Google’s Space AI Data Centres Raise New Questions as Pope Leo Warns of a Threat to Humanity

Google’s research initiative examines whether low Earth orbit could eventually support scalable machine learning infrastructure powered by solar energy.

The experiment comes as Pope Leo XIV has intensified his warnings about the social and ethical consequences of artificial intelligence. During his visit to France on September 25, the pope cautioned that AI could contribute to a future in which a “paradise of machines” undermines human agency and daily life.

Google’s Project Suncatcher will send four Trillium TPUs into orbit aboard a SpaceX Transporter-18 mission with Planet to test their resilience to radiation, thermal extremes, and launch vibration. Source: Google via X

The two developments address different sides of the same rapidly expanding technology. Google is investigating how to increase the physical capacity available for AI computing, while Pope Leo is emphasizing how that growing capability should be governed and directed.

Google says Project Suncatcher remains a research effort rather than an operational space-based AI data center. The immediate objective is to determine whether its hardware can withstand the demanding conditions of orbit and whether the engineering challenges can eventually be overcome.

Hardware survival

Project Suncatcher’s first orbital test is being developed with satellite company Planet and is scheduled to fly on SpaceX’s Transporter-18 rideshare mission. The mission will collect data on how Google’s TPUs respond to launch forces, radiation, and extreme thermal conditions in space.

Space-based AI infrastructure could harness near-constant sunlight in low Earth orbit, generating up to eight times more solar power for large-scale workloads. Source: Google via X

Launch itself presents a significant mechanical challenge. Google says a spacecraft can experience acceleration of up to about 10 times Earth’s gravity during its journey to low Earth orbit, while individual components can encounter substantially higher forces. Engineers therefore subjected the satellite hardware to vibration testing across three axes to reproduce conditions associated with a rocket launch.

Radiation represents another concern. Outside Earth’s atmosphere, electronics are exposed to solar events and cosmic rays that can cause errors in computing systems. Google tested its Trillium TPUs at the UC Davis Crocker Nuclear Laboratory using a proton beam while the chips were running AI workloads.

According to Google, initial testing indicated that the Trillium TPUs could withstand a total ionizing radiation dose greater than the amount expected during a five-year space mission. The orbital test is intended to determine how those results translate into actual operating conditions.

Google introduced Trillium as its sixth-generation TPU in 2024. The company says the chip delivers 4.7 times the peak compute performance per chip of its previous TPU v5e generation and is more than 67% more energy efficient.

Cooling in space

Operating AI data centres in orbit creates a separate engineering problem: removing heat.

AI accelerators generate substantial amounts of heat, but conventional cooling systems often depend on moving air or liquid through equipment. In the vacuum of space, there is no atmosphere to carry heat away through ordinary airflow.

Google says its Project Suncatcher team is testing heat pipes and radiators to manage thermal loads. The company has already conducted experiments inside a thermal vacuum chamber designed to reproduce the conditions that electronics would experience in orbit.

io.net argues decentralized compute can reduce centralized bottlenecks and control constraints, offering a practical GPU-based alternative for AI workloads. Source: @ionet via X

The cooling question is particularly important for any future attempt to scale space-based AI infrastructure. A constellation carrying larger numbers of TPUs would need to continuously manage the heat generated by those processors while maintaining reliable performance.

Energy availability is one of the main reasons Google is investigating the concept. The company says satellites in low Earth orbit can receive near-constant sunlight and potentially generate up to eight times more solar power than equivalent solar panels on Earth.

That does not mean orbital AI data centres are ready to replace terrestrial facilities. Google describes Suncatcher as a long-term research project, with the current mission focused on gathering engineering data rather than deploying commercial computing capacity.

Satellite interconnectivity

Scaling AI computing beyond a single satellite would require multiple spacecraft to operate as a coordinated network. Google is therefore also examining how satellites could exchange large volumes of data while moving rapidly around Earth.

The company’s proposed architecture would place multiple TPU chips on individual satellites and connect spacecraft using optical laser links. Google says future satellites would need to determine their own positions and those of neighboring spacecraft with extremely high precision to maintain these connections.

The company plans to test this technology during a two-satellite mission in 2027. The objective is to investigate whether high-bandwidth, short-distance laser communications can provide the connectivity required for distributed AI processing in orbit.

The concept builds on Google’s original Project Suncatcher announcement in November 2025, when the company described a potential network of solar-powered satellites equipped with TPUs. At that stage, Google said its research included satellite constellation design, control, communications, and radiation testing.

The broader AI infrastructure market is simultaneously expanding on Earth. Google has continued to invest in conventional TPU capacity as demand for machine learning computing grows. In May 2026, Google and Blackstone announced a planned TPU cloud venture with an initial $5 billion equity commitment and an expected 500 megawatts of capacity coming online in 2027.

That contrast highlights the experimental nature of Suncatcher: Google is expanding terrestrial AI infrastructure while separately examining whether some future computing workloads could operate beyond the planet.

Just the beginning

Google’s Project Suncatcher is still at an early research stage. The immediate test is not designed to demonstrate a commercial space-based AI data center but to establish whether the underlying hardware and supporting systems can function reliably in orbit.

The project also reflects a broader challenge facing the AI industry: expanding computing capacity while dealing with energy, cooling, communications, and infrastructure requirements. Google’s approach is to investigate whether access to continuous sunlight and orbital infrastructure could eventually provide another way to scale machine learning systems.

Pope Leo called for international AI governance, warning that a “paradise of machines” could erode human ethical judgment and urging greater education in responsible technology use. Source: Reuters via X

Pope Leo’s concerns focus on a different dimension of that expansion. In his September 25 address in France, he warned against losing human agency as AI becomes increasingly integrated into daily life, saying there was an “urgent need for education in ethical discernment.”

His position is also outlined in Magnifica Humanitas, his May 2026 encyclical on safeguarding the human person in the age of artificial intelligence. The Vatican document argues that AI should remain subject to human responsibility and oversight while recognizing its potential benefits. It also highlights the energy and resource demands associated with increasingly powerful AI systems.

The Vatican has therefore framed AI governance not simply as a technical issue but as a question involving human dignity, accountability, and the common good. In the encyclical, Leo argues that technological progress should remain directed toward human welfare rather than allowing efficiency or economic power to become the sole measure of progress.

Google’s space research and Pope Leo’s warnings do not directly address the same problem. One concerns the physical limits of AI infrastructure; the other concerns the social and ethical limits of increasingly powerful technology. Together, however, they illustrate how the next phase of AI development is expanding beyond model performance to questions about energy, infrastructure, governance, and humanity’s role in directing the technology.