Goldman Sachs Raises 2031 LEO Satellite Forecast to 300,000
Claire Weston
Goldman Sachs raised its 2031 low-Earth-orbit satellite forecast from 42,000 to 305,000 — a 634% jump driven not by satellite internet but by an entirely new use case: space-based data centers.
How dramatic is this revision?
Goldman's previous 2031 LEO satellite forecast stood at 42,000 units. The new figure — 305,000 — is more than 7× higher.
Near-term numbers barely change: the 2026 forecast rises only ~9% to 13,088. The real break comes in 2029, when the new estimate jumps to 90,042 — a 278% year-on-year surge.
This means → Goldman sees an entirely new demand wave arriving after 2029, bending the growth curve from a linear climb into an exponential takeoff.
What are space data centers — and why do they reshape the curve?
Space data centers — server facilities operating in orbit — become the largest growth driver after 2029. They account for 63% of new installations in 2029, rising to 79% by 2031, or 241,486 satellites.
Two core advantages: virtually unlimited low-cost solar power in orbit, plus edge-computing capability — data processed on the satellite itself, with no need to beam it back to the ground.
In plain terms = the two biggest headaches for ground-based data centers are electricity bills and cooling. Space eliminates both bottlenecks, making it theoretically ideal for compute workloads.
Goldman acknowledges, however, that the technology remains unproven — the single largest source of uncertainty in the entire forecast.
How critical is China's role?
In the base case, China reaches 23,750 satellites by 2031 — 8% of the global total. In the bull case ("blue sky scenario"), that jumps to 101,148 — 26% of the world's fleet.
As of end-2025, Chinese operators have just 253 satellites in orbit, yet have filed applications with the ITU for over 200,000 medium- and low-orbit satellites.
This means → Goldman reads those mass filings as a strategic move to lock in spectrum and orbital slots, not a near-term launch schedule. The biggest gap between the two scenarios comes down to whether China can convert filings into actual hardware in orbit.
Why is rocket technology the hardest bottleneck?
Satellites must be built *and* launched — launch capacity determines whether forecasts become reality. SpaceX currently launches roughly once every three days; in April 2026 it set a record with two rockets in 19 hours.
Next-generation reusable rockets will lift LEO payloads from today's ~17,500 kg to 100–150 tonnes — a 6–9× increase in capacity.
On the Chinese side, the Long March 10 completed a first-stage recovery in July 2026; Zhuque-3 and Galactic Energy's PALLAS-1 both target key test flights in 2026.
Put simply = rocket capacity is the "transport infrastructure" behind the entire forecast — if the road isn't built, the satellites stay on paper.
Can this forecast actually come true?
Goldman flags two layers of uncertainty: technology (can space data centers scale from concept to deployment?) and geopolitics (will China's constellation plans proceed on schedule?).
The bull case puts 2031 global installations at 395,624 — roughly 30% above the base case — but requires above-plan rocket capacity, faster satellite-communications commercialization, and China's 200,000-unit ambition materializing long-term, all at once.
This reflects a bet on a technology inflection point, not a linear extrapolation: if space data centers prove viable, the satellite industry's logic shifts from "communication coverage" to "orbital compute" — a fundamentally different market scale.
Content is for reference only, not financial advice.