Elon Musk has put a number on how big SpaceX wants its compute ambitions to get, and it is enormous. In a reply on X on 11 November, he wrote that SpaceX is “aiming for 1TW/year from Earth, rising to over 1PW/year from the Moon.”1 In follow-up posts the same day, he explained why he counts power in space at a premium, and he gave a nearer-term marker: a chance of about half a gigawatt of “actually usable” compute next year.23
This is the first ENN article in a new thread tracking SpaceX’s push to build AI compute at a scale far beyond today’s data centers, including in orbit and, one day, from the Moon.
What Musk posted
The headline figures came in a reply to the account @IterIntellectus. Musk wrote that SpaceX is aiming for 1TW a year from Earth, rising to over 1PW a year from the Moon.1 For scale, a petawatt is a thousand terawatts, so the lunar figure sits in a different league from the Earth one. He then tied the goal to SpaceX’s value:1
That’s why I say that if we achieve this outcome, SpaceX will obviously be worth more than Earth, because space is so staggeringly enormous!
The framing is conditional. Musk says “if we achieve this outcome,” which marks this as a stated aim, not a delivered result.1 The ambition is exactly what ENN readers have come to expect from him: set the goal at civilizational scale, then build toward it.
@IterIntellectus Actually, we’re aiming for 1TW/year from Earth, rising to over 1PW/year from the Moon. That’s why I say that if we achieve this outcome, SpaceX will obviously be worth more than Earth, because space is so staggeringly enormous!
— @elonmusk Nov 11, 2026
Why he says 100GW in space beats 100GW on Earth
In a second reply, to @TheFlintFace and @IterIntellectus, Musk made the case for putting compute in space rather than simply building more on the ground. “100GW in space is roughly equal to 800GW on Earth,” he wrote.2 By that estimate, every gigawatt in orbit is worth about eight on the ground.
His reasoning is about how solar power behaves on Earth. Solar on Earth, he wrote, “must be paired with massive batteries to continue powering compute at night and supplemented with combustion or nuclear power during winter.”2 He closed the point bluntly: “These are not the same at all.”2
That argument matches what supporters of space-based data centers have long highlighted. Coverage of the field describes the advantages proponents cite as free, limitless power from the Sun and none of the environmental costs of building these facilities on Earth, where opposition is starting to grow.4
@TheFlintFace @IterIntellectus 100GW in space is roughly equal to 800GW on Earth. Moreover, solar on Earth must be paired with massive batteries to continue powering compute at night and supplemented with combustion or nuclear power during winter. These are not the same at all.
— @elonmusk Nov 11, 2026
The near-term number: about half a gigawatt
The third post brought the vision down to next year. Replying to @zerohedge, Musk wrote that “SpaceX has a chance of going about half the GW next year of actually usable compute.”3 He did not say in that post where that compute would sit.
He leaned hard on the phrase “actually usable,” calling it “an important metric.”3 He explained why: “It takes us ~2 months working 24/7 to get a large compute cluster working reliably. I’d guess closer to 6 months for others.”3 In other words, Musk is measuring progress by compute that is running reliably, not by hardware sitting in a building, and he says his team brings clusters online roughly three times faster than he expects others to.
ENN has also covered SpaceX’s plans on the ground, including the $24B Starbase supercomputer plan.
@zerohedge SpaceX has a chance of going about half the GW next year of actually usable compute. “Actually usable” is an important metric. It takes us ~2 months working 24/7 to get a large compute cluster working reliably. I’d guess closer to 6 months for others.
— @elonmusk Nov 11, 2026

How this fits the vision he has described before
These posts build on ideas Musk has laid out over time. In earlier coverage originally from Ars Technica, after the outlet wrote about using autonomous assembly to build large data centers in space, Musk replied on X that Starlink satellites could do the job: “Simply scaling up Starlink V3 satellites, which have high speed laser links would work,” he said. “SpaceX will be doing this.”4
In a later post, made in a conversation with Cathie Wood of Ark Invest and quoted on the same forum thread, Musk sketched the path in more detail. He wrote that satellites with localized AI compute, beaming results back from low-latency, sun-synchronous orbit, “will be the lowest cost way to generate AI bitstreams in <3 years."4 He added that "1 megaton/year of satellites with 100kW per satellite yields 100GW of AI added per year," connecting via high-bandwidth lasers to the Starlink constellation.4
Then he went further, to the Moon. “The level beyond that is constructing satellite factories on the Moon and using a mass driver (electromagnetic railgun) to accelerate AI satellites to lunar escape velocity without the need for rockets,” he wrote, adding that this “scales to >100TW/year of AI and enables non-trivial progress towards becoming a Kardashev II civilization.”4
Set side by side, the 11 November figures are larger than those earlier ones: 1TW a year from Earth against 100GW a year, and over 1PW a year from the Moon against more than 100TW a year.14 The direction is the same, with the targets pushed higher.

Why SpaceX is the company to try
The coverage reposted on the forum makes the case for taking SpaceX seriously here. It describes SpaceX as owning and operating significantly more space-based infrastructure than any other company or country, and says Musk’s interest “significantly raises the profile of the nascent industry.”4 It also notes that Starlink has already defied conventional wisdom by delivering high-speed broadband to millions of customers worldwide while making a profit, so if Musk believes the Starlink architecture can be applied to data centers, “it will be difficult for the industry to ignore.”4
Later reporting by Eric Berger, quoted in the same thread, said that in the near term SpaceX plans to develop a modified version of the Starlink satellite to serve as a foundation for building data centers in space.4 Berger also quoted Abhi Tripathi, a long-time SpaceX employee now at the UC Berkeley Space Sciences Laboratory, who believes that once Musk realized Starlink satellites could be architected into a distributed network of data centers, “the writing was on the wall.”4
The idea is drawing other big names too. The same coverage notes that former Google chief executive Eric Schmidt acquired Relativity Space because of his interest in space-based data centers, and that Amazon founder Jeff Bezos predicted gigawatt-scale data centers will be built in space within the next 10 to 20 years.4
The skeptics’ case
Not everyone is convinced. Critics, as summarized in the coverage, say it is economically impractical to build these facilities in space and that supporters underestimate the technology needed to make it work.4 Musk’s answer, in his 11 November posts, is about energy: he argues that ground-based solar needs massive batteries and backup generation, so space power is worth roughly eight times as much per gigawatt.2 His emphasis on “actually usable” compute also shows he is measuring the effort by working output, not announcements.3
What to watch next
- The half-gigawatt marker. Musk says SpaceX has a chance of about half a GW of actually usable compute next year. That is the clearest near-term test of these posts.3
- Bring-up speed. His claim of about two months of 24/7 work to get a large cluster running reliably is a benchmark to watch as new clusters come online.3
- Starlink-based compute satellites. Musk has said scaled-up Starlink V3 satellites with laser links could serve as orbiting data centers, and that “SpaceX will be doing this.”4
- The Moon. The over-1PW-a-year goal depends on the lunar step Musk has described: satellite factories on the Moon and a mass driver launching AI satellites without rockets.14
For now, these are Musk’s stated goals and estimates, posted in his own words. They are also a clear statement of where he wants SpaceX to go, and of his conviction that the best place to put the next generation of compute is up where the Sun never sets.



