Orbital AI Data Centers — Economic Competitiveness Timeline > Pages > What is the cost per kilogram to deliver payload to LEO, and how will it evolve from 2026 through 2040?

Launch Cost per Kilogram to LEO

What is the cost per kilogram to deliver payload to LEO, and how will it evolve from 2026 through 2040?

Answer

Current customer pricing to LEO ranges from $2,500-7,000/kg depending on vehicle and mission type (Falcon 9 dedicated vs. rideshare). SpaceX's internal marginal cost is far lower, approximately $629/kg on Falcon 9. Starship, once operational for payload delivery (expected 2027-2028 for commercial customers), will drive dramatic cost reductions, but there is a critical distinction between SpaceX's internal cost and the price charged to external customers. For a vertically integrated SpaceX-xAI operation, internal transfer pricing could approach marginal cost. Central estimate: costs fall from ~$2,500/kg in 2026 to ~$500/kg by 2030 and ~$75/kg by 2040. Optimistic: $100/kg by 2030 and $20/kg by 2040. Conservative: $1,200/kg by 2030 and $300/kg by 2040.

Evidence

Current Pricing (2025-2026)

  1. Falcon 9 customer price is ~$2,500/kg; Falcon Heavy ~$1,500/kg, representing a 30x reduction from Space Shuttle era costs. -- aei-launch-costs

  2. As of December 2025, Falcon 9 rideshare missions to LEO cost roughly $6,500/kg, an increase from a low of $5,000/kg before inflation drove up costs. -- arena-space-lasers

  3. In March 2026, SpaceX increased Falcon 9 dedicated launch prices from $70M to $74M and rideshare from $6,500/kg to $7,000/kg, contradicting the narrative of continuously falling launch costs. -- payload-falcon9-price-hike

  4. SpaceX's internal marginal cost on Falcon 9 is ~$629/kg (~25% of $2,600/kg customer price). Total marginal launch cost ~$10.5-11M. Upper stage ~$7M, booster amortized ~$1M, propellant ~$250K. -- nbf-falcon9-true-cost

  5. Falcon 9 reusable: $1,500/kg (at full payload utilization); expendable: $2,720/kg. Falcon Heavy: $1,400/kg. New Glenn: ~$3,000/kg. Vulcan Centaur: ~$4,000/kg. -- spacenexus-launch-economics

  6. Reusable Falcon 9 currently delivers cost to orbit of roughly $3,600/kg per TechCrunch's assessment, citing a different payload utilization assumption than the theoretical maximum. -- techcrunch-orbital-brutal

  7. Current Falcon 9 cost to orbit ~$2,700/kg. Orbital data center viability requires <$100/kg, a 27x reduction. -- introl-orbital-dc-race-2026

Starship Projections

  1. Starship build cost ~$90M. Per-flight marginal cost target $2M. At 6 reuses: $94/kg; 20 reuses: $33/kg; 50 reuses: $19/kg; 70 reuses: $14/kg (at 200t payload capacity). -- nbf-starship-roadmap

  2. At 240t payload (higher estimate), costs drop further: 6 reuses: $78/kg; 20 reuses: $27/kg; 70 reuses: $14/kg. -- nbf-starship-roadmap

  3. First known Starship customer price: $90M for Voyager Technologies Starlab station launch in 2029, implying ~$600/kg at 150t payload capacity. -- indexbox-starship-90m

  4. Starship target cost: $10-50/kg to LEO. If achieved, this represents a 1,000-5,000x decrease from Shuttle era. -- spacenexus-launch-economics

  5. Musk has set an aspirational target of $10/kg for Starship, with propellant costs ~one-third. This would require a $1.5M total launch cost for 150 tons. Citigroup notes this "would require significant improvements in materials design, propulsion technology, and operating costs." -- aei-launch-costs

Analyst Forecasts

  1. Citigroup 2022 research note: base case ~$100/kg by 2040. Bull case ~$33/kg by 2040 (later revised to ~$30/kg in other reporting). Bear case ~$300/kg (rockets reused only ~10 times). -- citi-gps-space-2022

  2. Bain & Company envisions Starship reducing cost per kg by 50-80x from current levels, implying $31-50/kg from a ~$1,500-2,500/kg baseline. -- aei-launch-costs

  3. Google's Project Suncatcher white paper states economic viability for orbital data centers requires launch costs below $200/kg, projected to be achievable in the mid-2030s. -- google-suncatcher

  4. Space-based data center proponents assume costs will reach $350-500/kg "in the near future" for designs to become competitive with ground-based data centers. -- arena-space-lasers

Starship Development Status

  1. Starship V3 first launch approaching in March 2026. Payload capacity target: 100t to LEO. Upgraded Raptor V3 engines (280tf, ~50% more than Raptor 1). Not yet operational for commercial payload delivery. -- payload-falcon9-price-hike

  2. Rational Futures consultancy argues SpaceX will not charge much less than its best competitor (e.g., Blue Origin's New Glenn at ~$70M). This means external customer prices may not reflect internal cost reductions for years. -- techcrunch-orbital-brutal

  3. Musk envisions 10,000+ annual Starship launches, requiring as few as 20-30 physical Starships cycling every ~30 hours. SpaceX "will launch more AI than the cumulative amount on Earth of everything else combined." -- musk-2026

Pricing Strategy & Internal vs. External Cost

  1. SpaceX sells Falcon 9 launches at major markup over internal cost. Cost savings fund Starlink development rather than benefit external customers. Company has no competitive pressure to lower customer prices. -- spacenews-categorical-imperative

  2. The ratio of internal cost to customer price on Falcon 9 is approximately 1:4 (25% cost, 75% margin). For a vertically integrated SpaceX-xAI, orbital data center launches would be at internal cost, not customer price. -- nbf-falcon9-true-cost

Learning Curves

  1. Per Aspera projects <$500/kg to LEO pricing achievable "if Starship and/or reusable rockets from others succeed," placing this in their mid-2030s "Walk" phase. -- peraspera-realities

  2. The $39B 2035 orbital DC market projection "assumes Starship economics. If launch costs remain above $500/kg, the addressable market shrinks substantially." -- introl-orbital-dc-race-2026

Analysis

Current State: A Wide Gap Between Cost and Price

The launch cost landscape in 2026 is characterized by a striking divergence between SpaceX's internal costs and customer pricing. Falcon 9's customer price has actually been increasing -- from $62M to $67M to $70M and now $74M for dedicated launches ([evidence:payload-falcon9-price-hike.1], Falcon 9 price rose to $74M in March 2026), while rideshare pricing has climbed from $5,000/kg to $7,000/kg. Yet SpaceX's internal marginal cost has been falling; at ~$629/kg ([evidence:nbf-falcon9-true-cost.1]), SpaceX captures roughly 75% gross margin on external launches.

This cost-price gap is crucial for modeling orbital data center economics. SpaceX-xAI, as a vertically integrated entity, would launch its own satellites at internal marginal cost, not at the external customer price. This means the "effective" launch cost for xAI orbital data centers could be 2-4x lower than what third-party operators face. The analysis below distinguishes between internal cost (relevant for SpaceX-xAI) and market price (relevant for all other operators).

Starship: The Pivot Point

Starship is not yet operational for commercial payload delivery. As of March 2026, the vehicle is in its test flight phase (Flight 12 approaching, first V3 flight imminent -- [evidence:payload-falcon9-price-hike.1]). The vehicle targets 100-150t to LEO in reusable configuration and 200-300t expendable.

The cost trajectory for Starship depends critically on three variables:

1. Reuse rate achieved. The NextBigFuture analysis ([evidence:nbf-starship-roadmap.1]) shows costs are extremely sensitive to reuse count: $94/kg at 6 flights drops to $33/kg at 20 flights and $14/kg at 70 flights. For context, Falcon 9 boosters have achieved 25+ reuses, so high Starship reuse is plausible but not guaranteed -- Starship is far more complex than a Falcon 9 booster, and its upper stage (Ship) must also survive reentry, which is an unsolved challenge at scale.

2. Flight cadence. Musk envisions 10,000+ flights per year ([opinion:musk-2026.1]), which would amortize fixed costs aggressively. Current Falcon 9 cadence is ~100 flights/year. Going from ~100 to 10,000 requires a 100x scaling in launch infrastructure, propellant supply, payload availability, and regulatory throughput. Even reaching 500-1,000 flights/year by 2030 would be extraordinary.

3. Payload utilization. The nbf-starship-roadmap figures assume full 200t payload utilization. Real-world payload utilization is typically 50-80% due to volume constraints, orbit requirements, and integration factors. This could increase effective $/kg by 25-100%.

The Critical Distinction: Cost vs. Price vs. Internal Transfer Price

Three separate cost concepts are relevant:

For the orbital data center use case, the relevant figure is the internal transfer price for SpaceX-xAI, or the customer price for third-party operators. The TechCrunch/Rational Futures analysis ([opinion:techcrunch-orbital-brutal.2]) argues convincingly that SpaceX will price Starship just below competitors like New Glenn (~$70M) for external customers, maintaining fat margins rather than passing cost savings through.

Scenario Construction

Optimistic scenario assumes: Starship V3 achieves operational status in 2027; rapid ramp to 20+ reuses by 2028; 500+ flights/year by 2030; xAI uses internal transfer pricing near fully-loaded cost. By 2030, Starship achieves 20-30 reuses at scale with high payload utilization, yielding ~$100/kg. By 2040, 50-70+ reuses and thousands of flights/year drive costs to ~$20/kg. This aligns with the optimistic end of Citigroup's $30/kg bull case ([evidence:citi-gps-space-2022.1]) and nbf's 50-reuse projection of $19/kg ([evidence:nbf-starship-roadmap.1]).

Central scenario assumes: Starship becomes operational for payload delivery in late 2027/2028; moderate ramp in reuse (10-15 flights per vehicle by 2030); 100-300 flights/year by 2030; customer pricing remains well above marginal cost for external buyers; SpaceX-xAI gets ~50% discount vs. external price. By 2030, effective cost ~$500/kg. By 2040, ~$75/kg as Starship matures and competition from Blue Origin's New Glenn pressures prices down.

Conservative scenario assumes: Starship development delays (operational 2029+); reuse proves harder than expected (limited to 5-10 flights per vehicle through 2035); cadence ramps slowly; upper stage reusability never fully solved; competition remains limited. This aligns with Citigroup's bear case of $300/kg by 2040 ([evidence:citi-gps-space-2022.1]) and the ESPI assessment that $10M/flight Starship is "unrealistic in the near-term" ([evidence:introl-orbital-dc-race-2026.1]).

Key Uncertainties

  1. Starship upper stage reusability: The booster (Super Heavy) has already demonstrated catch landing, but the Ship upper stage must survive reentry from orbital velocity -- a fundamentally harder problem than booster recovery. If Ship reusability fails, costs remain at the $500-1,000/kg level indefinitely.

  2. Cadence scaling: Going from 100 Falcon 9 flights/year to 1,000+ Starship flights/year requires massive infrastructure investment. Launch pad availability, propellant production, and regulatory approvals are all bottlenecks.

  3. Competition effects: Blue Origin's New Glenn, Rocket Lab's Neutron, and potentially Chinese competitors could either pressure SpaceX to lower prices or alternatively establish a price floor well above marginal cost. Blue Origin filing Project Sunrise for 51,600 orbital DC satellites suggests competition is emerging.

  4. Internal vs. external pricing: The SpaceX-xAI merger creates a vertically integrated entity that can launch at internal cost. This is a structural advantage of perhaps 2-4x over competitors using market pricing. Our central and optimistic scenarios assume access to something approaching internal pricing for the cost-optimized deployment context specified in the assumptions.

  5. Payload utilization: Real-world $/kg is typically 25-100% higher than theoretical $/kg due to payload integration, volume constraints, and orbit-specific requirements. The values above are effective costs, not theoretical minima.