Elon Musk: In the future, superintelligent computing power will overwhelmingly be deployed in space
Elon Musk posted on X on November 2, 2026, saying that future superintelligent computing power will obviously be overwhelmingly deployed in space. By “overwhelmingly,” he meant that the ratio would directly round to 100%. He also added that anyone who does not understand this is a complete fool.
This statement appeared in his reply to a post by Sequoia partner Shaun Maguire discussing SpaceX’s computing costs and deployment speed. Musk had also posted on October 31 that ground-based computing would be “negligible” compared with orbital computing, and predicted that the monthly addition of orbital computing deployment would probably surpass ground-based deployment by the end of 2028 and would significantly exceed it in 2029.
SpaceX has renamed its AI business Super Intelligence and is advancing orbital computing satellite plans such as Starmind, aiming to bypass the power and cooling constraints of ground-based data centers by using near-continuous solar power and vacuum radiative cooling.
Original source:
Title: Future super intelligence compute will obviously be overwhelmingly in space. Publisher: Elon Musk (original X post) Publication date: 2026-11-02 URL: https://x.com/elonmusk/status/2117272748112154651 Source type: original post by the person involved Verification status: Full text of the original post and the surrounding reply context were read in this instance (verified on 2026-11-02) Supporting content: Supports all core facts in the body, including the explicit statement that “future superintelligent computing power will obviously be overwhelmingly deployed in space” and that it would “round to 100%”
ABAB AI Insight
Elon Musk has long tied SpaceX’s long-term growth narrative to AI computing power. After SpaceX announced the acquisition of xAI in February 2026, he publicly positioned orbital data centers as one of the company’s core directions and repeatedly emphasized the bottlenecks of ground-based power and cooling in later remarks. SpaceX has renamed its AI-related business from “Artificial Intelligence” to “Super Intelligence,” and its third-quarter report showed that revenue from this business rose 830% year over year to $7.56 billion, exceeding the combined revenue from its rocket and connectivity businesses.
This statement, that it would be “overwhelmingly” and even “rounded to 100%,” further strengthens his October 31 view that “ground-based computing will be negligible, orbital computing monthly deployment will probably surpass ground-based deployment by the end of 2028, and in 2029 it will significantly exceed it.” SpaceX is advancing the Starmind satellite constellation plan, aiming to deploy as many as 1 million computing nodes, with each satellite carrying about 150 kW of peak computing payload, relying on solar power and space radiative cooling. Musk had also proposed launching computing satellites into deep space from the Moon using electromagnetic mass drivers, claiming a scale of hundreds to thousands of TW per year.
Compared with ground-based hyperscale data centers such as xAI’s Colossus, the core difference in the orbital approach lies in energy access and cooling methods: on the ground, it is constrained by grid capacity, land, and water cooling, while in orbit it can obtain near-continuous solar power and dissipate heat directly into a vacuum through radiation. Musk sees orbital computing as a path around these constraints and predicts that its cost will eventually fall below that on the ground. But this path still depends on conditions that have not yet been fully verified, including Starship’s high-frequency launch capability, the scale of satellite manufacturing, and orbital traffic management.
This statement further shifts SpaceX’s valuation narrative from rockets and Starlink toward being the dominant player in the orbital deployment of superintelligent computing power. It reinforces the company’s differentiated position in the AI infrastructure race, but it also concentrates substantial capital expenditure and launch-pace execution risk within the same time window. Indicators to watch include whether SpaceX subsequently discloses specific launch timelines for Starmind or AI satellites, and whether the actual deployed orbital computing power in 2027–2028 comes close to its previously stated 13–20 GW range.
ABAB News · Cognitive Law
Musk’s use of “overwhelmingly” as rounding to 100% is a further reinforcement of orbital computing’s dominant position.
Ground-based computing is constrained by power and cooling, while the orbital approach bypasses these limits with continuous solar power and vacuum heat dissipation.
The move from “monthly deployment surpassing ground-based deployment by the end of 2028” to “rounding to 100%” is a progressive expression of the same narrative.