The Next Cloud War Could Be in Space: SpaceX vs Amazon
The Cloud Is Moving Beyond Data Centers
For years, the cloud computing battle has been fought inside huge terrestrial data centers. Amazon Web Services, Microsoft Azure and Google Cloud competed over servers, storage, artificial intelligence and enterprise customers. But a new infrastructure layer is emerging above them: low-Earth-orbit satellite networks.
SpaceX and Amazon are approaching that opportunity from opposite directions. SpaceX begins with rockets and Starlink, then moves upward into connectivity, AI and potentially orbital computing. Amazon begins with AWS, one of the world's most important cloud platforms, and is adding Amazon Leo as a satellite extension of that ecosystem. The result is more than a contest over rural broadband. It could become a battle over how data moves from remote devices to the cloud - and eventually where some of that computing happens.
SpaceX Starts With a Massive Lead
SpaceX already operates at a scale that Amazon will need years to match. In its June 2026 quarterly filing, SpaceX reported more than 10,200 Starlink broadband and mobile satellites in low Earth orbit and 12.0 million Starlink subscribers as of 30 June 2026. That subscriber figure excludes many managed enterprise and government customers negotiated directly with the company.
The advantage is not only the size of the constellation. SpaceX designs and manufactures the satellites, operates the network and owns the rockets that deploy most of it. This vertical integration means the company can add capacity without depending entirely on another launch provider. Starship could strengthen that advantage further: SpaceX says its V3 Starlink satellites are designed for much higher capacity and deployment at a scale that Falcon 9 cannot match.
Amazon Is Building a Different Kind of Rival
Amazon is much further behind in orbit. Its second-quarter 2026 results said Amazon Leo had nearly 400 satellites in orbit, enough for the company to begin initial satellite internet service during 2026. The planned constellation is larger than 3,000 satellites, with more than 80 launches booked across multiple launch providers.
Those numbers make Amazon look like a late entrant if the competition is measured only by satellites and subscribers. But Amazon does not need to copy Starlink exactly. Its strongest advantage already exists on the ground: AWS.
Why the AWS Connection Changes the Competition
Amazon Leo has been designed with enterprise networking in mind. The Leo Ultra terminal is intended to deliver up to 1 Gbps download and 400 Mbps upload, but the more important feature for corporate customers may be private connectivity. Amazon says Leo can connect directly to AWS workloads without sending traffic across the public internet.
That creates a simple proposition for businesses operating far from fiber. A mining company, shipping fleet, energy site or remote factory could connect through Leo and route data directly into AWS for storage, analytics and AI. Instead of buying satellite connectivity as a separate service and then building the cloud connection around it, Amazon can make the satellite link part of the cloud architecture itself.
Enterprise Customers Could Be the First Real Battlefield
The early commercial fight is likely to appear in industries where reliable connectivity matters more than consumer pricing. Aviation, maritime transport, logistics, energy, agriculture and telecommunications all operate assets in locations where terrestrial networks are weak or unavailable.
SpaceX already sells Starlink for fixed businesses, ships, aircraft and mobile operations. Its scale gives it a large installed base and years of operational data. Amazon, however, is using existing enterprise relationships to build its own pipeline. Delta Air Lines plans to begin installing Amazon Leo on an initial 500 aircraft in 2028, and Amazon has announced agreements with companies in aviation, telecommunications, logistics and energy.
This is where the two strategies become clear. SpaceX can tell a customer that it already has a global network and controls the infrastructure used to expand it. Amazon can tell the same customer that connectivity can become another native part of the AWS environment it may already use. Both approaches reduce complexity, but from different ends of the technology stack.
SpaceX Has a Wild Card: Computing in Orbit
The most ambitious part of the SpaceX strategy is no longer connectivity. In its 2026 prospectus, the company described plans to expand from terrestrial AI infrastructure toward orbital AI compute. SpaceX says it intends to deploy its first modular orbital AI compute systems and begin monetizing capacity toward the end of the decade.
This remains a long-term plan rather than an established cloud business, and the engineering challenges are substantial. Hardware in orbit must deal with radiation, thermal management, maintenance, launch costs and limited physical access. Yet the strategic idea is important: SpaceX is not only asking how to connect users to data centers. It is asking whether some future data centers should leave Earth.
Two Stacks, Two Very Different Advantages
The competition can be understood as two technology stacks growing toward each other. SpaceX begins with launch, adds satellites and connectivity, and is now adding AI infrastructure and a roadmap for orbital compute. Amazon begins with cloud computing and enterprise software, then adds satellites as a new access network into AWS.
SpaceX therefore has the stronger physical infrastructure position. Its launch cadence, satellite manufacturing and existing Starlink network are difficult to reproduce. Amazon has the stronger cloud distribution position. AWS already sits inside the technology architecture of companies and governments around the world, giving Leo a natural route into enterprise budgets and workloads.
Direct-to-Device Could Expand the Market Again
A second front is emerging around phones and connected devices. SpaceX has been developing Starlink Direct to Cell to extend mobile coverage beyond terrestrial towers. Amazon moved in the same direction in 2026 by announcing an agreement to acquire Globalstar, giving it access to satellite assets, spectrum and operating experience for future direct-to-device services.
If satellite connections eventually become a normal fallback for smartphones, vehicles and sensors, the addressable market becomes far larger than households willing to install a dish. Connectivity from orbit could become an invisible layer of the internet - available whenever terrestrial infrastructure disappears.
So Who Is Winning?
Today, SpaceX is clearly ahead in satellite broadband. Starlink has the operating constellation, subscribers, launch infrastructure and commercial experience. Amazon Leo is still moving from deployment into service and has a large gap to close.
But the long-term cloud question is less settled. Amazon does not need to beat Starlink in every consumer market if Leo becomes a valuable enterprise on-ramp to AWS. SpaceX, meanwhile, does not need to remain only an internet provider if its AI infrastructure and orbital-compute plans become commercially viable.
The most important competition may therefore be over control of the full data path: how information is collected, transported, processed and delivered. The cloud was once defined by buildings filled with servers. By the end of this decade, part of that infrastructure may extend into orbit - and SpaceX and Amazon are already building two very different routes to get there.
