Google’s research initiative examines whether or not low Earth orbit may ultimately assist scalable machine studying infrastructure powered by photo voltaic power.
The experiment comes as Pope Leo XIV has intensified his warnings concerning the social and moral penalties of synthetic intelligence. Throughout his go to to France on September 25, the pope cautioned that AI may contribute to a future through which a “paradise of machines” undermines human company and each day life.
Google’s Mission Suncatcher will ship 4 Trillium TPUs into orbit aboard a SpaceX Transporter-18 mission with Planet to check their resilience to radiation, thermal extremes, and launch vibration. Supply: Google through X
The 2 developments tackle completely different sides of the identical quickly increasing know-how. Google is investigating find out how to improve the bodily capability accessible for AI computing, whereas Pope Leo is emphasizing how that rising functionality needs to be ruled and directed.
Google says Mission Suncatcher stays a analysis effort fairly than an operational space-based AI knowledge middle. The instant goal is to find out whether or not its {hardware} can stand up to the demanding situations of orbit and whether or not the engineering challenges can ultimately be overcome.
{Hardware} survival
Mission Suncatcher’s first orbital take a look at is being developed with satellite tv for pc firm Planet and is scheduled to fly on SpaceX’s Transporter-18 rideshare mission. The mission will accumulate knowledge on how Google’s TPUs reply to launch forces, radiation, and excessive thermal situations in area.
House-based AI infrastructure may harness near-constant daylight in low Earth orbit, producing as much as eight occasions extra solar energy for large-scale workloads. Supply: Google through X
Launch itself presents a major mechanical problem. Google says a spacecraft can expertise acceleration of as much as about 10 occasions Earth’s gravity throughout its journey to low Earth orbit, whereas particular person elements can encounter considerably increased forces. Engineers subsequently subjected the satellite tv for pc {hardware} to vibration testing throughout three axes to breed situations related to a rocket launch.
Radiation represents one other concern. Outdoors Earth’s environment, electronics are uncovered to photo voltaic occasions and cosmic rays that may trigger errors in computing programs. Google examined its Trillium TPUs on the UC Davis Crocker Nuclear Laboratory utilizing a proton beam whereas the chips have been operating AI workloads.
In accordance with Google, preliminary testing indicated that the Trillium TPUs may stand up to a complete ionizing radiation dose better than the quantity anticipated throughout a five-year area mission. The orbital take a look at is meant to find out how these outcomes translate into precise working situations.
Google launched Trillium as its sixth-generation TPU in 2024. The corporate says the chip delivers 4.7 occasions the height compute efficiency per chip of its earlier TPU v5e era and is greater than 67% extra power environment friendly.
Cooling in area
Working AI knowledge centres in orbit creates a separate engineering downside: eradicating warmth.
AI accelerators generate substantial quantities of warmth, however typical cooling programs usually depend upon transferring air or liquid via gear. Within the vacuum of area, there isn’t a environment to hold warmth away via strange airflow.
Google says its Mission Suncatcher workforce is testing warmth pipes and radiators to handle thermal masses. The corporate has already performed experiments inside a thermal vacuum chamber designed to breed the situations that electronics would expertise in orbit.
io.web argues decentralized compute can scale back centralized bottlenecks and management constraints, providing a sensible GPU-based different for AI workloads. Supply: @ionet through X
The cooling query is especially essential for any future try and scale space-based AI infrastructure. A constellation carrying bigger numbers of TPUs would wish to repeatedly handle the warmth generated by these processors whereas sustaining dependable efficiency.
Vitality availability is likely one of the fundamental causes Google is investigating the idea. The corporate says satellites in low Earth orbit can obtain near-constant daylight and probably generate as much as eight occasions extra solar energy than equal photo voltaic panels on Earth.
That doesn’t imply orbital AI knowledge centres are prepared to switch terrestrial amenities. Google describes Suncatcher as a long-term analysis challenge, with the present mission centered on gathering engineering knowledge fairly than deploying industrial computing capability.
Satellite tv for pc interconnectivity
Scaling AI computing past a single satellite tv for pc would require a number of spacecraft to function as a coordinated community. Google is subsequently additionally inspecting how satellites may trade massive volumes of knowledge whereas transferring quickly round Earth.
The corporate’s proposed structure would place a number of TPU chips on particular person satellites and join spacecraft utilizing optical laser hyperlinks. Google says future satellites would wish to find out their very own positions and people of neighboring spacecraft with extraordinarily excessive precision to keep up these connections.
The corporate plans to check this know-how throughout a two-satellite mission in 2027. The target is to research whether or not high-bandwidth, short-distance laser communications can present the connectivity required for distributed AI processing in orbit.
The idea builds on Google’s unique Mission Suncatcher announcement in November 2025, when the corporate described a possible community of solar-powered satellites outfitted with TPUs. At that stage, Google mentioned its analysis included satellite tv for pc constellation design, management, communications, and radiation testing.
The broader AI infrastructure market is concurrently increasing on Earth. Google has continued to put money into typical TPU capability as demand for machine studying computing grows. In Could 2026, Google and Blackstone introduced a deliberate TPU cloud enterprise with an preliminary $5 billion fairness dedication and an anticipated 500 megawatts of capability coming on-line in 2027.
That distinction highlights the experimental nature of Suncatcher: Google is increasing terrestrial AI infrastructure whereas individually inspecting whether or not some future computing workloads may function past the planet.
Only the start
Google’s Mission Suncatcher remains to be at an early analysis stage. The instant take a look at is just not designed to reveal a industrial space-based AI knowledge middle however to determine whether or not the underlying {hardware} and supporting programs can operate reliably in orbit.
The challenge additionally displays a broader problem dealing with the AI trade: increasing computing capability whereas coping with power, cooling, communications, and infrastructure necessities. Google’s method is to research whether or not entry to steady daylight and orbital infrastructure may ultimately present one other approach to scale machine studying programs.
Pope Leo known as for worldwide AI governance, warning {that a} “paradise of machines” may erode human moral judgment and urging better training in accountable know-how use. Supply: Reuters through X
Pope Leo’s issues give attention to a special dimension of that enlargement. In his September 25 tackle in France, he warned in opposition to shedding human company as AI turns into more and more built-in into each day life, saying there was an “pressing want for training in moral discernment.”
His place can be outlined in Magnifica Humanitas, his Could 2026 encyclical on safeguarding the human individual within the age of synthetic intelligence. The Vatican doc argues that AI ought to stay topic to human duty and oversight whereas recognizing its potential advantages. It additionally highlights the power and useful resource calls for related to more and more highly effective AI programs.
The Vatican has subsequently framed AI governance not merely as a technical situation however as a question involving human dignity, accountability, and the frequent good. Within the encyclical, Leo argues that technological progress ought to stay directed towards human welfare fairly than permitting effectivity or financial energy to develop into the only measure of progress.
Google’s area analysis and Pope Leo’s warnings don’t instantly tackle the identical downside. One issues the bodily limits of AI infrastructure; the opposite issues the social and moral limits of more and more highly effective know-how. Collectively, nevertheless, they illustrate how the following section of AI development is increasing past mannequin efficiency to questions on power, infrastructure, governance, and humanity’s position in directing the know-how.
Ahmed Ishtiaque Ahmed Ishtiaque Read More
