specification that admits government, international and commercial providers on equal terms describe a different arrangement from a single agency-owned network. Speculative Whether that produces more capacity or merely relocates the shortage is the open institutional question of this subject, and it will be answered by whether anyone actually flies a commercial relay constellation.

Established And the sentence this brief would leave a reader with. Deep space demonstrably needs infrastructure, and the evidence for that is now a documented incident rather than a projection. Whether it can get it is a different claim, and the funded column is short, late, and dominated by antennas rather than by anything in orbit.

These horizons track the exists / funded / proposed sort, because the movement of an item between those bins is the only thing that changes this brief's answer:

• 10 yr: Established Expect the funded column to deliver: DSS-23, the lunar exploration upgrades across six antennas, and the three lunar ground sites at White Sands, in South Africa and in Australia. Frontier Expect Gateway's HALO to launch no earlier than 2027, and expect Artemis III to fly two crewed vehicles on the same network, which is the stress test the 2024 papers are written about. Frontier The item to watch is cryogenic transfer: both Artemis landers require it, NASA ranks it 21st of 187 shortfalls, and nobody has done it between two spacecraft.
• 25 yr: Frontier This is the window in which a lunar navigation and communications layer either exists as a service or does not. Today it is one flown receiver, a specification stack, a pre-Phase A architecture and one announced commercial effort. Speculative Propellant depots at boil-off rates well under 0.05% per day are design studies; a flown depot in this window would be the single largest change to every architecture in this category. Speculative Areostationary relays at Mars are the recommended answer to the 2030s capacity problem and remain a recommendation.
• 50 yr: Speculative A solar-system-wide infrastructure with routine refuelling, relays at more than one destination and navigation services beyond cislunar space is the coherent long-horizon case, and every element of it currently sits in the proposed column. Frontier The historical base rate argues for something less: continued rationing, incremental antenna programmes, and architectures that route around missing capabilities the way one Mars ascent-vehicle study routed around cryogenic depots by choosing storables.
• 100 / 250+ yr: Handwave Beyond useful forecasting for a subject whose construction rate is measurable in decades per antenna. Speculative The structural observation that survives at this range is that communication and navigation capacity are shared public goods with no natural owner, and that every forty-five years of this field's history has been about who gets to use them next week.

• Depends on One dependency, on FR-I-24, and it is the proposed column that requires it: areostationary relays at Mars, a lunar navigation constellation, propellant depots and any assembled large aperture all need orbital assembly and servicing before they need anything else. The dependency is documented rather than assumed — the servicing programme's own fiscal-year 2024 reporting states that in-space fluid transfer technology shortfalls still exist and are actively being addressed, with unfinished development efforts remaining. The infrastructure on this page and the manufacturing capability it depends on are blocked at the same place, on the same problem, in the same programme office. A second dependency is a price rather than a result and is therefore recorded in prose: every element in the proposed column is costed per kilogram to its destination, against a launch anchor of roughly $1,000 per kilogram.
• Enables Everything that operates beyond low Earth orbit depends on what this brief describes, which is why no typed enabling edge is claimed to any individual brief — an edge to each would be an edge to most of the category. The concrete enabling relationships are the ones the record documents: Mega Telescopes depends on network capacity in the most direct way possible, since a flagship observatory's operations were measurably degraded by contention with one uncrewed test flight; Advanced Nuclear Propulsion and every other high-energy vehicle class depends on refuelling that has never been demonstrated between two spacecraft; and Spaceports and the lunar surface briefs take over where this one stops.
• Adjacent Deep Space Communications owns communications as engineering — optical link budgets, photon-counting detector arrays, coding, delay-tolerant networking, relay architectures — and this brief owns communications only as capacity: how much exists, who competes for it, and what happens when it runs out. Orbital Shipyards owns assembly as a facility. Space Law and Governance owns lunar spectrum coordination and the interoperability arrangements that let government, international and commercial providers share one architecture. Outside this map: ground-segment engineering, cryogenic fluid mechanics, satellite navigation, and the launch industry whose per-kilogram price sets the cost of everything in the proposed column.

Established “The Deep Space Network capacity problem is a projection.” It is an outturn. NASA states that supporting even a single crewed vehicle in cislunar space pushes the network's present-day assets to their limit, and that Artemis I asset contention led to major impacts on even flagship robotic missions like the James Webb Space Telescope. Established A degraded flagship observatory is a measurement, not a forecast, and Artemis III will fly two crewed vehicles on the same network.

Frontier “So the network is about to collapse.” Also wrong, and in the opposite direction. It has been formally forecast to exceed capacity in 1979, 1999, 2000, 2006, 2018 and 2024, and each time it was managed through with scheduling, multiple-spacecraft-per-antenna techniques and incremental upgrades. Frontier The honest statement is that the network has been in a capacity crisis for forty years and has been rationed through it every time. What changed with Artemis is that the rationing stopped being invisible.

Established “In-space refuelling has been demonstrated.” It has not. NASA's own 2025 guidelines document states that the storage and transfer of cryogenic propellants in space has never been demonstrated to any scale, and that no one has ever transferred cryogenic propellants between two spacecraft. Established The one milestone — SpaceX's March 2024 liquid-oxygen transfer — was between tanks on a single vehicle. Both Artemis landers depend on the capability that has not been shown, and NASA ranks it 21st of 187 technology shortfalls.

Speculative “Propellant depots exist.” Every depot in the fetched record is a design study, including the well-cited one achieving boil-off well under 0.05% per day with passive thermal control — a figure produced by authors in the launch-vehicle industry, which is interest running with the finding. Nothing has been built.

Established “Optical communications removes the deep-space bandwidth constraint.” It removes a large part of the inner-system constraint and then obeys the inverse-square law like everything else. 267 megabits per second at 19 million miles; 8.3 megabits per second at 249 million miles. Established “DSOC is ongoing.” Its 65th and final pass was in September 2025. The demonstration is complete, and an earlier version of this page left the impression that it was still running.

Speculative “LunaNet is a lunar network.” It is a specification — a set of requirements enabling interoperability across government, international and commercial providers, described by its own governance paper as a lunar internet. Speculative A global lunar position, navigation and timing system is at pre-Phase A. The one flown data point in the entire line is a receiver that picked up GPS and Galileo signals at the Moon.

Speculative “Commercial relay services will offload the network.” The request for proposals was issued in the first quarter of 2023, its terms require providers to return data without creating further burdens on NASA's direct-to-Earth systems, and one company has announced an effort. An announced effort is not a service, and the contractual clause is better read as an acknowledgement of scarcity than as a solution to it.

Frontier “Gateway is funded and on schedule.” Gateway is the one element with flight hardware in hand and a stated launch year: HALO has arrived in the United States from Turin, its power system has had initial activation, final outfitting is underway, and launch is no earlier than 2027. Speculative NASA's own page gives no funding figures and carries a live notice that it is being updated to align with recent programme and national space policy announcements. Hardware in hand is not budget stability, and this page asserts only the first.

Frontier “OSAM-1 completed its refuelling demonstration.” No source obtained for this rewrite states the mission's programmatic status, and none is asserted here. What the fiscal-year 2024 reporting does say is quotable as it stands: fluid transfer technology shortfalls still exist and are actively being addressed, with unfinished development efforts remaining. That language reads like a post-mortem and is carried as language rather than as a status.

Established And the framing verdict. Deep space needs infrastructure — that half is now evidenced by an incident rather than a model. Can it get it? The funded column contains antennas, one module, a roadmap and a procurement; the construction rate is roughly a decade per antenna programme; and the single most-assumed capability in every architecture has never been performed by anyone. That is not a refutation of the framing. It is the difference between a need and a plan, stated at the length the evidence supports.