As the artificial intelligence buildout strains terrestrial infrastructure, a growing chorus of industry voices is looking to the skies. Orbital data centers could one day ease the burden, but significant technological advances are still needed before the concept becomes practical, according to industry experts who spoke with CNBC.
SpaceX CEO Elon Musk is aiming for a launch as early as late 2027, but several observers see meaningful scale arriving only in the next decade.
“It’s a next-decade event,” Evelyn Chow, portfolio manager at Neuberger, told CNBC’s “Squawk Box Asia” on Wednesday. “We’ll need to see significant satellite launches over the next four to five years, and a concomitant buildout of connectivity, before we can start to contemplate true scale in orbital data centers.”
Blaine Curcio, founder of Orbital Gateway Consulting, called the 2030s timeline a “fair assessment,” but he cautioned that SpaceX has a history of defying skeptics. “Had you asked any satellite industry expert in the late 2010s whether SpaceX could launch 10,000 satellites by 2025, every single one of them, myself included, would have said ‘no chance,'” he said.
Cooling in a vacuum
One of the most fundamental obstacles is cooling. Terrestrial data centers lean on liquid cooling systems to manage the intense heat generated by computing hardware. But in the vacuum of space, that approach doesn’t translate directly.
“Cooling is a huge issue,” Chow said. “It also needs to be tolerant to radiation.”
Without Earth’s atmosphere to dissipate heat, orbital facilities must rely on radiative cooling, a far less efficient process. And radiation exposure in orbit poses an ongoing threat to sensitive electronics, requiring shielding and hardened components that add weight and cost.
Fast-moving tech could make launches obsolete
Curcio also flagged the blistering pace of hardware development as a serious hurdle. Graphics processing units, or GPUs, are evolving so quickly that any data center launched today could be outdated within a couple of years.
“GPUs are evolving so quickly that if you launch a state-of-the-art data center into space, at an enormous cost, it might be obsolete in a couple of years,” he said.
That cost-benefit calculus is central to the debate. Launching payloads remains expensive, and if the hardware inside those payloads loses its edge quickly, the economics could collapse before the systems even begin operating at full capacity.
The problem of getting data back to Earth
Even if cooling and hardware challenges are solved, a more basic issue remains: communication. High-volume data generated in orbit must be transmitted back to Earth for AI systems to use it.
Transcelestial, a company specializing in laser-based communications, is working directly on this challenge. “Anyone can build data centers, but if you can’t talk to these AI systems, then those data centers are useless,” said Rohit Jha, CEO and cofounder of Transcelestial.
Laser links may offer a path to much higher bandwidth than traditional radio-frequency systems, but building a reliable, global network of ground stations and relay satellites is still a work in progress.
Jha added that orbital data centers could take another five to seven years to reach hyperscale, a stage that would likely require nuclear power to sustain the immense electricity demands of AI workloads.
Combined, these constraints paint a picture of an industry that is intellectually compelling but operationally far off. For investors and engineers alike, the near-term focus remains on building the satellite constellations and connectivity networks that could one day underpin a space-based computing ecosystem.
Source: www.cnbc.com — https://www.cnbc.com/2026/09/07/data-centers-space-elon-musk-obstacles.html
This article is for informational purposes only and does not constitute financial, investment, tax, or legal advice. Do your own research and consult a licensed professional before making financial decisions.
