The New Frontier: Payload Economics — How Starship Changes Everything

Part 3 of a 6-part series. Part 1: The Vision | Part 2: The Timeline

V3 Flies

May 22, 2026. Starship V3 — stretched tanks, 33 Raptor 3 engines, catch-ready hardware — lifted off from Starbase, Boca Chica, on its first integrated flight test. IFT-12. The upgraded heat shield held through re-entry. Around twenty mock Starlink V3 satellites deployed from the payload bay. Super Heavy executed a controlled descent and splashed down in the Gulf of Mexico (catch attempt deferred for the first V3 flight — smart). The Ship completed a controlled splashdown in the Indian Ocean.

Musk posted on X: “Epic first Starship V3 launch & landing! You scored a goal for humanity.”

During the pre-launch rehearsal, the vehicle loaded over 5,000 metric tonnes of propellant — liquid methane and liquid oxygen — into the stretched tanks. That’s roughly the mass of 100 fully loaded semi-trucks, pumped aboard in under an hour. The entire stack stood about 123 metres tall, making it the largest and most powerful rocket ever to fly successfully.

V3 hardware flew, deployed payloads, survived re-entry, and splashed down intact. Real hardware. Real flight. The production pipeline behind it — ~10 more ships and ~5 boosters planned for the rest of 2026 — tells you SpaceX isn’t treating this as a one-off stunt. They’re building a fleet.

And everything I’m about to write depends on that fleet.

Because the story of space isn’t really about rockets. It’s about what a kilogram costs to get up there — and what happens when that number drops by three orders of magnitude.

Sixty Years of Getting Cheaper (Slowly)

To understand why Starship matters, you need to see the cost curve it’s bending. Every rocket ever built has charged a toll per kilogram to low Earth orbit (LEO, ~200-400 km altitude). That toll has defined what’s possible in space. Here’s what the last six decades look like:

  • Saturn V (1967-1973): as much as $46,000/kg to LEO in some analyses that include full programme costs — in 1960s dollars. Adjusted for inflation, estimates range even higher. Expendable. Every launch threw away $2-3 billion worth of hardware into the ocean. But it put humans on the Moon, so nobody cared about the bill — until Congress did.

  • Space Shuttle (1981-2011): ~$18,000-54,000/kg depending on how you amortise the programme costs. The wide range tells you something: the Shuttle was “reusable” in the sense that you could refurbish it for the price of building a new one. Each turnaround took months and cost hundreds of millions. It could carry ~27,500 kg to LEO but averaged only ~5 flights per year across the fleet.

  • Ariane 5 (1996-2023): ~$8,000-12,000/kg. Europe’s workhorse. Reliable. Expendable. Good for GTO dual-manifest missions, but fundamentally a use-once vehicle.

  • Atlas V / Delta IV Heavy (2002-2024): ~$4,000-10,000/kg. ULA’s national security workhorses. Reliable to the point of boring (a compliment in rocketry). But expensive, and never designed for cost reduction — designed for assured access.

  • SLS Block 1 (2022-present): ~$10,000-20,000+/kg. NASA’s Senate Launch System. It works — Artemis I proved that. But at $2.2 billion per launch for a rocket that puts ~95,000 kg into LEO and flies once or twice a year, the economics are brutal. Every SLS launch costs roughly the same as SpaceX’s entire Starship development through 2023.

  • Falcon 9 (reusable, 2015-present): $2,700/kg at list price ($67M for ~22,800 kg). On high-utilisation internal missions (Starlink), the effective cost drops to $1,500-2,000/kg. SpaceX has landed Falcon 9 boosters over 350 times. This is the vehicle that proved orbital-class reusability works at industrial scale. It broke the cost curve — but it’s approaching its floor.

  • Falcon Heavy (reusable, 2018-present): ~$1,000-1,500/kg. Three Falcon 9 cores strapped together. Up to ~63,800 kg to LEO. The cheapest operational heavy-lift option before Starship.

  • Starship V3 (target, fully reusable, high cadence): $10-100/kg to LEO.

Read that last line again. Then read the first one. That’s a drop from $46,000/kg to $10/kg — roughly 4,600x cheaper. Even the conservative near-term estimate of $100-200/kg represents a 20-30x improvement over today’s best option.

The magnitude of change goes beyond incremental improvement. Think of the difference between artisanal hand-delivery and containerised shipping. And that analogy isn’t casual — it’s the entire point of this article.

The First Real Price Signal

Talk is cheap. Aspirational cost targets on conference slides are cheaper. What matters is what someone actually agreed to pay.

In early 2026, Voyager Space disclosed in its 10-K SEC filing that it had contracted with SpaceX for a dedicated Starship launch to deliver the Starlab commercial space station to orbit, targeted for around 2029. The price: $90 million.

Let’s do the arithmetic. Starship V3’s reusable payload capacity to LEO is approximately 100,000–150,000 kg (100–150 tonnes) — SpaceX states >100 t, with V3 targeting the higher end. At $90M and the conservative 150t end:

$90,000,000 ÷ 150,000 kg = $600/kg

That’s 4-7x cheaper than Falcon 9 at list price. For a dedicated mission. On what is essentially an early-production vehicle with minimal flight heritage.

This is the launch of the cost curve, not the destination. SpaceX’s internal economics will drive this figure down relentlessly as cadence increases, production scales, and rapid reuse becomes routine. The $90M number is what you charge when you’ve flown a handful of V3 missions and every launch still involves significant one-off engineering review. Imagine what ship #50 costs to operate versus ship #3.

For context: SpaceX has invested over $15 billion in Starship development to date. That’s a staggering number for a private company, but it’s roughly 7 SLS launches. The difference is that SpaceX gets a reusable fleet out of it. NASA gets 7 splashdowns.

How Starship Economics Actually Work

The magic of Starship’s cost structure isn’t any single breakthrough. It’s three things compounding simultaneously: cheap propellant, full reusability, and high cadence. Remove any one of them and the economics don’t work. Together, they’re transformative.

Propellant: The Cheapest Part of the Rocket

Starship runs on methalox — liquid methane (CH₄) and liquid oxygen (LOX). Both are cheap industrial commodities. Methane is natural gas, cooled down. LOX is air, separated and liquefied.

The propellant cost for a fully loaded Starship + Super Heavy stack is approximately $1-2.4 million per flight. That’s for roughly 5,000 tonnes of propellant.

For comparison, a single Falcon 9 launch at list price is $67 million. The propellant alone for Starship costs less than 4% of a Falcon 9 ticket. The rest of the cost is hardware amortisation, operations, range fees, and margin.

This is a critical insight: the fuel is essentially free at scale. Just like jet fuel is a small fraction of an airline ticket price. The cost driver is everything else — and “everything else” is what reusability and cadence attack.

Reusability: Amortising the Hardware

A Falcon 9 booster has demonstrated 20+ flights on a single unit. Starship aims to exceed that significantly — the design target is for both the booster (Super Heavy) and the ship (Starship) to fly hundreds of times with minimal refurbishment between flights.

Here’s why that matters. Say a Starship costs $100 million to build (a rough estimate for early production). If you fly it once and throw it away, that’s $100M in hardware cost per flight. Fly it 10 times, it’s $10M. Fly it 100 times, it’s $1M. Fly it 1,000 times (ambitious, but SpaceX talks about aircraft-like operations), it’s $100,000.

Add propellant ($1-2.4M), ground operations, range fees, and you get to a marginal cost of $2-10 million per flight at high cadence. At 100–150 tonnes to LEO, that’s (using the high end):

  • $2M flight: $13/kg
  • $5M flight: $33/kg
  • $10M flight: $67/kg

These numbers assume mature operations with rapid turnaround. We’re not there yet. But the IFT-12 success — catching a V2 booster on IFT-7, flying V3 hardware successfully — shows the trajectory is real.

Cadence: The Hidden Multiplier

This is the factor most people underestimate. The cost per flight isn’t just about the vehicle — it’s about the fixed infrastructure: launch pads, ground crew, mission control, manufacturing facilities, quality assurance teams. These costs exist whether you launch once a month or once a day.

SpaceX’s Starbase facility is being built for multiple launches per day. Not per week. Per day. The integration tower (“Mechazilla”) is designed for rapid catch-and-restack operations. The propellant farm can support back-to-back launches.

At one launch per week, your fixed costs divide by 52. At one launch per day, they divide by 365. At three launches per day (SpaceX’s medium-term target for Starbase + Cape Canaveral combined), they divide by ~1,000.

This is the airline model. A Boeing 777 costs $350 million. Airlines don’t make money by flying it once — they make money by flying it 14-16 hours per day, every day, for 25 years. SpaceX is building the same operational cadence for orbital rockets.

Atomic Age Space Fuel Station — orbital refuelling, retro-futurism style

The Refuelling Revolution

Here’s where the economics get really interesting — and where Starship stops being “just” a LEO truck and becomes an interplanetary transport system.

The Problem

Starship can put ~150 tonnes into LEO with full reusability. That’s extraordinary for Earth-to-orbit. But going beyond LEO — to the Moon, Mars, or anywhere in deep space — requires much more delta-v (velocity change), which means much more propellant. A fully loaded Starship in LEO doesn’t have enough fuel left to reach the lunar surface with a useful payload.

The Solution: Orbital Refuelling

Instead of building a bigger rocket (the traditional approach — and the reason Saturn V was so enormous), SpaceX’s plan is to refuel Starship in orbit. Launch a Starship to LEO mostly empty of payload but full of propellant. Transfer that propellant to a waiting “depot” Starship. Repeat until the depot is full. Then the mission Starship tops off its tanks from the depot and departs for the Moon or Mars with a full propellant load and a full payload.

The first orbital refuelling demonstration is targeting June 2026 — just weeks from now. The test will involve two Starships: a “target” depot vehicle and a “chaser” tanker. This is one of the most consequential spaceflight milestones since the first orbital docking in 1966 (Gemini 8).

The engineering challenges are real. Cryogenic propellant (liquid methane at -161 °C, liquid oxygen at -183 °C) doesn’t behave well in microgravity. It floats around the tank in unpredictable blobs instead of settling at the bottom where you can pump it out. The solution involves settling thrust — small engine burns that push the propellant to the tank bottom — and sophisticated thermal management to minimise boil-off.

What Refuelling Enables

A lunar mission using current V3 hardware requires approximately 8-16 tanker flights to fully fuel a Starship HLS (Human Landing System) in LEO. That sounds like a lot, but consider: at high cadence, SpaceX could launch a tanker every few days. The total propellant delivery campaign might take 4-8 weeks. The Starship V4 (targeted for the late 2020s, with 200-300 tonnes to LEO) reduces this to 5-6 tanker flights.

The multiplication factor is dramatic. Without refuelling, Starship delivers ~150 tonnes to LEO and nothing beyond. With refuelling, it delivers 50-100+ tonnes to the lunar surface and potentially 100+ tonnes to Mars transit. Refuelling effectively multiplies useful payload by 3-5x for destinations beyond LEO.

An uncrewed HLS lunar landing demonstration is targeted for June 2027. If orbital refuelling works in 2026 and the lunar demo works in 2027, crewed lunar landings via Starship become a near-term reality rather than a distant aspiration.

The Cislunar Train

Think of it as a railroad — not a single heroic journey, but a system of regular routes and infrastructure:

  • Earth-to-LEO: Starship launches from Starbase or Cape Canaveral. Cheap, high-cadence, routine. This is the “first mile” — analogous to a truck bringing goods to a rail terminal.

  • LEO propellant depot: A Starship variant parked in orbit, accumulating fuel from sequential tanker flights. This is the marshalling yard — where you assemble the resources for the next leg.

  • LEO-to-lunar orbit: A refuelled Starship HLS departs with full tanks and a full payload bay. This is the mainline freight run.

  • Lunar surface operations: Starship HLS lands, delivers cargo, and eventually returns to lunar orbit for refuelling and reuse.

  • Eventually: Lunar ISRU (In-Situ Resource Utilisation). Manufacture propellant from lunar water ice (H₂O → H₂ + O₂) at the south pole. This replaces Earth-launched propellant for the return trip, cutting the number of required tanker flights dramatically. This is Part 4’s topic — and it changes everything again.

The ultimate architecture includes propellant depots at Earth-Moon L1 or L2 Lagrange points or in lunar orbit — “gas stations” positioned along the route. Pre-position fuel, and missions become smaller, more frequent, and cheaper. You stop needing heroic one-shot expeditions. You build a supply chain.

This is how the American West was settled. Not by individual wagon trains making desperate crossings, but by the railroad, the telegraph, and the supply depot turning a frontier into a territory and then a state. The cislunar train is the same concept, scaled up by about 384,400 km.

Wright’s Law: Why Costs Keep Falling

There’s a pattern in manufacturing that has held for over a century, across industries from Ford’s Model T to Boeing’s aircraft to solar panels to lithium-ion batteries. It’s called Wright’s Law (after Theodore Wright, who described it for aircraft production in 1936):

Every cumulative doubling of units produced yields a consistent percentage decrease in cost per unit.

The typical learning rate is 15-25% — meaning each doubling reduces unit cost by 15-25%. This isn’t magic. It’s the compounding effect of workers gaining experience, processes being refined, tooling being optimised, suppliers scaling up, and design engineers eliminating unnecessary complexity.

SpaceX has already demonstrated Wright’s Law in action with Starlink satellites. The first Starlink sats (v0.9, 2019) cost an estimated $1 million+ each. By the time SpaceX had manufactured over 7,000 units (late 2024), costs had dropped to an estimated $250,000-300,000 per satellite. With over 10,000 cumulative units by 2026, costs are likely below $200,000 each. That’s a 5x+ reduction driven purely by production learning.

The same dynamic applies to Starship itself. Ship #1 (the first Starship prototype that flew) was essentially hand-built, with enormous non-recurring engineering costs baked in. Ship #30 benefits from refined manufacturing processes, established supply chains, trained workforce, and optimised designs. Ship #100 will cost substantially less than Ship #30. The learning curve is relentless.

And it doesn’t stop at the vehicle. Everything that rides on Starship benefits from the same law:

  • Orbital compute satellites: The first orbital data processing unit costs a fortune. The hundredth costs a fraction. The ten-thousandth is commodity hardware.
  • Lunar habitat modules: Mass-produced pressurised modules, leveraging the same manufacturing lines over and over.
  • Solar arrays, radiators, docking adapters — every component follows the curve.

Wright’s Law is why SpaceX’s strategy of building the machine that builds the machine is so powerful. The factory is the product. Starbase functions as a rocket production line that happens to have a launch pad at the end of it.

The Comparison That Matters

Here’s the number that reframes everything. Forget rockets for a moment. Think about how much it costs to move a kilogram of stuff on Earth:

  • Container ship (Shanghai to Rotterdam, ~19,000 km): ~$0.01-0.02/kg
  • Rail freight (cross-continental): ~$0.03-0.05/kg
  • Truck freight: ~$0.10-0.30/kg
  • Air freight (intercontinental): ~$2-5/kg
  • Express air courier (FedEx/DHL priority): ~$20-80/kg

Now, space:

  • Falcon 9 to LEO: ~$2,700/kg
  • Starship near-term (early operations): ~$100-200/kg
  • Starship long-term (high cadence, mature ops): ~$10/kg

At $2,700/kg, only things that are extraordinarily valuable per kilogram make economic sense in orbit: telecommunications satellites ($5,000-50,000/kg in value), scientific instruments, military assets. The market is limited to things that must be in space and that can justify outrageous shipping costs.

At $100/kg, a new category opens up. That’s roughly the cost of expedited air freight for bulky cargo. Suddenly, space stations with thick shielding become affordable. Orbital manufacturing pilot plants with heavy equipment become viable. You’re no longer constrained to featherweight, hyper-optimised hardware.

At $10/kg, you cross a threshold that changes the nature of what space is for. That’s cheaper than many types of terrestrial shipping when you factor in insurance, customs, and handling for high-value goods. At $10/kg, you can send steel, concrete, water, bulk supplies to orbit. You can build large structures. You can resupply a lunar base with food and equipment routinely. You can send 100 tonnes to Mars for the price of a nice house.

Here’s the historical parallel that I keep coming back to: the gap between air freight and container shipping is what created global trade. Before the standardised shipping container (1956), international trade was expensive, slow, and limited to high-value goods. The container reduced shipping costs by ~95% and enabled the globalisation of manufacturing. Suddenly, it made economic sense to manufacture goods in Shenzhen and sell them in Stuttgart.

The gap between Falcon 9 and Starship is what creates the space economy. Not “space exploration” — the space economy. An economy where things are manufactured, traded, and consumed in orbit, on the Moon, and eventually on Mars.

Space Vegas — when launch costs drop, services follow. Every frontier gets its entertainment district.

What Cheap Launch Enables: The Cascade

Each order-of-magnitude reduction in launch cost doesn’t just make existing activities cheaper — it enables entirely new categories of activity that were previously impossible. This is a cascade, and it’s the most important concept in space economics.

At ~$2,700/kg (Falcon 9 era — now)

  • Communications and Earth observation satellites
  • GPS and navigation constellations
  • Scientific missions (space telescopes, planetary probes)
  • ISS resupply and crew rotation
  • Starlink broadband constellation
  • Space tourism for the ultra-wealthy ($55M per Axiom mission)

This is the current market. It’s worth roughly $400-500 billion per year globally and growing. But it’s fundamentally limited to things that must be in space and that can justify premium shipping costs.

At ~$600/kg (early Starship — Voyager contract pricing)

  • Commercial space stations (Starlab, Axiom Station, Orbital Reef) become economically viable to build and resupply
  • Orbital manufacturing pilot plants — pharmaceutical crystallisation, ZBLAN fibre optics, high-purity semiconductors
  • Mega-constellations beyond Starlink — thousands of satellites for 6G, IoT, Earth monitoring
  • Space tourism drops from $55M to maybe $5-10M per seat — still exclusive, but 10x more accessible

At ~$100/kg (Starship mature operations)

  • Orbital data centres become economically competitive with terrestrial facilities in specific use cases (free solar power at 1,361 W/m² unattenuated, vacuum cooling, no real estate costs, no permitting battles)
  • Large orbital structures — rotating habitats, manufacturing platforms, fuel depots
  • Routine lunar cargo delivery — tonnes of supplies per mission instead of hundreds of kilograms
  • In-space manufacturing at industrial scale, not just lab experiments
  • The in-space manufacturing market is estimated at $1.5 billion in 2026, growing at 23.7% CAGR. At $100/kg launch costs, that growth rate could accelerate dramatically.

At ~$10/kg (Starship long-term, high cadence)

  • Bulk construction materials to orbit — water shielding for radiation protection, metal for structures, regolith simulant for experiments
  • Lunar base resupply becomes routine logistics, not a heroic mission
  • Mars missions become affordable for national space agencies and eventually private ventures — 100 tonnes to Mars transit for $1 million in launch costs
  • Orbital cities — rotating habitats with thousands of residents become architecturally and economically conceivable
  • Asteroid mining return missions — send heavy processing equipment to near-Earth asteroids
  • Space-based solar power — the mass penalty that killed every previous SBSP proposal becomes manageable

Each 10x cost reduction doesn’t just expand the existing market — it creates markets that didn’t exist before, couldn’t exist before, because the economics were impossible. This is the cascade.

Orbital Manufacturing: The Near-Term Killer App

Among the cascade effects, one stands out as the most likely near-term economic driver: orbital manufacturing, particularly orbital data centres.

Why Space for Compute?

The logic is disarmingly simple:

  • Free energy: Unattenuated solar flux in LEO is ~1,361 W/m², available ~60% of the time (eclipse periods). No atmosphere, no clouds, no seasonal variation. Solar panels in orbit generate roughly 3-4x more energy per square metre than the best terrestrial installations.

  • Free cooling: Vacuum is the ultimate heat sink for radiative cooling. No air conditioning. No water. No cooling towers. Data centres on Earth spend 30-40% of their energy budget on cooling. In space, that cost approaches zero (you still need radiators, but the thermal environment is far more favourable).

  • No land costs, no permitting: Try building a 500 MW data centre anywhere on Earth. You’ll fight planning boards, environmental reviews, NIMBYism, and water rights for years. In orbit, there’s no zoning board. No neighbours. No water table concerns.

  • No geopolitical jurisdiction issues: An orbital data centre doesn’t sit in any country. For certain classes of computation — AI training, financial modelling, sensitive data processing — jurisdictional neutrality has value.

What’s Already Happening

  • Axiom Space launched the AxDCU-1 (Axiom Data Center Unit-1) to the ISS in late 2025. It’s a proof-of-concept: a small compute module using neural processing units (NPUs) optimised for space — radiation-hardened, low power, designed for the thermal environment. Still a proof-of-concept, not production — but the first real hardware test of orbital compute.

  • Google is reportedly in discussions with SpaceX about orbital data centre architecture. No confirmed contracts yet, but the interest from a hyperscaler validates the concept.

  • TERAFAB — a concept for dedicating 80% of 1 TW/year of chip production capacity to space-based NPUs. That’s an audacious number. To put it in perspective, global semiconductor fab capacity in 2026 is roughly 3-4 TW/year equivalent. TERAFAB would represent a fundamental reorientation of chip manufacturing toward orbital deployment. This is aspirational, not imminent — but the fact that serious people are modelling it tells you where the trendline points.

The Economics

At current launch costs (~$2,700/kg on Falcon 9), orbital data centres make no economic sense. The launch cost alone to put a 10-tonne compute module in orbit would be $27 million — before you’ve built the module, the solar arrays, or the radiators.

At $100/kg (mature Starship), that same 10-tonne module costs $1 million to launch. Now the equation changes. If the module generates revenue for 5-10 years, the launch cost becomes a minor line item compared to the hardware and operations costs. The free energy and cooling start to dominate the total cost of ownership calculation.

At $10/kg, you’re launching 10 tonnes for $100,000. At that point, orbital compute doesn’t just compete with terrestrial — it wins for specific workloads, particularly energy-intensive AI training where power cost dominates.

The Competition: Who Else Is in the Game?

SpaceX isn’t operating in a vacuum (well, technically they are, but you know what I mean). The launch market is responding to Starship’s looming dominance, and several competitors are relevant:

  • Blue Origin (New Glenn / Blue Moon): New Glenn is now flying and offers partial reusability (booster landing). The Blue Moon Mark 2 lunar lander, selected by NASA for Artemis V, could potentially deliver cargo to the lunar surface before SpaceX’s HLS — Blue Origin has a simpler architecture that doesn’t require orbital refuelling for initial missions. Blue Origin won’t match Starship’s $/kg economics, but they don’t need to — they need to be the reliable second option. For lunar cargo specifically, Blue Moon’s direct-flight architecture (no orbital refuelling required) could be operational sooner.

  • Rocket Lab (Neutron): Not competing on payload mass — Neutron targets ~13,000 kg to LEO. But Rocket Lab is building the Electron/Neutron combination to serve the small-to-medium sat market with rapid cadence. They’re the FedEx to SpaceX’s container ship: smaller parcels, faster turnaround, dedicated orbits. Different market, not directly threatened by Starship.

  • ULA (Vulcan Centaur): The replacement for Atlas V and Delta IV Heavy. Competitive for national security missions where dual-source procurement is mandated. Not cost-competitive with Starship, but has the advantage of NSSL (National Security Space Launch) certification and relationships.

  • China (Long March 9 / CZ-9): China’s super-heavy lift vehicle is in development, targeting first flight around 2030-2033. Partially reusable designs are being tested. China has the industrial capacity and political will to build a Starship competitor, but they’re at least 4-5 years behind on the reusability curve. The Long March 10 (crew-rated, targeting 2027 for lunar missions) is more near-term but smaller.

  • ESA / ArianeGroup (Ariane 6): Just entering service, already arguably obsolete economically. Ariane 6 is expendable and targets ~$8,000-10,000/kg. Europe is studying next-generation reusable launchers (SALTO, Themis) but nothing is funded for production before the early 2030s. Europe risks being locked out of the cheap-launch economy if it doesn’t accelerate.

  • India (NGLV / SHLV): India’s Next Generation Launch Vehicle and Super Heavy Lift Vehicle are in early development. India has the engineering talent and cost advantages (ISRO’s budgets are famously efficient) but is starting from expendable architectures.

The honest assessment: nobody is within 5 years of matching Starship’s $/kg. Blue Origin is the closest in ambition, but New Glenn is roughly comparable to Falcon 9 in economics, not Starship. The rest of the world is competing for second place in a race where first place is about to lap the field.

That said, competition matters — not for cost, but for resilience. A space economy that depends entirely on one vehicle from one company is fragile. The 2003 Columbia disaster grounded the Shuttle fleet for 2.5 years. If Starship had a similar stand-down, the entire space economy would freeze. Blue Origin, Rocket Lab, and others provide essential redundancy.

What Could Go Wrong

I’ve been painting an optimistic picture. Here’s the cold water.

Reusability Might Not Scale as Hoped

Falcon 9 boosters have flown 20+ times, but with refurbishment between flights. Starship’s economics depend on hundreds of flights per vehicle with minimal turnaround. No rocket has ever demonstrated that. Aircraft do it, but aircraft don’t experience the thermal and structural extremes of Mach 25+ re-entry. The V3 heat shield held on IFT-12 — but will it hold on flight 50? Flight 200? One unexpected failure mode in the thermal protection system could force expensive refurbishment cycles that wreck the cost model.

Orbital Refuelling Is Unproven

Nobody has ever transferred thousands of tonnes of cryogenic propellant between vehicles in orbit. The June 2026 demo will be a first. Boil-off management, fluid dynamics in microgravity, autonomous rendezvous and docking at the rates required for lunar campaigns — these are hard engineering problems. If refuelling proves more difficult or lossy than expected, the tanker flight count goes up, and cislunar economics get worse.

Cadence Requires Demand

SpaceX can build launch capacity for 100+ flights per year. But who’s buying? Starlink is the anchor customer, but even Starlink’s constellation completion only requires a finite number of launches. The cascade effects described above — orbital manufacturing, lunar bases, Mars missions — require other companies and governments to invest at scale. If demand doesn’t materialise at the pace SpaceX builds supply, the fixed-cost amortisation doesn’t work, and $/kg stays higher than the theoretical floor.

Regulatory and Environmental Constraints

Each Starship launch involves enormous quantities of propellant, generates significant noise and vibration, and produces exhaust products. Launching multiple times per day from a single site will face environmental review, FAA licensing constraints, and potential community opposition. SpaceX is building ocean launch platforms partly to mitigate this, but those platforms add cost and complexity.

Single Point of Failure

SpaceX is a private company controlled by one person. Its priorities could shift. Musk’s attention could be diverted (it certainly splits across many ventures). A catastrophic failure — loss of crew on a Starship mission — could ground the fleet for years. The space economy can’t be built on a single company’s roadmap, no matter how impressive that company is.

The $10/kg Target May Be Aspirational

The marginal cost model assumes aircraft-like operations, thousands of flights per vehicle, and minimal per-flight refurbishment. It also assumes propellant costs don’t rise (methane is a fossil fuel — carbon pricing could increase costs), that range fees stay low (unlikely if launch cadence exceeds current range capacity), and that insurance costs don’t spike after an inevitable anomaly. $100/kg feels achievable within 5-7 years. $10/kg might take 15 years or might require a fundamental shift in launch operations — ocean launch, autonomous operations, minimal ground crew.

I believe the trend is real. I also believe the timeline is uncertain and the final floor price is higher than the most optimistic projections.

But here’s the thing worth remembering: even at $200/kg — 10× cheaper than Falcon 9 — the transformation is revolutionary. At that price, orbital data centres become viable, lunar resupply becomes routine, and the cascade effects described above still trigger. The difference between $200/kg and $10/kg is the difference between a space economy and a space civilisation. But $200/kg alone already changes the world.

When Does Space Pay for Itself?

This is the question that separates space enthusiasts from space economists. The enthusiasts say: “we should go because it’s inspiring.” The economists say: “show me the revenue model.”

Here’s the honest answer: space already pays for itself in some domains, and Starship expands those domains dramatically.

Already Profitable (Today)

  • Communications satellites: A multi-billion dollar industry for decades. GEO comsats generate revenue from day one.
  • Starlink: On track for $6-7+ billion in annual revenue by 2026. Profitable as a standalone business. This is SpaceX’s real business model — Starship exists partly to deploy Starlink V3 at scale.
  • Earth observation: Companies like Planet, Maxar, and Umbra sell satellite imagery to governments, agriculture, insurance, and defence.
  • Launch services: SpaceX’s launch business alone is profitable. Every Falcon 9 flight is a revenue event.

Near-Term Viable (2027-2032)

  • Commercial space stations: Axiom, Starlab (Voyager/Airbus), Orbital Reef (Blue Origin/Sierra Space). Revenue from research, manufacturing, tourism, media. Viable at $600/kg launch costs.
  • Orbital manufacturing (specialty): ZBLAN fibre optics, pharmaceutical crystallisation, exotic alloys. Small-volume, high-value products where microgravity enables quality improvements worth the shipping cost.
  • Lunar cargo delivery: NASA and ESA are committed to sustained lunar presence. Government contracts provide baseline revenue.

Medium-Term (2032-2040)

  • Orbital data centres: At $100/kg, the first commercial orbital compute facilities become viable for energy-intensive AI workloads.
  • Lunar ISRU: Mining water ice, producing propellant, reducing the cost of cislunar operations. Revenue model: selling propellant to other missions.
  • Space tourism at scale: If a Starship ticket drops to $100,000-500,000 (still expensive, but within reach of the global top 1%), tourism alone could be a multi-billion dollar market.

Long-Term (2040+)

  • Asteroid mining: Precious metals, platinum-group elements, water. The economics only work at very low launch costs and with proven ISRU technology.
  • Space-based solar power: Beaming energy to Earth. Requires massive structures in orbit — only viable at $10-50/kg.
  • Mars settlement: Not a revenue generator in any conventional sense, but a civilisational project that becomes possible at Starship costs.

The Diamandis vision — 500,000 to 1 million orbital satellites, launch cadence of potentially one per hour, the TERAFAB concept requiring 10 million tonnes per year to orbit (which translates to roughly 274 Starship launches per day) — that’s the far end of the curve. It sounds insane today. So did “10,000 Starlink satellites” in 2015.

The Frontier of Independence

There’s a concept I keep coming back to when thinking about space economics, and it’s this: at what price point does a human activity in space stop depending on Earth for its survival?

At $2,700/kg, everything in space is utterly dependent on Earth. Every bolt, every meal, every litre of water launched from the ground. Orbital assets are outposts — maintained at enormous expense, abandoned if funding dries up. That’s the ISS model.

At $100/kg, you can afford to send redundancy. Spare parts. Extra food. Backup systems. Thick shielding. The margin of safety goes up, and the dependency on just-in-time resupply from Earth goes down. Outposts become stations.

At $10/kg, combined with ISRU (manufacturing from local resources), you reach a frontier of independence — where a lunar or orbital settlement can survive a temporary interruption in Earth supply. Not indefinitely, but long enough to ride out a launch stand-down, a funding dispute, or a political crisis. Stations become settlements. Settlements eventually become communities.

That’s the real inflection point. Not when space becomes profitable — it already is. Not when launch costs hit some magic number. The inflection point is when the cost of keeping people alive in space drops below the economic value they create there. That’s when space stops being an expense and starts being an economy.

Starship doesn’t get us all the way there. But it gets us close enough to see the destination.

What’s Next: Part 4 — Lunar Industrialisation

Cheap launch gets mass to orbit. Orbital refuelling gets mass to the Moon. But what happens when you can make things on the Moon?

In Part 4, we’ll dig into lunar industrialisation — water ice mining at the south pole, propellant production, regolith processing, and the economics of building a permanent industrial presence on a world with no atmosphere, 1/6 gravity, and 14-day nights. We’ll look at NASA’s Artemis programme, Blue Origin’s Blue Moon, SpaceX’s HLS, and the nascent lunar economy that’s starting to take shape.

The Moon isn’t a destination. It’s a depot.

This is Part 3 of “The New Frontier” — a 6-part series exploring humanity’s expansion into space, from the engineering to the economics to the ethics.

  • Part 1: The Vision — Why space matters now
  • Part 2: The Timeline — From IFT-1 to IFT-12 and beyond
  • Part 3: Payload Economics — How Starship Changes Everything (you are here)
  • Part 4: Lunar Industrialisation — coming soon
  • Part 5: Mars and Beyond — coming soon
  • Part 6: The Ethical Frontier — coming soon

Krzysztof Sajna is an IT engineering manager who spends too much time reading SEC filings and rocket engine specifications. He writes at sajna.space and can be found on LinkedIn.