Elon Musk’s post at 06:12Z on 4 November, “Data centers in space looking pretty appealing right now”, had 1,515,108 views by 07:05Z. SpaceX’s filings describe such a plan: on 30 January it asked the US Federal Communications Commission for authority to launch and operate up to one million satellites as the “SpaceX Orbital Data Center system”.
That is about 90 times the more than 11,100 Starlink satellites SpaceX says it has in low-Earth orbit, by our arithmetic. Its IPO prospectus says it expects to begin deploying orbital AI compute satellites “as early as 2028”.
Business Pill · Lead Time
A short Business Engineer explainer on lead times: why big infrastructure takes years from plan to operation. It teaches the general idea only and does not discuss SpaceX, the FCC or space data centers.
The key insight: Musk’s post gives no reason, but SpaceX’s own filings set out an orbital data center plan: an FCC application for up to one million satellites, about 90 times its Starlink fleet by our arithmetic, and a prospectus that says the first orbital AI compute satellites could deploy “as early as 2028”.
The FCC’s Space Bureau accepted the application for filing on 4 February and set comments due on 6 March. Its notice says the system “will operate at altitudes ranging from 500 km to 2,000 km”, in 30-degree and sun-synchronous orbits, and will “primarily rely on optical intersatellite-links” that may connect with SpaceX’s Starlink satellites.
The notice quotes SpaceX calling the system the “first step towards becoming a Kardashev II-level civilization – one that can harness the Sun’s full power”. SpaceX asked for waivers including one “to exempt the System from the NGSO milestone requirements and deployment obligations” and one from the surety bond requirements.
What SpaceX has filed on orbital data centres and the targets those filings state: the FCC application of 30 January 2026 for up to one million satellites (FCC Public Notice DA 26-113), the 12 June prospectus goal of 100 GW of compute to orbit a year, the 10-Q for the quarter to 30 September (1.7 GW of nameplate compute on the ground, three Starship launches in nine months), and the prospectus’s ‘as early as 2028’ and end-of-decade targets. The ~90x is our arithmetic: 1,000,000 / 11,100.
What Was Posted on X
@SPCX100T Terrestrial compute will be peanuts compared to orbital compute.
Orbital compute deployed per month probably exceeds terrestrial around the end of 2028 and then far exceeds it in 2029.
Elon Musk’s post of 31 October, a reply to another account, with his timing for orbital compute: deployed per month, it ‘probably exceeds terrestrial around the end of 2028’. It is a prediction worded with ‘probably’; SpaceX’s prospectus gives ‘as early as 2028’ for its first orbital AI compute satellites.
SemiAnalysis’s post of 3 November, headlined ‘SpaceX Will Be The World’s Biggest AI Cloud Making 13GW in 1.5 Years A Reality’, which appeared in Musk’s profile timeline. It concerns SpaceX’s AI capacity; we did not read the linked newsletter, and SpaceX’s 10-Q reports 1.7 GW of nameplate compute at 30 September.
SpaceX completing the build of 13GW from here through 2027 “would be the largest amount of datacenter capacity ever built by any one company in that timeframe, roughly equivalent to all of Google’s datacenter capacity from 1999-2026 in just 1.5 years.”
Trung Phan relays a line on SpaceX completing 13GW through 2027: it “would be the largest amount of datacenter capacity ever built by any one company in that timeframe, roughly equivalent to all of Google’s datacenter capacity from 1999-2026 in just 1.5 years.” The comparison is the quoted source’s claim, not a SpaceX figure.
Building the equivalent of 17 years of capacity in 18 months.
Sunny Madra (@sundeep), quoting Trung Phan’s post on SpaceX’s 13GW build, writes: “Building the equivalent of 17 years of capacity in 18 months. All on @nvidia gear!” The 17-year framing and the Nvidia claim are his.
What the Prospectus Says
SpaceX’s final IPO prospectus, filed on 12 June, says: “Our goal over time is to launch 100 gigawatts of compute to space each year.” It says that “will require thousands of launches per year and the transport of approximately one million metric tons to orbit annually”.
At the 100 metric tons Starship V3 is designed to carry, that is about 10,000 launches a year, by our arithmetic. The prospectus also names a lower bar: about 10 gigawatts a year, “which we believe would still enable a commercially attractive AI compute business with strong economic returns”.
On timing, it says that “By the end of the decade, we intend to deploy the first modular orbital AI compute shells and begin monetizing capacity through the sale of AI software and AI compute”. Its risk factors say a constellation “potentially numbering up to one million satellites” will need approvals, and that “there can be no assurance that such approvals will be obtained on acceptable timelines, terms, or at all”.
The Condition in Musk’s Award
The prospectus describes Musk’s “AI CEO Award”, granted in March 2026 in place of his xAI Award. Its 302.1 million performance-based restricted shares vest on market value milestones from $1.065 trillion to $6.565 trillion and on the “completion of non-Earth-based data centers capable of delivering 100 terawatts of compute per year”.
One hundred terawatts is 1,000 times the 100-gigawatt yearly goal, by our arithmetic.
What the Latest 10-Q Shows
SpaceX’s 10-Q for the quarter to 30 September reports nameplate compute of 1.7 GW, installed on the ground, against 0.6 GW a year earlier. The 100-gigawatt yearly goal is about 59 times that installed base, by our arithmetic; one is a yearly target, the other installed capacity.
The filing reports three Starship launches in the nine months to 30 September, and says “To date, all Starship launches have been classified as internal”. It says SpaceX allocates a significant amount of launch capacity to Connectivity and expects “to allocate a significant amount to our SI segment in the future”. Our search of the 10-Q text found three uses of “orbital”, all about launches.
What Rivals Have Filed
Blue Origin filed with the FCC on 19 March for up to 51,600 satellites under the name Project Sunrise, Advanced Television reported. SpaceX’s one million is about 19 times that, by our arithmetic. Google says its Project Suncatcher prototype, launched on 1 October, sent the first AI answer from a data center class TPU in orbit, and names twin interconnected satellites in 2027 as its next step, as we reported.
Update, 4 Nov: an Nvidia executive amplifies the 13-gigawatt line. At 23:34 UTC on 3 November, Sunny Madra, whose X profile describes him as a VP at Nvidia and previously COO and President of Groq, posted: “Building the equivalent of 17 years of capacity in 18 months. All on @nvidia gear!” He was quoting a post by Trung Phan that relays a line from a SemiAnalysis subscriber report of 3 November: SpaceX completing 13GW through 2027 “would be the largest amount of datacenter capacity ever built by any one company in that timeframe, roughly equivalent to all of Google’s datacenter capacity from 1999-2026 in just 1.5 years” (SemiAnalysis). Both posts are embedded in What Was Posted on X above.
Where the numbers come from. The 13 gigawatts is Musk’s own: on SpaceX’s 20 October webcast he said “I would be surprised if we brought on less than 13 GW of compute by the end of next year”, an “incremental 10 GW of compute online in 2027” on top of “roughly 3 GW maybe 2.5 to 3 GW” by the end of this year, with next year “primarily Rubin, Vera Rubin compute” (SpaceX’s edited transcript). The Google comparison is SemiAnalysis’s; the free section we read does not show the Google figure behind it, and we have not verified it. The “17 years” is Madra’s wording; 1999 to 2026 is 27 years by our count. We reported SemiAnalysis’s 3.6 gigawatts under construction and >9 gigawatts identified here.
Against the filing, by our arithmetic: 13 gigawatts by the end of 2027 is about 7.6 times the 1.7 gigawatts of nameplate compute SpaceX’s 10-Q reports at 30 September. The measures differ: Musk spoke of compute brought online, the filing reports nameplate compute installed on the ground.
As we read the three filings together, the FCC application sets the ceiling in satellites, the prospectus sets the goal in gigawatts and tons, and the 10-Q reports where SpaceX stands today: 1.7 GW installed on the ground and three Starship launches in nine months.
As we read the prospectus, the measure that links them is launch: it says 100 gigawatts a year needs about one million metric tons to orbit annually, which at Starship V3’s designed 100 metric tons is about 10,000 launches a year, by our arithmetic.
SpaceX IPO prospectus, 12 June 2026
“Our goal over time is to launch 100 gigawatts of compute to space each year.”
Three Implications
FOR SPACEX SHAREHOLDERS The prospectus ties part of Musk’s AI CEO Award to non-Earth-based data centers capable of delivering 100 terawatts of compute per year, 1,000 times the 100-gigawatt goal by our arithmetic.
FOR THE FCC PROCESS The Space Bureau accepted the application on 4 February; SpaceX asked for waivers including one from the NGSO milestone requirements and deployment obligations.
FOR RIVAL ORBITAL COMPUTE PLANS Blue Origin filed for up to 51,600 satellites, as reported by Advanced Television; Google names twin interconnected Suncatcher satellites in 2027 as its next step.
The framework puts it this way: “The May 2026 Map of AI argued that the seven-layer AI stack is now governed from the bottom — energy, silicon, compute capacity.”
As we read it, SpaceX’s filings address the bottom of that stack from orbit: solar power in sun-synchronous orbit for energy, and gigawatts of compute launched per year for compute capacity. The framework does not discuss SpaceX’s FCC application or orbital data centers.
We read the FCC’s 4 February notice, not SpaceX’s full application, and we did not search the FCC docket for later filings; we make no claim on whether the application has been granted. Musk’s 4 November post gives no reason and does not mention the filings.
The prospectus states goals and expectations, not commitments with dates beyond “as early as 2028” and “By the end of the decade”. The Blue Origin figures are as reported by Advanced Television; we did not read that filing. We make no forecast on launches, approvals or the award.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Musk’s post this morning drew over 1.5 million views in under an hour. SpaceX’s filings list an FCC application for up to one million orbital data center satellites, about 90 times SpaceX’s Starlink fleet by our arithmetic, a prospectus goal of 100 gigawatts of compute to orbit a year, and a Musk award condition 1,000 times that goal. The latest 10-Q shows 1.7 GW installed on the ground and three Starship launches in the first nine months of the year.
A note on sourcing. The FCC application details come from the FCC Space Bureau’s public notice DA 26-113 of 4 February 2026; SpaceX’s goals and Musk’s award terms come from its 12 June 2026 prospectus on SEC EDGAR; current figures come from its 10-Q for the quarter to 30 September 2026. Blue Origin’s figures are as reported by Advanced Television. Musk’s posts were read via api.fxtwitter.com on 4 November 2026. Ratios are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
