AWS CEO Matt Garman Pours Cold Water on Orbital AI Data Center Hype

The vision of solving the world’s insatiable appetite for artificial intelligence computing power by launching data centers into orbit is facing a reality check from one of the industry’s most prominent figures. Matt Garman, the CEO of Amazon Web Services (AWS), recently expressed strong skepticism regarding the near-term practicality of space-based infrastructure, suggesting that such projects are unlikely to be economically feasible for at least the next ten to twenty years.

The Economic and Engineering Reality of Space

While the concept of utilizing orbital space to host massive server farms has gained traction among some of the world’s wealthiest tech leaders, Garman highlighted the stark contrast between current terrestrial capabilities and the requirements of space deployment. Speaking on the Sources Podcast, Garman pointed out that while the idea of tapping into free solar energy in orbit sounds appealing, the logistical costs of rocket launches remain a prohibitive factor.

Garman emphasized the sheer scale of modern cloud operations to illustrate the gap. “We as a world don’t do one launch a day today,” he noted, contrasting this with the rapid pace of terrestrial expansion where Amazon is currently installing one GPU every 14 seconds. Beyond the frequency and cost of launches, Garman identified significant technical barriers, specifically the difficulty of heat dissipation in the vacuum of space and the constant exposure to radiation, which would necessitate specialized, expensive hardware construction.

Differing Visions Among Tech Billionaires

The skepticism from the AWS chief stands in contrast to the ambitious goals set by other industry titans. Elon Musk, the CEO of SpaceX and Tesla, has been a vocal proponent of orbital data centers, viewing them as a necessary evolution to meet the soaring energy demands of AI. Earlier this year, SpaceX revealed plans to develop a massive “constellation of a million satellites” designed to function as orbital data centers, even initiating a hiring drive for engineers to bring the project to life. In January, Musk suggested that space could become the most cost-effective location for data centers within a three-year window.

Jeff Bezos, the founder of Amazon and Blue Origin, has also expressed interest in the concept. During a CNBC interview in May, Bezos described the prospect of space-based data centers as “very realistic,” though he cautioned that the timelines proposed by some advocates—often cited as two or three years—were likely overly ambitious.

The Search for Alternative Frontiers

Garman’s assessment of unconventional data center locations extended beyond space. He also cast doubt on the practicality of underwater data centers, noting that they face many of the same environmental and maintenance challenges as orbital facilities. However, he did identify a more terrestrial alternative that he finds plausible: deserts.

According to Garman, desert environments offer a more realistic frontier for future data center expansion. He noted that these regions provide “relatively good environmental areas,” access to heat sinks, and vast space for solar energy harvesting. Despite the potential for such remote locations, Garman cautioned that moving infrastructure away from populated areas could have negative economic consequences, as local communities would lose out on the tax revenue and high-paying employment opportunities that typically accompany large-scale data center developments.

While Garman acknowledged that the engineering challenges of space-based computing are “technically solvable,” he maintained that the industry is far from seeing these facilities become a standard part of the global tech stack. For now, the focus remains on terrestrial solutions, with Garman also defending the water-usage efficiency of Amazon’s current facilities, noting that their largest sites consume less water than 40 typical US households.

Source: businessinsider.com