Google and SpaceX in Talks to Build Orbital Data Centres for AI Computing

Space-based infrastructure pitched as future of AI compute, though costs remain significantly higher than ground-based alternatives

edit

By LineZotpaper

Published13 May 2026

Read Time3 min

Google and SpaceX are in discussions to construct data centres in Earth's orbit, according to a report by TechCrunch published Monday, with the two companies positioning space as a long-term home for artificial intelligence computing infrastructure — despite current costs far exceeding those of conventional land-based facilities.

Google and SpaceX are exploring a partnership that would place data centres into orbit, marking one of the most ambitious proposals yet to expand computing infrastructure beyond Earth's surface, according to a report from TechCrunch.

The talks represent a convergence of two of the technology sector's most resource-intensive ambitions: Google's surging demand for AI computing power and SpaceX's push to commercialise low Earth orbit through its Starlink satellite network and Falcon rocket capabilities.

Details of the proposed arrangement remain limited. It is unclear at this stage what form the orbital data centres would take, how they would be powered, or how data would be transmitted to and from Earth at the speeds required for AI workloads. Neither Google nor SpaceX has made a public statement confirming the discussions.

Why Consider Space?

Proponents of space-based computing point to several theoretical advantages. Satellites in orbit have uninterrupted access to solar energy and the vacuum of space offers natural cooling — two factors that represent enormous operational costs for traditional data centres on the ground. As AI model training and inference demand ever-greater compute density, the appeal of an environment unconstrained by terrestrial land, power grid, and water cooling limitations grows.

For SpaceX, a deal of this kind would represent a significant commercial milestone, turning its launch and satellite infrastructure into a platform for enterprise cloud services — extending its business well beyond rocket launches and consumer broadband.

Costs Remain a Major Hurdle

Despite the theoretical appeal, the economics of space-based computing remain deeply challenging. Launching hardware into orbit costs orders of magnitude more than installing equivalent equipment in a ground-based facility. Maintenance, repair, and hardware replacement — routine tasks in a terrestrial data centre — are enormously complicated and expensive in space.

Industry analysts have long cautioned that space computing is a long-term proposition at best. For the foreseeable future, hyperscale cloud providers such as Google, Microsoft, and Amazon are expected to continue expanding their ground-based infrastructure to meet AI demand.

The report does not indicate a timeline for any agreement, and preliminary discussions between large technology companies frequently do not result in formal partnerships.

Broader Context

The talks come as the global race to secure AI computing capacity intensifies. Google, Microsoft, Amazon, and Meta have each committed to spending tens of billions of dollars on data centre expansion in 2025 and 2026. At the same time, concerns around energy consumption, water usage, and the availability of suitable land are placing pressure on the industry to explore unconventional solutions.

SpaceX, meanwhile, has been actively courting enterprise customers for its Starlink network and has signalled broader ambitions to monetise its position in low Earth orbit.

Analysis

Why This Matters

If realised, orbital data centres could reshape how AI infrastructure is planned and financed over the coming decades, creating a new class of space-based commercial asset.
The talks signal that Big Tech's appetite for compute is pushing companies to explore infrastructure solutions once considered the domain of science fiction.
A Google-SpaceX partnership of this scale would deepen the commercial stakes in low Earth orbit, with implications for regulation, international competition, and space sustainability.

Background

The idea of computing in space is not new — researchers and startups have proposed satellite-based edge computing for years — but it has remained largely theoretical due to prohibitive launch costs and the complexity of operating hardware in orbit.

The landscape began to shift as SpaceX dramatically reduced the cost of reaching orbit through reusable rockets and began deploying its Starlink constellation of thousands of satellites. This gave the company both launch capability and an existing orbital infrastructure that could, in principle, support additional payloads.

Meanwhile, the explosion in demand for AI computing since 2022 has strained terrestrial data centre supply chains, power grids, and cooling water supplies. Hyperscalers are facing growing political and environmental resistance to new ground-based facilities in many regions, pushing strategists to think creatively about future capacity.

Key Perspectives

Google: As one of the world's largest consumers of computing infrastructure, Google has strong incentives to diversify its capacity options and explore long-term alternatives before terrestrial constraints become acute. A space-based solution could also provide strategic differentiation from cloud rivals.

SpaceX: An agreement with Google would validate SpaceX's ambition to turn its launch and orbital assets into an enterprise platform, generating recurring revenue that complements its launch business and Starlink consumer subscriptions.

Critics/Skeptics: Many engineers and economists argue the economics simply do not work at present — launch costs, radiation hardening of hardware, latency, and the impossibility of on-site maintenance make space data centres far more expensive and less reliable than ground alternatives. Some also raise concerns about adding to orbital congestion and space debris.

What to Watch

Any formal announcement or memorandum of understanding between Google Cloud and SpaceX, which would signal talks have advanced beyond early-stage exploration.
Developments in launch cost reduction and in-space servicing technology, which are the primary economic barriers to making orbital computing viable.
Regulatory scrutiny from bodies such as the FCC and international counterparts regarding spectrum allocation and orbital slot permissions for data-transmitting satellite infrastructure.

Sources

Report: Google and SpaceX in talks to put data centers into orbit — TechCrunch