Launching materials to orbit still costs roughly $1,000/kg — over $900,000 per ton. Alexander Wyglinkski (WPI) notes the expense of moving all components into orbit, with assembly and maintenance costs on top, makes the economics "extremely challenging" . The European Parliament's research service flags launch costs as the single biggest barrier
.
AI GPUs and accelerators become obsolete in 2–3 years. On Earth, racks are swapped and boards replaced continuously. In orbit, every upgrade requires a launch, docking, or robotic servicing — a logistical nightmare that critics say makes orbital AI compute impractical for cutting-edge workloads .
AI inference often requires real-time responses. Orbital data centers introduce signal travel time (latency) that makes them unsuitable for many mainstream AI applications. Analysts at Stanford Tech Review and New Space Economy conclude orbit is credible only for specialized, latency-tolerant workloads, not as a general-purpose replacement for terrestrial compute .
Former NASA officials have called the plan "ridiculous," warning that a million-satellite constellation would dramatically increase collision risk and exacerbate the orbital debris problem . Every collision creates more debris, potentially triggering a cascading Kessler syndrome that could render low Earth orbit unusable.
OpenAI's Sam Altman called the idea "ridiculous" in its current form. A Gartner report described the excitement around orbital data centers as "peak insanity" and a "bubble," adding that practical applications won't arrive "for decades, if ever" .
Space radiation causes random bit flips in electronics accumulates over time, degrading performance. This is a well-known problem for satellite computing that requires expensive hardening and limits the useful lifespan of orbital hardware .
Launching a million satellites would produce enormous carbon emissions and atmospheric pollution from rocket exhaust. Critics argue the environmental cost of rocketing compute hardware into orbit likely outweighs any terrestrial energy savings .
Bottom line: SpaceX has unveiled detailed satellite designs and filed with the FCC for a million-satellite orbital compute network, positioning it as the growth story for the upcoming IPO. The technology relies on continuous solar power, passive cooling, and Starship's low launch costs. Skeptics point to extreme cost, hardware obsolescence, latency, debris risk, and broad industry dismissal as reasons the project may remain a long shot — or a speculative bubble — for years to come.
Comments
0 comments