Technology

Tech giants sketch timelines for data centres in orbit, Blue Origin files FCC plan for a 50,000-satellite constellation and Musk promises first AI satellite within a year

AWS chief calls the economics unlikely for a decade or two

Images

Data centers in space: How tech execs view the timeline Data centers in space: How tech execs view the timeline newsweek.com

Blue Origin has filed plans with the Federal Communications Commission for a constellation of more than 50,000 satellites to support computing in space, the most concrete entry yet in a year of orbital data centre announcements. Elon Musk said in July that SpaceX will launch its first AI satellite within a year and build a large-scale deployment by 2028; the company's finance chief, Bret Johnsen, says the first orbital compute satellite is on target for 2027.

The pressure originates on the ground. The Department of Energy estimates that data centres, driven by AI development, will account for up to 12 percent of US electricity consumption by 2028, and residents across the country and abroad are fighting proposed projects over water, electricity and utility bills. Accenture found that 80 percent of executives consider space-based technologies key to future revenue growth; the space economy could exceed 1.8 trillion dollars by 2035, with more than 60,000 satellites expected in orbit within four years.

SpaceX flew 165 orbital flights in 2025 — roughly 85 percent of all American launches, up from 25 in 2020 — and has asked the FCC for approval to launch up to one million solar-powered satellites. The launch record is the argument for the timelines; the counterargument comes from Amazon Web Services, whose chief, Matt Garman, told the Source podcast last week that it is "highly unlikely" orbital data centres will be economically feasible or practical in the next decade or two, citing rocket launches that remain "relatively infrequent and expensive."

The Government Accountability Office's Karen Howard puts truly large-scale facilities in the mid-2030s, citing power demand, cooling and data transfer systems that cannot add excessive size or launch weight, plus hefty federal investment in infrastructure and supply chains. Radiation and orbital traffic raise the bill further.

The operating economics do not transfer. A failed cooling fan on Earth is a service visit; in orbit it waits for the next launch window — the ones Garman calls infrequent and expensive. Small-scale processing chips on single satellites already exist; the large-scale version is the one nobody has priced.

Blue Origin has not announced a timeline for its constellation. Its filing promises "a new compute tier that operates independently of Earth-based constraints" — the constraints being, on current evidence, electricity, water and the communities objecting to both.