Link to the code: Zae-Project / arkspace-core

Reusable Rockets and the Satellite Economy: How Launch Costs Are Reshaping Orbital Infrastructure in 2026

Every proposal for orbital computing infrastructure, whether SpaceX’s 1 million satellite data center or the Google Project Suncatcher TPU constellation, rests on an assumption that launching mass to orbit will become progressively cheaper. Without that assumption, none of these proposals are economically coherent. In 2026, the reusable launch vehicle industry is the mechanism making that assumption defensible.

The Cost Curve That Changed Everything

The baseline for launch cost before commercial reusability is approximately $9,000 per kilogram to low Earth orbit. That figure reflects expendable rockets from the 2000s: United Launch Alliance Delta IV, Ariane 5, Soyuz. Every mission consumed a new vehicle.

Falcon 9 with first-stage recovery cut that to roughly $2,700/kg at the advertised commercial rate. More significant is what reusability costs SpaceX internally: refurbishing a recovered Falcon 9 first stage costs under $300,000. Building a new one costs approximately $30 million. A booster completing its 26th flight, as Falcon 9 B1078 did on the March 1, 2026 Starlink 10-41 mission, has amortized its manufacturing cost across 26 missions. The marginal cost of each additional flight approaches the cost of propellant and refurbishment, not a new vehicle.

The practical consequence: SpaceX flew 22 Starlink missions in the first 10 weeks of 2026, adding 566 satellites to the constellation. That launch pace is physically possible only because boosters are not discarded after each use.

Market Scale in 2026

The reusable launch vehicle market, covering commercial launch services that recover and reuse first-stage hardware, is valued at approximately $1.38 billion in 2025 by Allied Market Research, with projections to $3.56 billion by 2035. The commercial satellite launch segment accounts for 51.4 percent of the market in 2025, confirming that constellation deployment is the primary driver.

A separate January 2026 market report from Yahoo Finance pegged the market at $2.49 billion in 2025 growing to $2.75 billion in 2026, reflecting different scope assumptions (whether government and military contracts are included). Both figures point in the same direction: the market is growing, satellite deployment is the primary demand source, and the growth rate is tied directly to constellation expansion plans from Starlink, Kuiper, and new entrants like Logos Space.

The 2026 LEO mega-constellation race covered in detail elsewhere involves three operators deploying or approving large constellations within a single year. Each of those constellations depends on reusable launch economics to be viable at scale.

Rideshare: Enabling the Small Satellite Economy

Reusability benefits large constellation operators most directly, but rideshare programs extend the cost reduction to smaller operators.

SpaceX’s Transporter and Bandwagon rideshare missions allow small satellite operators to purchase a portion of Falcon 9 capacity, reducing the effective per-kilogram cost for operators who cannot fill an entire fairing. Dedicated Falcon 9 customer missions have averaged only 3,370 kg of payload, roughly 19 percent of total capacity. Rideshare converts that unused capacity into revenue while driving down the minimum mission cost for smaller operators.

The practical effect is visible in the smallsat ecosystem. D-Orbit’s AIX constellation and Starcloud’s GPU cluster demonstrations both reached orbit on commercially viable economics that would not exist at 2010 launch prices. A 100 kg satellite that costs $270,000 to launch versus $900,000 is a categorically different business proposition.

The Starship Trajectory

Starship represents the next step in the cost reduction curve. Current estimates for Starship at partial reusability put launch costs at $78–$94 per kilogram to LEO, with a longer-term target below $100/kg as both stages achieve rapid reuse.

At $100/kg, the orbital infrastructure economics change fundamentally. A 500 kg satellite with a $50,000 launch cost can carry significantly more payload mass for the same mission cost as a 100 kg satellite at $270,000 launch cost. More importantly, the replacement cycle changes: operators can afford to refresh constellations more frequently, maintaining higher average capability without multi-year deployment gaps.

SpaceX’s 1 million satellite orbital data center filing implicitly relies on Starship-class economics. Launching 1 million satellites on Falcon 9 at current rates would cost hundreds of billions of dollars in launch costs alone. At Starship’s projected rates, the economics compress into a range that might be financially justifiable given projected compute revenue.

Commercial Space Stations and the Maintenance Dependency

Reusable launch economics also underpin the commercial space station development underway in 2026. Axiom Space, Voyager/Starlab, and the other ISS successor programs all require regular crew rotation and cargo resupply. The cost of those missions, which would be prohibitive at expendable rocket rates, is manageable with reusable vehicles.

The same logic applies to on-orbit servicing missions. The Astroscale ELSA-M debris removal mission and the Tetra-5 refueling demonstration both operate in an economic context where launch costs are low enough to make servicing commercially viable rather than a government-funded one-off.

Debris and the Sustainability Tension

Lower launch costs create a paradox for orbital sustainability. Cheaper access to orbit means more satellites, which means more debris risk. The cascade dynamics analyzed in detail elsewhere show that the Kessler threshold, the point at which debris collisions generate more debris faster than natural deorbit removes it, is a real operational constraint for dense constellations.

The reusable launch industry’s growth therefore has a direct dependency relationship with active debris removal capability. Astroscale, ClearSpace, and the Northrop Grumman MEV program are not independent of the launch market: they are the cleanup infrastructure that makes continued launch growth sustainable.

The Orbital Computing Economic Case

The core economic question for proposals like China’s Three-Body Computing Constellation and ESA’s ASCEND is whether the cost of deploying and maintaining compute in orbit can fall below the revenue generated by compute services sold to ground users. Today the answer is no. The marginal cost of orbital compute is orders of magnitude higher than cloud data center compute.

What reusable launch economics provide is a credible trajectory toward cost parity. If launch costs fall from $2,700/kg to under $100/kg over the next decade, and satellite mass fractions for compute hardware improve in parallel, the crossover point for specific applications, particularly applications that benefit from low latency to polar or maritime users, could arrive within the 2030s.

The current Starlink orbital computing filing and the Starcloud demonstrations are early tests of that thesis at small scale. The reusable launch economics will determine whether those tests can become viable commercial services.

Path Forward

The reusable launch vehicle market will not grow in isolation from the satellite industry it serves. The two trajectories, launch cost reduction and constellation scale, are coupled. Each supports the other: cheaper launches enable more satellites, more satellites generate more demand for launches, and higher launch demand enables more booster reuse and further cost reduction.

The risks to this trajectory are real. Regulatory constraints on mega-constellations, spectrum congestion, and debris accumulation could limit how many additional satellites the orbital environment can absorb. The reusable launch market is a necessary but not sufficient condition for orbital infrastructure at scale.

For the immediate term, 2026’s launch pace has already validated one thing: the factory-floor approach to satellite deployment, driven by booster reuse, can sustain cadences that would have been technically impossible five years ago.

Official Sources