Chris Young’s Beyond Earth column explores the intersection of space technology and policy, providing thought-provoking commentary on the latest advancements and regulatory developments in the sector.
Starlink started with a promise to connect the world. Now Elon Musk wants it to power AI, and cool the planet.
The world’s richest man has announced two potential new use cases for his company’s orbital network. Firstly, Starlink could be used to meet the surging, unsustainable demand for data centers on Earth. “Simply scaling up Starlink V3 satellites, which have high-speed laser links, would work,” Musk wrote on social media platform X. “SpaceX will be doing this.”
More ambitiously, Musk also wrote that an AI satellite constellation could be used to cool our planet and tackle global warming.
Both ideas address real-world crises — unsustainable energy demand and runaway global heating — but they come with staggering technical, environmental, and ethical implications.
And both, conveniently, justify sending thousands more satellites into low Earth orbit, regardless of the risks to science, safety, or sustainability.
SpaceX already dominates the satellite launch sector. It is responsible for roughly 70 percent of the approximately 12,500 active satellites in orbit today.
The AI boom, led by OpenAI’s public launch of ChatGPT in 2022, has created a surging power demand. Estimates suggest this demand – driven by the data centers used to power and train AI – will continue to skyrocket for the foreseeable future. A Goldman Sachs report from earlier this year, for example, estimates that AI-driven energy demand could rise 165 percent by 2030.
Those energy requirements are completely unsustainable, especially in the context of the ongoing climate crisis and nations failing to meet emissions reduction targets. Now, Musk aims to tackle this problem with SpaceX’s satellite infrastructure. The tech mogul tends to target the big problems with his companies: Tesla’s mission is to reduce our reliance on fossil fuels; SpaceX’s is to make humanity a multiplanetary species; Neuralink’s is to mitigate the existential threat of AI, etc.
Musk isn’t the first to suggest data centers could operate in space. He isn’t even the first tech baron to float the idea. Earlier this year, former Google CEO Eric Schmidt told Congress that energy demand will increase from three percent to 99 percent of total generation, due to the rising use of AI in data centers. Schmidt also claimed he acquired space startup Relativity Space – and placed himself as CEO – to build data centers in space.
The jury is out regarding the feasibility of sending data centers to orbit. In an interview with Interesting Engineering, University of Regina astronomer Samantha Lawler said, “This is stupid, for many reasons. How does putting data centers in space help anything at all? It’s my understanding that data centers require a lot of swapping components and fixing things and updating regularly.”
“All of that would be much harder in space,” she continued. “Launching that much material into space is also hard. And what happens if or when it gets damaged by space debris?”
However, a team of researchers from NTU Singapore recently claimed space data centers could be achieved by leveraging existing launch and satellite infrastructure. These could use AI accelerators to process raw data in orbit. Arguably, though, feasibility doesn’t equate to sustainability. Instead, it kicks the can down the road – by addressing the problem of unsustainable power demand on Earth, it contributes to the overcongestion of space.
Never one to shy away from controversy, Musk has now also joined one of the most controversial scientific debates of our time: In the face of rising global temperatures and a grindingly slow transition away from fossil fuel reliance, should governments be looking to cool the planet through solar geoengineering?
According to Musk, dimming the Sun is perfectly feasible using satellite technologies available today. “A large solar-powered AI satellite constellation would be able to prevent global warming by making tiny adjustments in how much solar energy reached Earth,” Musk posted on X on November 3.
Earlier this month, the United Nations once again warned that we are not on track to meet the carbon emissions reduction goals set by the Paris Agreement in 2015. Some scientists claim we must consider alternative, extreme measures to cool our planet, such as solar geoengineering.
However, a recent paper by a team at Columbia University warns that this method could have a dramatic backfire. “There are a range of things that might happen if you try to do this – and we’re arguing that the range of possible outcomes is a lot wider than anybody has appreciated until now,” V. Faye McNeill, an atmospheric chemist and aerosol scientist at Columbia’s Climate School and Columbia Engineering, explained in a press release.
Lawler was equally sceptical of Musk’s Mr. Burns plans. “We know exactly what we need to do to “prevent global warming” which is stop burning fossil fuels,” she told IE.
“Blocking sunlight while continuing to burn fossil fuels and raise carbon dioxide levels in the atmosphere is not a solution, it’s a temporary fix,” Lawler continued. “And if there are enough satellites in orbit to block a significant fraction of sunlight, we’ll have serious collision risk problems.”
Both of Musk’s plans lead to the same scenario, according to Lawler. Namely, more congestion in an already overcrowded space.
When IE first spoke to Lawler in 2022, the University of Regina astronomer warned that Starlink satellites were setting us on the path to Kessler Syndrome – a cascading effect whereby every satellite collision creates debris that increases the risk of more collisions. As space debris would be flying at several kilometers per second, a clean-up operation would be akin to “collecting bullets,” Lawler said at the time.
Since then, things are “much worse”, Lawler explained to IE in 2025. “There are thousands more satellites in orbit than there were in 2022. My research data has more satellite streaks in it, and I see way more satellites in my dark skies. The reentry rate has also climbed considerably. Starlink is designed with only a 5-year lifetime, which means they must be replaced at a stunning rate.”
According to the astronomer, SpaceX is averaging between one and two Starlink reentries per day. Between November 2024 and May 2025, they burned up almost 500 Starlink satellites, a SpaceX report points out.
“They claimed in this report that they would try to have Starlinks reenter over the Pacific Ocean, just in case they’re not burning up completely, but I have seen absolutely no evidence of that,” Lawler said. “Starlink reentries have been widely reported over many inhabited locations.”
There is a growing risk of SpaceX debris hitting private property. In 2024, a chunk of a Starlink satellite fell over farmland in Saskatchewan, Canada. In 2022, debris from SpaceX’s Crew-1 capsule fell onto a sheep field in southern New South Wales, Australia.
As Lawler points out, “There are no robust safety testing requirements for satellite reentries. Of course, if they’re burning up completely, that’s a huge amount of weird metal and plastic that’s being added to the upper atmosphere, and that’s actually a lot scarier than a few pieces hitting the ground, because we don’t actually know how that will affect atmospheric chemistry.”
Since 2022, the IAU Centre for the Protection of the Dark and Quiet Sky from Satellite Constellation Interference (CPS) has campaigned for stronger regulations regarding satellite operations. IE also reached out to members of the IAU CPS for this article, but they were unable to secure approval to reply due to the ongoing US government shutdown.
“Interestingly, many satellite builders and operators that I’ve spoken with also want stronger regulations in orbit,” Lawler told IE. “I know there are discussions happening at the UN level, but they are painfully slow, and meanwhile, Starlink is launching batches of 60 more satellites into orbit every few days.”
“There are reports that the Chinese Space Station has been damaged by a debris strike, and a collision in Starlink’s orbit could increase debris so much that human spaceflight is no longer an option,” she continued. “We desperately need regulation of satellites in orbit: without regulation, we’ll hit the safe operating limit soon enough.”
This is corroborated by a recent SpaceX filing with the FCC. The filing revealed that Starlink satellites required 50,000 collision avoidance maneuvers between 2019 and 2023. According to Hugh Lewis, a professor of astronautics at the University of Southampton in the UK, if those trends continue, Starlink satellites will have to perform roughly a million maneuvers every six months by 2028.
SpaceX’s own reports show they performed a collision avoidance maneuver every 2 minutes on average between November 2024 and May 2025. “That is terrifying,” Lawler exclaimed. “Every two minutes, there is a chance to make a mistake that could put us into Kessler Syndrome, the runaway collisional cascade that could destroy a large fraction of satellites in orbit and make it difficult for us to use low Earth orbit for decades to centuries.”
If Starlink is going to evolve from broadband service to orbital climate infrastructure, regulators can’t afford to keep looking the other way.
Musk’s habit of launching first and explaining later has already redrawn the rules of commercial space. But as Starlink shifts from internet access to AI acceleration and solar geoengineering, the consequences are no longer theoretical.