Goldman Sachs lifts 2031 LEO satellite forecast 634% on space data centers

Goldman Sachs now expects 305,293 low-Earth orbit satellites by 2031, a 634% increase from its prior forecast, driven by a new demand source: space-based data centers.

Goldman Sachs raised its 2031 low-Earth orbit satellite installation forecast by 634% to 305,293 units, citing space data centers as the primary growth engine starting in 2029, according to a July 10 research report.

"Space data centers offer unlimited access to low-cost solar power and edge computing capabilities that can process satellite data directly in orbit," the Goldman Sachs research team wrote in the report, marking a structural shift in how Wall Street values the LEO satellite industry.

The revision represents one of the largest single-forecast adjustments in the sector's history. Goldman's previous baseline of 42,000 units for 2031 has been superseded by a model in which space data centers account for 79% of all new LEO installations by that year, or 241,486 satellites. Satellite internet, the traditional demand driver, would contribute just 63,807 units. The inflection point arrives in 2029, when space data center installations leap to 63% of new capacity from zero.

The forecast implies a massive expansion in the total addressable market for satellite manufacturers, launch providers and ground equipment makers. Companies such as SpaceX, AST SpaceMobile and Rocket Lab stand to benefit, though Goldman cautioned that the space data center concept remains unproven at scale. Chinese operators, which have filed for more than 200,000 orbital slots with the International Telecommunication Union, represent a structural wild card.

The China Variable

Under Goldman's baseline scenario, China accounts for 23,750 LEO satellites by 2031, or 8% of the global total. In the firm's "blue sky" scenario — which assumes better-than-expected rocket capacity, faster satellite communications commercialization and full execution of Chinese constellation plans — that figure jumps to 101,148 units, or 26% of the global fleet. China currently operates just 253 LEO satellites, making the gap between its filed applications and actual deployment the widest in the industry.

Rocket Capacity as the Binding Constraint

The entire forecast hinges on launch vehicle availability. Current reusable rockets can deliver about 17,500 kilograms to LEO per flight. Next-generation vehicles, targeting 100 to 150 tons of payload capacity, would represent a sixfold to ninefold improvement. SpaceX already launches Starlink satellites every three days on average and set a record in April 2026 with two launches in 19 hours. In China, four reusable rocket programs are advancing: the Long March 10 completed a first-stage soft landing splash test in February 2026 and a net recovery on July 10; the Zhuque-3 achieved a partial test flight success in December 2025 with a recovery test flight planned for the second quarter of 2026; the Hyperbola-3 targets its first launch by the end of 2026; and Galactic Energy's PALLAS-1 aims for a debut launch this year.

For investors, the key question is whether space data centers can transition from concept to commercial reality within three years. If they do, the satellite industry's addressable market expands by an order of magnitude, benefiting launch providers and satellite manufacturers. If they don't, the 2029 inflection point in Goldman's model disappears, and the 634% forecast revision becomes a cautionary tale about extrapolating unproven technology. SpaceX's Starlink, the only operational mega-constellation, remains the benchmark against which all other LEO business models are measured.

This article is for informational purposes only and does not constitute investment advice.