Google is reportedly in advanced discussions with SpaceX to launch artificial intelligence data centers into orbit, according to a new report from the Wall Street Journal. The deal would see SpaceX rockets carrying Google's server infrastructure beyond Earth, tapping into the growing belief that space-based computing is essential for the next generation of AI workloads.

The talks come just months after SpaceX acquired Elon Musk's AI company, xAI, in a move Musk framed as necessary to scale AI beyond the constraints of terrestrial data centers. In a statement at the time of the acquisition, Musk argued that "current advances in AI are dependent on large terrestrial data centers, which require immense amounts of power and cooling." He added that "global electricity demand for AI simply cannot be met with terrestrial solutions, even in the near term, without imposing hardship on communities and the environment. In the long term, space-based AI is obviously the only way to scale."

Google's interest in orbital data centers was first hinted at late last year when the company announced Project Suncatcher, an initiative to launch prototype satellites by 2027. The project aims to "one day scale machine learning compute in space," according to internal documents reviewed by the Wall Street Journal. Google CEO Sundar Pichai confirmed the company's ambitions in February during the AI Impact Summit in New Delhi, India, remarking that he never imagined he'd "one day be spending time with teams figuring out how to put data centers into space."

The Terrestrial Data Center Bottleneck

Traditional data centers are the backbone of the modern internet and cloud computing, but they are struggling to keep up with the exponential growth of artificial intelligence. Training large language models such as GPT-4 or Google's Gemini requires vast amounts of electricity—often measured in megawatts per facility—and generates enormous heat that must be actively cooled. Many data centers rely on water or refrigerants, straining local water supplies and contributing to carbon emissions. As AI adoption accelerates, analysts predict that data centers could consume up to 10% of global electricity by 2030, up from roughly 1% today.

Space offers a compelling alternative. In orbit, solar panels can generate power continuously without atmospheric interference, potentially reducing reliance on fossil fuels. The vacuum of space naturally provides near-perfect insulation, and heat can be radiated away efficiently using passive cooling systems. Moreover, latency-sensitive AI applications could benefit from orbital relays that connect directly to satellites, bypassing congested terrestrial networks.

However, space-based data centers come with their own challenges. Launching heavy server racks into orbit is extremely expensive, and current launch costs—while declining thanks to reusable rockets like SpaceX's Falcon 9—still run into thousands of dollars per kilogram. Powering and cooling equipment in microgravity requires specialized engineering, and maintaining uptime through radiation hardening and debris avoidance adds complexity. Despite these hurdles, companies like SpaceX, Google, and even Microsoft have been exploring the concept for years.

A History of Space Computing

The idea of putting computers in space is not new. Since the dawn of the Space Age, satellites have carried onboard processors for navigation, communication, and data collection. But only recently has the sheer processing power required for AI made terrestrial expansion impractical, pushing the industry to look upward. In 2023, Microsoft partnered with the U.S. Department of Defense to test edge computing in orbit using its Azure cloud platform. Amazon Web Services has also launched its AWS Snowcone devices on the International Space Station to demonstrate space-based data processing.

SpaceX's ambitions go beyond mere experimentation. The company recently filed with the Federal Communications Commission (FCC) for permission to launch "a million satellites" to support a space-based AI infrastructure. While such a massive constellation seems far-fetched, SpaceX already operates the Starlink constellation of over 6,000 satellites, providing low-latency internet globally. Adding compute nodes to these satellites—or launching dedicated orbital data centers—could create a distributed AI network accessible from anywhere on Earth.

Google's Project Suncatcher is specifically designed to test the feasibility of hosting machine learning workloads in space. The prototype satellites are expected to carry purpose-built chips optimized for AI inference, such as Google's own Tensor Processing Units (TPUs). Google has declined to provide a timeline beyond the 2027 target, but the company's track record with space—having launched experimental satellites through its Google Maps and Earth initiatives—suggests it is serious about making orbital computing a commercial reality.

If the deal with SpaceX goes through, Google would be the first major cloud provider to secure dedicated launch capacity for orbital data centers. The partnership could also give SpaceX a substantial revenue stream ahead of its planned IPO, which is expected to value the company at $1.75 trillion. A deal of this scale would send a strong signal to investors that SpaceX is not just a rocket manufacturer but a full-service space infrastructure provider capable of supporting next-generation computing.

Environmental and Regulatory Considerations

Proponents of space-based AI argue that it could reduce the environmental footprint of data centers by leveraging orbital solar power and eliminating the need for water cooling. However, critics point out that launching rockets produces significant carbon emissions and space debris. Each Falcon 9 launch releases roughly 200-300 metric tons of CO2, though reusable boosters reduce the per-kilogram impact over time. Furthermore, orbital data centers would require ongoing maintenance, which could increase the frequency of launches.

Regulators are also taking note. The FCC is currently reviewing SpaceX's application for a million-satellite constellation, which would dwarf existing orbital traffic and raise concerns about collision avoidance and light pollution. The International Telecommunication Union (ITU) has also set limits on geostationary orbit slots, but low Earth orbit is currently unregulated in terms of satellite density. Google and SpaceX would need to navigate these legal landscapes carefully to avoid delays or public backlash.

Despite these challenges, the momentum behind orbital AI data centers continues to build. Last week, Anthropic—another leading AI company—announced a partnership with SpaceX to utilize xAI's data centers in Memphis, Tennessee, with a focus on future space development. This suggests that the industry is coalescing around space as the next frontier for AI infrastructure, with Google, SpaceX, xAI, and Anthropic all investing heavily in the concept.

For now, the talks between Google and SpaceX remain in the early stages, and no official agreement has been signed. Both companies declined to comment on the report. Yet the convergence of their strategies—Google's Project Suncatcher, Musk's xAI acquisition, and SpaceX's constellation ambitions—paints a clear picture: the race to put AI in orbit is real, and the winners may well determine the future of computing.

Source: Mashable News