SpaceX Orbital Data Center Valuation Cockpit (beta)

Sensitivity Analysis (Tornado)

Each bar shows how far the value today moves when one assumption is set lower or higher, with everything else unchanged. Red is the downside, green the upside. The longer the bar, the more that assumption matters.

Impact of changing each variable to its Low / High value (others at base) on PV Valuation ($ Trillions). The dashed line marks the current SPCX market cap. Red shows the downside and green the upside relative to the current case.

Path to the 2040 end state

Monte Carlo: range of valuation outcomesPrototype

Nobody knows the exact numbers, so the model runs 10,000 scenarios, each with slightly different assumptions around your settings. The chart shows how often each value came up: red bars are below today’s market value, green bars above it. Switch on the Advanced view to see and change the ranges.

Illustrative only. 10,000 simulated outcomes. Each draw picks the inputs from skewed ranges around your current slider settings, keeps related inputs consistent (more flights, lower launch cost), and treats the discount rate and the tax rate as fixed at your slider values, because it already prices in execution risk. The spread reflects the assumed ranges (shown in the Advanced view), not data. The chart's right edge is the 99th percentile; the last bar collects everything above it.

How this model works, and what it leaves out

• Orbital against ground, per megawatt. The model compares what one megawatt of AI compute costs per year in orbit with what it costs in a ground data center. Whatever orbital saves, times the compute capacity deployed, is the annual saving. It is a cost comparison, not a revenue forecast.
• Ground side. The yearly cost of a ground data center is the depreciation of its buildings, power and cooling systems (about $1.86M per MW), its electricity (8¢ per kWh at a power usage effectiveness of 1.2, running all year), and about $0.61M for maintenance, staff, property tax, insurance and network. At the base case that adds up to $3.31M per MW per year.
• Orbital side. Each AI satellite is rated at 150 kW of solar power, delivers 120 kW of compute on average, weighs about 2.1 tonnes and costs the production-cost slider ($0.75M in the base case). Starship carries 70 of them per launch (150 tonnes to sun-synchronous orbit) at the launch-cost slider. Production and launch cost are spread over the satellite lifetime (5 years in the base case). There is no electricity bill in orbit; ground stations, staff, insurance and network add $0.35M per MW per year. Capacity is counted at peak power, as in the research report.
• Deployed capacity. The end state is a steady state: the terminal launch rate (10,000 flights a year in the base case) sustained for one full satellite lifetime, giving 10.5 MW per launch × 10,000 × 5 years = 525 GW in orbit. The model does not cap this against demand, so it assumes the market can use that much compute.
• Valuation. The annual saving is taxed at 21% (a P/E multiple applies to after-tax earnings), multiplied by the terminal P/E, and discounted from the end-state year back to today. The end-state year slider only changes how many years that is: the same end state reached later is worth less today. The discount rate is a real rate (16% in the base case), because all values are in constant 2025 dollars with no inflation added. A less aggressive plan carries less risk, which is why the Lower-Risk Case uses a lower rate.
• Who keeps the saving. The model assumes SpaceX sells orbital compute at the price of ground compute and keeps the whole saving as operating profit. If competition or customers take part of it, the value is lower.
• Cash flows until the end state. The model counts only the value of the business in the end-state year. It assumes all profits up to then are reinvested in building the satellites, the launch fleet and the ground segment, so no interim cash flow is received and no separate cost of building the constellation is deducted. The ramp-up path from today to the end state is not modeled; the path charts and the week simulation are illustrations of the 2040 base case.
• What the comparison leaves out. AI chips are excluded on both sides (the satellite cost excludes AI silicon and the ground cost covers the facility, not the servers), so the model compares infrastructure, power and launch. No cost of capital is charged on either side, and ground costs are held at today's level. Launch failures, space debris, radiation damage, communication delays, spectrum and regulation are not modeled. Only the orbital data center business is valued, not Starlink or any other SpaceX business.
• Timing and use. Present value is as of today so that it compares with today's market cap. The Monte Carlo ranges are judgment, not data. This is a scenario tool built on stated assumptions, not a forecast or investment advice.