Communicating with Earth when the answer may take tens of minutes

The day Houston cannot answer in time

Imagine an air leak in a Martian workshop. An alarm appears, a technician must choose between isolating a line and shutting down a pump, and the crew asks Earth for advice. Even in favorable geometry, the answer cannot return on terrestrial conversational timescales. At large separation, the radio round trip alone approaches three quarters of an hour before anyone on Earth has read the message, reconstructed the context and drafted a response. Communication delay therefore becomes an architectural constraint: the settlement needs local procedures, models, authority and data to survive between exchanges with Earth.

NASA currently gives an Earth–Mars one-way light time of roughly 3 to 22.4 minutes depending on planetary geometry. The arithmetic is already decisive: a three-minute one-way delay means at least six minutes round trip; 22.4 minutes means 44.8 minutes. Acquisition, routing, queues, human interpretation and response preparation come on top of that. This is not simply “the Internet with lag.” It is an operating culture in which immediate causality must remain local.

NASA’s Autonomous Systems and Operations work treats autonomy as a mission capability rather than a software convenience for exactly this reason. A Mars crew cannot wait for terrestrial flight controllers to approve every action. Earth remains enormously valuable for expertise, deferred analysis, strategic planning, tele-science and updates. Immediate safety, vehicle control, fire response and fault isolation, however, must be designed to work through periods of silence.

Machines face the same constraint. A rover seeing an obstacle, a power network sensing an overload or life support detecting rising carbon dioxide needs a loop far faster than interplanetary light time. Communications architecture therefore cannot be separated from autonomy, control, cybersecurity and operational governance.

The useful question is not merely how many minutes separate Mars from Earth, but which decisions become impossible to delegate once delay exceeds the dynamics of the event. A valve may need to close in seconds, a rover may brake in fractions of a second and a leak may need isolation within minutes, while a complex medical consultation may tolerate delayed exchange. A settlement therefore needs decision horizons: fast physical loops remain local, mission-level choices can be shared, and strategic decisions can wait for Earth. Light-time becomes an architecture of responsibility.

Delay also changes documentation. A Martian crew cannot depend on procedures that exist only in terrestrial cloud services. Electrical drawings, maintenance history, configuration records, emergency medical knowledge, maps, software, manufacturing models and recovery procedures must be stored locally, versioned and usable offline. At this scale, the network is not only a means of conversation; it becomes the distributed operational memory of the settlement.

A Mars network already exists — but it is only an embryo of human infrastructure

In 2026 the Mars Relay Network is an international constellation of four orbiters: Mars Odyssey, Mars Express, Mars Reconnaissance Orbiter and ExoMars Trace Gas Orbiter. Surface rovers do not attempt to send all science directly to Earth. They use proximity links to orbiters that have more power, larger antennas and better opportunities to contact terrestrial networks. NASA notes that surface-to-orbiter links can reach data rates up to about 2 Mbit/s under suitable conditions.

That architecture is already a systems lesson. A rover does not need to carry by itself the antenna, power and geometry required for every long-haul function. It delegates transport to an orbital layer. The concept resembles terrestrial networking: an end user does not communicate directly with every server on the planet; intermediate infrastructure carries and aggregates traffic. Mars therefore already has, in miniature, something resembling access networks and backhaul.

NASA’s 2026 MRN statistics also show that relays are not interchangeable. For February through May 2026, the agency lists average relay volumes of about 850.6 Mbit per day for MRO and 2,228.1 Mbit per day for TGO, while Mars Express primarily provides backup support. Those values are not a human-settlement requirement. They describe today’s robotic system and illustrate how capacity depends on orbit, pass geometry, spacecraft power, science priorities and Earth contacts.

A human presence would multiply the variety of traffic: telemedicine, maintenance video, mapping, science, manufacturing files, software, environmental monitoring, family communications, education, robotic operations and safety logs. The network can no longer be merely an opportunistic secondary function of science orbiters. It becomes infrastructure that must be intentionally sized, redundant and maintainable.

The current Mars Relay Network is already a powerful architectural lesson. Surface vehicles do not spend all their mass and energy maintaining high-volume direct links to Earth; much of their data is sent to orbiters that can relay it onward. A human settlement would need to turn that scientific arrangement into dedicated, renewable infrastructure: orbital coverage, local surface networking, mobile-vehicle access, common timing and the ability to continue after losing one relay.

The difference is traffic volume and criticality. A city would generate industrial telemetry, medical video, manufacturing files, scientific data, personal communications, local navigation and safety commands. Traffic classes would be unavoidable: life-critical command, alarm, telemaintenance, navigation, science, bulk transfer and entertainment. When bandwidth shrinks, the network must already know what it can sacrifice and what it must preserve.

The life of a data bundle: from rover to mission control

The word “relay” hides a chain of operations. A rover or lander first prepares the data it wants to transmit, considering onboard memory, available energy and the next orbital pass. When a compatible orbiter becomes visible, a proximity link can be established using frequencies suited to the short range. The orbiter receives the data, stores it, and later forwards it to Earth when geometry and scheduling allow. The data may therefore already be hours old before it reaches a Deep Space Network or partner antenna.

The Mars Relay Network Handbook published in August 2026 is valuable because it treats this network as a service with interfaces, capabilities and responsibilities. A relay is not permanently available magic: orbiters have trajectories, pointing constraints, their own science missions and shared communications resources. For human settlement, that operational reality becomes a lesson: bandwidth has to be scheduled like power and water.

MaROS: the infrastructure already has a coordination layer

The Mars Relay Operations Service — MaROS facilitates telecommunications coordination between landed and orbiting spacecraft operators. The “Martian internet” therefore begins not with antennas but with organization: who requests a pass, who resolves conflicts, how data flow is tracked, how incidents are investigated and how agencies cooperate. A human settlement would add permanent medical, safety, navigation, industrial-control and civil-communications needs to that scientific coordination.

The scale change is enormous. A rover can wait for another pass if a science image is delayed. A hospital cannot always treat clinical data the same way. A human network needs quality of service, priority, redundancy and pre-emption rules so that critical information can move ahead of bulk traffic without making the network useless for science and industry.

The Martian Internet will not be the terrestrial Internet with twenty minutes of delay

Terrestrial Internet protocols grew up in a world where an end-to-end path usually exists and a lost packet can often be requested again within milliseconds. Interplanetary space breaks that intuition. Mars can occult a relay, an orbiter can leave the horizon, optical ground stations can lose weather availability, and a high-value link can be scheduled for another mission. The complete path may not exist at the moment data are created.

Delay/Disruption Tolerant Networking, or DTN, is designed for that condition. Its central idea is store-and-forward: a node keeps a data “bundle” in persistent storage, waits for the next useful contact, then forwards it. Bundle Protocol does not assume a continuous circuit from Mars to the ultimate destination. Contact plans, priorities, expiration times and custody become first-class engineering concepts.

This changes storage design. An instrument producing 50 Mbit/s for eight hours generates 50 Mbit/s ÷ 8 = 6.25 MB/s, and 6.25 × 3,600 × 8 = 180,000 MB, or roughly 180 GB. If the transmission window disappears, those 180 GB are not an exceptional error condition; they are a physical queue that must be stored, prioritized, verified and perhaps compressed. At city scale, networking and storage are inseparable.

The network must also know what can wait. A medical alert of a few kilobytes can matter more than a terabyte of mapping already backed up elsewhere. Critical commands should not sit behind entertainment video. The quality of a Martian Internet will therefore be measured less by a peak-rate headline than by whether it preserves the right information in the right order on bad days.

Many terrestrial protocols were designed around an end-to-end path that exists during the exchange and acknowledgements that return quickly enough to guide transmission. Interplanetary space breaks those assumptions through orbital geometry, occultations, solar conjunction, pointing limits and long latency. DTN treats disruption as a normal property. A node can accept a bundle, keep it on persistent storage and forward it at the next scheduled or discovered contact. Information begins to move like cargo through a logistics network.

That immediately creates governance questions. How long is a bundle valid? Who may assign emergency priority? What happens if duplicate commands exist after recovery? How is a message authenticated after waiting for hours inside a relay? Delay tolerance therefore has to be coupled with identity, time, integrity, expiration rules and command semantics. DTN is not merely a throughput trick; it becomes part of system safety.

Will Mars have its own GPS? Position, navigation and timing become a shared service

Earth GPS works because a constellation broadcasts extremely precise timing and orbital information. Mars has no equivalent constellation today. Rovers combine odometry, inertial sensing, imagery, orbital maps, radio and estimation support from Earth. That mosaic is adequate for slow robotic vehicles under strong supervision; it becomes far more restrictive when convoys, drones, construction machines, emergency vehicles and crews move over hundreds of kilometres.

The 2024 JPL study on Positioning, Navigation and Timing at the Moon and Mars explains why some lunar concepts do not scale outward. Weak terrestrial GNSS signals can still contribute to some lunar scenarios. At Mars they become impractical. Dedicated architecture, stable clocks, orbit determination and local services are needed.

Time then becomes distance. Light travels about 299,792,458 metres per second. A timing error of one microsecond — 1 µs = 0.000001 s — corresponds to about 299.8 metres of light-travel distance. Ten nanoseconds still correspond to roughly three metres. That does not mean a navigation system inherits those exact position errors, because multiple measurements are fused, but it shows why clocks are navigation sensors.

A Martian settlement could therefore grow its own PNT capability in stages: orbiting relays with stable clocks, surface beacons around inhabited zones, shared maps, local corrections, inertial navigation, vision and radio ranging. A “Mars GPS” may emerge first as islands of service around settlements before becoming a planet-wide system.

A Martian PNT service would probably not begin as a miniature copy of terrestrial GPS. It could start with stable clocks on a few relays, radio ranging, orbital maps, local beacons and fusion with onboard inertial and vision systems. As traffic grows, the value of a shared reference grows rapidly. Autonomous rovers, cranes, pressurized convoys, drones and EVA crews all need mutually consistent positions, maps and time to avoid conflicts.

Time is as important as position. A ranging measurement derived from propagation time can become wrong when clocks diverge. One microsecond corresponds to roughly three hundred metres of light-travel distance; that does not mean a complete PNT system automatically suffers a three-hundred-metre position error, because multiple measurements and geometry are combined, but it shows why synchronization cannot be treated as a convenience. A Martian city will eventually need timing as infrastructure.

Radio, Ka-band and laser: the future will probably be hybrid

Radio remains the robust, proven tool of deep space. Deep Space Network antennas communicate with distant spacecraft particularly in X and Ka bands. Radio offers mature acquisition and degraded-mode options, but a human mission’s demand for video, science, telemedicine, mapping and large software transfers pushes toward far higher capacity.

Optical communications concentrate energy into an extremely narrow beam. NASA/JPL’s DSOC demonstration showed data rates far beyond comparable radio systems and NASA reported in 2026 that the two-year demonstration completed reliable high-rate optical communications over interplanetary distances. That does not make radio obsolete. Optical links demand precise pointing and, for ground receivers, availability can be affected by weather and atmospheric conditions.

A credible architecture therefore layers multiple media: robust radio for commands, telemetry and degraded operation; higher-rate radio when geometry and resources permit; optical links for bulk data; local fiber and radio across the surface; orbital relays to aggregate traffic. Diversity reduces dependence on a single failure mode.

The lesson is industrial too. A superb optical terminal is useless if a pointing mechanism cannot be repaired. A high-performance Ka-band amplifier that has no spare path becomes a liability. Human networks must be designed around replaceable interfaces, low-rate survival modes and long periods of reduced capability.

Radio brings a deep heritage of robust commanding, telemetry and radiometric navigation. Higher radio frequencies can provide more bandwidth but demand tighter pointing and can be more sensitive to the terrestrial atmosphere. Optical communications can provide much higher rates for comparable aperture and power, as deep-space laser demonstrations have shown, yet require extremely accurate pointing and receiving sites that cannot see through clouds.

A resilient human architecture therefore benefits from diversity rather than technological dogma. Emergency commands may favor a mature low-rate radio path; bulk science may wait for an optical opportunity; surface networks can mix fiber, radio and other media. The important word is not 'laser' or 'Ka' but diversity: different physical media remove some common-cause failures and permit graceful degradation instead of total loss.

Solar conjunction: learning to operate when Earth becomes a distant partner

When the Sun approaches the Earth–Mars line of sight, solar plasma can seriously degrade radio propagation and increase the risk of corrupted commands. Robotic missions already plan conservative operating periods around solar conjunction. A human settlement cannot treat the event as a surprise: the geometry is predictable months in advance and should trigger a prepared operating mode.

Before conjunction the settlement can preload software, medical references, maintenance procedures, manufacturing files, education content, science plans and data expected to be useful during the reduced-contact period. Non-urgent activities are rescheduled. Decisions normally escalated to terrestrial authority receive explicit temporary delegation. Automated systems can adopt more conservative configurations while remote expertise is harder to reach.

NASA studies have also considered relay geometries intended to reduce conjunction outages. This illustrates that autonomy and infrastructure are complementary levers: make Mars more capable locally, and improve the network where practical. Neither eliminates the need to assume that contact can still be lost.

The honest test of a Martian communications architecture is simple: what happens if Earth disappears for a day, then for a week? Which services continue? Which degrade? Which decisions become local? What data must already be stored on Mars? That turns telecommunications resilience into an exercise in governance.

Solar conjunction is more than a slower connection. When Mars lies near the Sun in Earth's sky, solar plasma degrades radio propagation and increases the risk of corrupted commands. Missions adopt special procedures and often reduce commanding. A settlement should treat this recurrent geometry as an operational season: preload software, synchronize critical knowledge, prepare medical references, schedule industrial work and increase communication reserves before the link becomes poor.

This becomes a natural maturity test. If the settlement cannot operate for days with strongly reduced Earth support, it is not operationally autonomous. Conjunction can therefore be used as a planned resilience exercise that verifies local authority, automation, logistics and knowledge stores before a real emergency removes Earth unexpectedly.

Cybersecurity and authority: a network packet may command a pump, airlock or rover

On Earth, a cyber incident can stop a factory. On Mars, the boundary between cyber and physical safety is thinner. The same network carries commands to power, air, robots, airlocks, workshops and vehicles. A compromised identity, badly signed update or false timebase can therefore create immediate material consequences.

Security cannot rest on the belief that “nobody will attack Mars.” It begins with architecture: segmentation between life-critical and comfort networks, strong authentication, command signatures, auditable logs, local key management, least privilege, manual fallbacks and the ability to reject a syntactically valid command when it contradicts the locally observed physical state.

Earth–Mars delay also forces a doctrine of authority. Who may shut down a reactor? Who may close a shared airlock? Who can immobilize a rover owned by another organization? A terrestrial command received twenty minutes later may have become dangerous because the situation changed. Commands therefore need context, validity windows and local execution conditions.

This ties networking to governance. An information-autonomous settlement is not isolated from Earth; it is capable of deciding when an external message is advice, when it is strategic authority and when it can no longer be executed without local judgment.

On Mars, cybersecurity and industrial safety converge. A compromised credential may become an open valve; a corrupted update may disable life support; a false clock may invalidate a navigation sequence. Vital, scientific and personal networks should be segmented, privileges limited, software signed, known-good configurations retained and changes logged. Security recovery also has to work without Earth: revoking a key, isolating a segment or restoring a version cannot depend on an interplanetary round trip.

Autonomy does not mean software owns every decision. A pump may trip automatically on overpressure and a rover may avoid an obstacle, but a reconfiguration that permanently abandons part of a habitat or consumes strategic reserves belongs to a higher authority. Designing the network therefore also means designing boundaries of authority, escalation and human responsibility.

On Mars, a cyber incident can become a physical incident

A settlement network will connect systems that move matter and energy: pumps, valves, chargers, vehicles, doors, medical devices and crop systems. Cybersecurity therefore cannot be reduced to data theft. A malicious command or compromised update could create an air failure, collision or sterilization defect.

The first defence is architecture: segmentation between administrative networks and control systems, strong authentication, signed software, logging, least privilege, manual modes and offline backups. The second is recovery. A settlement must be able to rebuild a controller, restore a known configuration and continue operating even when part of the network is treated as hostile.

Earth delay makes local resilience more important. Terrestrial cybersecurity teams can help analyse an incident, but they cannot decide in real time which valves to isolate. Local operators need authority and tools, turning cybersecurity into a civil-safety competence.

From four people to a city: networking changes nature with scale

For four people, a local network may fit inside a handful of switches, radios and redundant computers. At twenty people, operations, health, laboratory work, maintenance and private use already compete. At one hundred, multiple teams, workshops, habitats and vehicles operate simultaneously. At one thousand, the problem resembles a critical utility: network core, local data centers, backups, identity services, public services and on-call teams.

Growth is not linear. Doubling population does not merely double traffic. The number of interactions grows, services diversify and a failure affects more people. Networking evolves from mission equipment into a civic institution with maintenance obligations, availability targets and separation between essential and optional services.

The 2026 National Academies science strategy reinforces the point: human campaigns on Mars aim to produce complex observations about life, geology, climate, water and the human environment. Science itself becomes a major consumer of storage, computation and telecommunications. A weak information infrastructure does not merely reduce comfort; it limits what human presence can learn from Mars.

The Martian network is therefore political infrastructure in the broadest sense: it determines who can communicate, which information survives disruption, how decisions are documented and how far a community can act without waiting for Earth. It may be one of the least visually dramatic systems and one of the most important in turning a base into a society.

With four people, much coordination can still happen by voice and a handful of consoles. At twenty, teams, workshops and shifts appear. At one hundred, identity management, traffic priorities, data retention and network maintenance become permanent jobs. At one thousand inhabitants, communications are no longer a mission subsystem; they are urban infrastructure with operators, physical redundancy, security governance, spares and defined levels of service.

Scaling also creates data-sovereignty questions. Medical records, manufacturing recipes, infrastructure drawings and scientific archives need replication, backups and ownership rules. 'Where is the truth?' becomes operational when two bases have spent hours editing different versions of a procedure. Version control, synchronization and conflict resolution become as important to continuity as antennas and transmitters.

2026: When Mars relay begins to look like infrastructure

For decades, Mars communications were assembled mission by mission. A rover carried a proximity radio, a science orbiter doubled as a relay, and the Deep Space Network returned the data to Earth. That opportunistic architecture has been remarkably successful, but it is not the scale of a human settlement. A crewed base produces life-critical telemetry, medical records, industrial models, navigation data, video, science, private communications and safety commands. It cannot treat communications as a useful side effect of aging science spacecraft.

On 14 May 2026 NASA issued a Request for Proposals for a Mars Telecommunications Network. The official notice calls for high-performance Mars telecommunications orbiters supporting surface, orbital and future human exploration, with the network expected to be ready to operate at Mars no later than 2030. That date is not a promise that a Martian Internet will be complete in 2030. The more important change is institutional: NASA is treating Mars communications as infrastructure worthy of dedicated vehicles and continuing service.

The existing Mars Relay Network explains the transition. Odyssey, Mars Reconnaissance Orbiter, Mars Express and Trace Gas Orbiter have served surface missions as relay assets. A short UHF hop from a rover to an orbiter can be far more efficient than a direct Earth link because the range is thousands rather than tens or hundreds of millions of kilometres. The orbiter stores data and forwards it when geometry and schedules allow. NASA's MaROS service makes this operational chain visible: data has a route, a pass, a relay, a delivery time and sometimes a retry.

For a city, that becomes a governance question. When does telecommunications stop being a mission subsystem and become a public utility? If hospitals, power plants, rovers and factories share the network, service classes, priorities, ownership and repair responsibilities must be explicit. A failed orbiter is no longer merely inconvenient for science; it may reduce medical capability or isolate a vehicle. The birth of the Mars Telecommunications Network is therefore a useful historical pivot from a network of missions toward infrastructure for society.

Mars Telecommunications Network: when a mission network becomes infrastructure

In May 2026 NASA released an industry request for proposals for a Mars Telecommunications Network intended to provide high-performance services to orbital, surface and future human missions, with capability requested at Mars no later than 2030. The historical importance lies less in the date than in the change of philosophy: instead of requiring every mission to carry nearly all of its own communications chain, the architecture begins to treat communications as a shared service.

Shared infrastructure changes mission economics. If a lander can rely on a powerful relay, its own long-range antenna can be smaller or its surface data rate much higher. If several vehicles use the same service, fixed orbital infrastructure costs can be distributed. But the shared dependency creates a common-mode risk: failure of the network affects many missions at once. The architecture therefore has to combine mutualization with resilience.

Eventually the issue becomes political. Who funds the network — NASA, ESA, private operators, the Martian settlement? Can a commercial operator prioritize its own vehicles? What minimum service remains available for emergencies? Who allocates spectrum and network identities? At what point does exploration infrastructure become a de facto public utility? Mars colonization begins to meet questions terrestrial societies already know from electricity, water and internet access.

DTN: design for the absence of connectivity

Terrestrial networking usually assumes that two endpoints can establish an interactive path with modest delay. Earth and Mars violate that assumption. Light time alone creates minutes of one-way latency, orbital motion creates visibility windows, an orbiter disappears behind the planet, and solar conjunction can make communications unreliable. Treating the system as ordinary Internet with a very long ping would build fragility into the architecture.

Delay/Disruption Tolerant Networking starts from the opposite assumption: disruption is normal. The Bundle Protocol can package information into units that are stored persistently at intermediate nodes and forwarded when a later contact becomes available. End-to-end connectivity need not exist when the data is generated. Routing can therefore depend on contact plans, priority, storage, expiration and custody rather than on a continuously open path.

That changes the meaning of network time. A leak alarm is nearly worthless if it arrives hours too late; a geological archive can wait; a signed medical record must remain authentic across multiple stores; a software update must never be mistaken for a complete file if only part arrived. The network must manage priority, integrity, lifetime and security, not merely move bytes.

A simple scenario shows the storage problem. A site generating 50 Mbit/s for eight hours without its primary link produces 50 × 28,800 = 1,440,000 megabits. Divide by eight to convert bits to bytes and the result is roughly 180 gigabytes before protocol overhead and replicas. Disruption therefore becomes a storage and policy problem as much as an antenna problem.

Inside a habitat, conventional low-latency networks will still make sense. Across planets, asynchronous networking is unavoidable. A Martian Internet will therefore be a federation of networks with radically different timing assumptions, linked by gateways that understand delay as a normal physical property.

Bundle Protocol, LTP and custody: design for disruption instead of suffering it

Much of terrestrial networking assumes that an end-to-end path exists while two systems communicate. Between Earth and Mars that assumption fails. Contacts are intermittent, delays are enormous and orbital geometry is partly predictable. Delay/Disruption Tolerant Networking turns the constraint into a model: a node can durably retain a unit of data — a bundle — until a useful contact becomes available.

LTP can provide reliable transfer across selected long-delay links, while custody mechanisms allow an intermediate node to accept responsibility for retaining data until it can forward them. Memory therefore stops being a temporary buffer and becomes a networking function.

A contact plan can exploit predictable geometry: a relay may be visible in forty-two minutes for an eight-minute pass, another station may become available later, and the router can select a path through time as well as space. A bundle bound for Earth begins to resemble freight moving through scheduled connections rather than a continuous phone call.

Prioritizing when everything cannot leave at once

Suppose a base accumulates 180 GB during an eight-hour interruption. A 200 MB emergency medical package, 2 GB of industrial fault logs and 150 GB of science imagery do not have the same urgency. The network needs service classes, expiration times and priority rules. But there is an obvious failure mode: if every department marks its traffic “critical,” priority stops meaning anything. Network governance therefore needs categories that survive operational and political pressure.

Security interacts with latency. A bundle may remain stored across several nodes for a long time. Integrity, authentication and access control must survive that journey. A command that arrives after the operational situation has changed can be dangerous even if it is authentic. Expiration and temporal context are therefore safety functions, not housekeeping.

Radio and optical links: bandwidth becomes a civic budget

Optical communications are compelling because a narrow beam can deliver very high data rates with limited transmitted power. JPL's DSOC demonstration achieved 267 Mbit/s at about 31 million kilometres and continued communicating at far greater distances. That does not mean a Mars colony will receive 267 Mbit/s at all times: range, pointing, weather at Earth terminals and station availability matter. It does demonstrate that high-rate deep-space optical communications have moved beyond a paper concept.

Human Mars networks will probably be hybrid rather than purely optical. Local links can use wired networks, UHF and other radio systems; relay orbiters can support several bands; Mars-to-Earth trunks may combine X/Ka radio and optical channels. The key word is diversity: different paths, different frequencies, different Earth stations and graceful degradation.

The architecture must then answer civic questions: how much energy is reserved for information, which data stays local, which services survive total Earth isolation, and which communication capability must survive the loss of a flagship relay? What a settlement chooses to store and protect reveals what it considers vital.

DSOC: turn a record into a human-scale data-transfer example

The Deep Space Optical Communications — DSOC demonstration achieved 267 Mbit/s from roughly 31 million kilometres. Converting the number gives intuition. A 100 GB file contains about 800 gigabits. In an idealized case where 267 Mbit/s remained constant and protocol overhead were ignored:

t = 800 Gbit ÷ 0.267 Gbit/s ≈ 2,996 s ≈ 49.9 min.

The calculation is deliberately ideal. Data rate falls with geometry and distance; terrestrial weather can close an optical site; pointing has to be extremely accurate; coding, protocols and contact schedules reduce useful throughput. But the example makes the scale difference between a few megabits per second and hundreds of megabits per second tangible.

For a settlement, optical links could carry bulk science, imagery, database replication, digital twins and cultural media. Radio will probably remain essential for robustness, commanding, poor-weather availability on Earth and many local links. The most resilient architecture is therefore likely hybrid rather than “laser only.”

Four, twenty, one hundred, one thousand people: scaling the network

For four people, simplicity is a safety feature. One habitat, a few rovers, a small number of relay paths, local caches and well-understood procedures may be enough. Critical backups can be physically separated and knowledge duplicated on independent media.

At twenty people the base becomes an organization. Multiple EVAs, laboratories, workshops and medical functions compete for capacity. Traffic must be segmented and logged; local services must remain usable when Earth is absent; storage becomes an operational library rather than a mission recorder.

At one hundred people, vehicles and industrial processes multiply the data faster than population alone would suggest. Identity management, off-site backup, network operations and cybersecurity become permanent jobs. Partial failure must not interrupt air, water or power.

At one thousand people the system resembles a critical infrastructure operator: several data centres, physically diverse paths, spectrum governance, orbital relays, local positioning and professional maintenance teams. Losing Earth for days should be a planned operating mode, not an apocalypse scenario.

A quantified scaling scenario — explicit assumptions, not a forecast

To make scale concrete, consider a pedagogical scenario in which the amount of data worth retaining averages 20 GB/day for four people, 80 GB/day for twenty, 500 GB/day for one hundred and 5 TB/day for one thousand. These are not NASA requirements or predictions. They simply illustrate how science, video, predictive maintenance, health records, digital twins and backups can grow with population.

Raw storage is not the hard part: seventy terabytes is modest by modern terrestrial standards. The challenge is maintaining hardware, verifying backups, powering and cooling servers, replicating data and deciding what must be synchronized to Earth. A settlement can therefore become highly autonomous locally while remaining scientifically and culturally connected to its origin planet.