SpaceX Revenue of $731.1 Billion by 2028? What Sequoia's Maguire Is Betting On
Shaun Maguire's model is less about launch counts than about linking rockets, connectivity, power and AI compute into one cost advantage. It is a shareholder's hypothesis, not reported results.
1. Who forecast $731.1 billion in 2028 revenue?
Shaun Maguire, a partner at Sequoia Capital, projects SpaceX revenue of $731.111 billion (731,111 million) in 2028. The path runs through $51.482 billion in 2026 and $273.757 billion in 2027. He points to Nvidia's revenue surge over the past three years as a comparison.
Maguire is a physicist and has led Sequoia's SpaceX investment since 2020. He also disclosed that Sequoia holds a large stake in SpaceX, so he is "extremely biased." The forecast is not reported results. It is his own 50th-percentile scenario.
He built an internal model in the fourth quarter of 2020, when Starlink was recording its first revenue. That model put 2026 Starlink annual revenue at $25,218,103,720. At the time it looked absurd, he recalls. Here too, what matters is less the record of past predictions than whether the new assumptions can come true.
2. When a rocket explodes, excuses don't work
Maguire's first competitive edge is people and organization. At SpaceX, whether a rocket flies or explodes is the harshest measure. Failures have to be traced physically, so technical results outrank internal politics, he argues. In his words, "physics beats politics."
He gives examples: if turbines are a bottleneck, build them in-house; if a power plant permit is blocked in Tennessee, run a plant in Mississippi and connect transmission lines. These are less verified facts about individual projects than illustrations of an execution culture that does not push problems outside the organization.
He sees a gigawatt-scale AI data center as 10 to 100 times harder than the data centers of tens of megawatts in the past. If that is right, the winners in AI may be decided not only by chip purchasing power but by the ability to solve power, facilities, networking and transport at once.
3. Starship's moat is the time rivals need to catch up
Maguire defines a moat this way: give the best people unlimited resources, and how long does it take them to reach today's technology? In his view, Starship is among the most complex systems humanity has built.
He estimates China is about five years behind TSMC in semiconductors and about ten years behind SpaceX in launch. He notes SpaceX's first booster recovery came in 2015 and China's in 2026, and argues the chip gap is narrowing while the space gap is widening. This is Maguire's assessment, not an independently confirmed industry gap.
The core is Starship's vertical integration. If design, production, testing and supply chain are combined in-house, a competitor must catch up not with one rocket but with the whole system that builds and improves it repeatedly. A large technical moat does not, however, guarantee a high return on investment.
4. Can SpaceX also produce AI compute most cheaply?
Maguire sees AI compute as closer to oil production than to the broadband overbuilt in the dot-com era. If prices fall, higher-cost suppliers are squeezed first and low-cost ones can hold up better.
He argues SpaceX could keep its edge even if compute cost per gigawatt falls to about $10 billion. He also considers it very unlikely that cost drops below $20 billion per gigawatt within roughly four years. These are his views, not confirmed market prices.
He stresses the assumption that revenue per gigawatt and capex per gigawatt move together. If selling prices fall and build costs fall in a similar direction, a low-cost operator has room to add capacity. He concedes, though, that power permitting and supply delays mean the revenue decline is not fully offset. The Anthropic customer concentration and contract termination risk flagged in an earlier Insight Times article is not removed by this logic alone. Cheap capacity can still see utilization fall if customers leave.
5. The $5 billion economics of one Starship launch
This is the most intuitive calculation of Starship's economics. Maguire puts the current Starlink constellation's annual revenue run rate at about $20 billion, built with about 400 Falcon 9 launches. He assumes one Starship can deploy satellites equivalent to about 20 Falcon 9 launches.
| Step | Maguire's assumption and result |
|---|---|
| Current constellation | About 400 Falcon 9 launches, annual revenue run rate of about $20 billion |
| Launch capacity conversion | 1 Starship = about 20 Falcon 9 |
| Equivalent constellation | About 20 Starship launches |
| Supported revenue per launch | About $1 billion a year |
| Assumed life of about 5 years | About $5 billion in lifetime revenue per launch |
The $5 billion is neither the selling price of a launch nor the net profit from one. It is a simple conversion of the revenue the satellites deployed in that launch could support over their lifetimes. Actual profit depends on the cost of building satellites, launching, operating and acquiring customers.
6. After Starlink: phones and orbital compute
Direct to Cell (DTC) lets ordinary phones connect directly to the satellite network. Maguire bets DTC revenue will pass Starlink's in 2029 or 2030. It is not current reported results but an outlook in which satellite deployment costs fall and the connectable market widens.
Orbital computing is the idea of placing compute facilities in space. Maguire sees its economic effect starting in earnest in 2029 and expects SpaceX to be as vertically integrated in chips as in rockets ten years from now. Between the technical concept and commercial monetization, steps remain to be verified.
7. What has to go right for the 2028 forecast
Compressed into annual figures, Maguire's company-wide model looks like this. All numbers are his model's, and should not be read as reported results or official company guidance.
| Metric | 2026 | 2027 forecast | 2028 forecast |
|---|---|---|---|
| Total revenue ($M) | 51,482 | 273,757 | 731,111 |
| Space segment revenue ($M) | 5,239 | 9,389 | 17,345 |
| Connectivity segment revenue ($M) | 19,094 | 40,567 | 88,538 |
| Total SI revenue ($M) | 27,149.8 | 223,802 | 625,228 |
| Operating profit ($M) | 10,256 | 167,374 | 483,855 |
| Capex ($M) | 88,008 | 290,015 | 368,992 |
| Change in cash ($M) | -77,752 | -122,641 | 114,863 |
| Starship flights per year | 2 | 38 | 250 |
| Year-end ground compute capacity | 2.5 GW | 10.2 GW | 21.0 GW |
For 2028 revenue of $731.111 billion and operating profit of $483.855 billion to hold, more Starship flights, gigawatt-scale buildout, compute sales and cost control all have to line up at once. In the model, the change in cash is -$77.752 billion in 2026 and -$122.641 billion in 2027, then turns to +$114.863 billion in 2028.
Maguire himself acknowledges wide error margins on revenue per gigawatt, capex per gigawatt and "terafab" capex, and that his post does not justify most of its numbers in detail. The model is less a target for investors to accept as is than a hypothesis showing what to verify.
What to Watch
- Starship flights: whether actual operations approach the model's path of 38 and 250 flights a year.
- Ground compute capacity: whether the expansion to 10.2 GW and 21.0 GW grows alongside customer usage.
- Revenue and capex per gigawatt: how much build-cost savings absorb falling selling prices.
- Change in cash: whether it turns positive in 2028 after the heavy investment.
- Contracts and utilization: whether major customers stay and replacements can be found.
Insight Times Editorial Desk
