FundaAI, a research newsletter published on Substack, wrote on 6 November that “SpaceX has ordered 6,000 space-specific racks for 2027 from a leading server ODM”, which it says equal 432,000 GPUs and 1.05 gigawatts of on-orbit compute (FundaAI). SpaceX has not confirmed the order; FundaAI does not name the manufacturer.
FundaAI puts the hardware value at “~$42โ45 billion” at terrestrial rack pricing, or about $7.0 million to $7.5 million per rack and roughly $97,000 to $104,000 per GPU, by our arithmetic.
Business Pill · Vertical Integration
An evergreen Business Engineer explainer on vertical integration. It teaches the general idea only and does not discuss SpaceX or FundaAI.
The key insight: FundaAI reports 6,000 space-specific racks ordered for 2027, equal to 432,000 GPUs and 1.05 GW; by our arithmetic that is $7.0 million to $7.5 million per rack at terrestrial pricing and 20% of the roughly 30,000-rack program it describes.
The research note says the order went to “the sole JDM (joint design manufacturer) on the program”, which is “due to complete NPI in May 2027, in time for the test satellite, and finish production in December”. NPI is new product introduction, the stage before volume manufacturing.
FundaAI puts the program’s lifetime volume at “~30,000 racks”, matching a first phase of about 30,000 satellites, or “~2.16 million GPUs at 72 per rack, equal to 5.25 GW (7.5 GW peak)”. It adds that “the first 6,000 racks are a fifth of that”.
FundaAI’s estimate of SpaceX’s space-compute rack program: 6,000 racks ordered for 2027 (1.05 GW, 432,000 GPUs) against a program lifetime of about 30,000 racks. Data: FundaAI, 6 November 2026, as published; an analyst estimate, not a SpaceX disclosure.
What Was Posted on X
Wow. According to FundaAI, SpaceX has ordered 6,000 racks for 2027 just for space-based data centers.
Jukan’s post at 23:55 UTC on 6 November, which spread FundaAI’s figure on X with a screenshot of the research note’s paragraph on 6,000 racks for 2027. The note puts the 2027 order at 1.05 GW; separately, it gives about 1 GW on orbit by the end of 2028 as its bull case.
One Rack per Satellite
FundaAI says each satellite carries one NVL72 rack, and that “SpaceX has disclosed AI1 at 250 kW peak and 175 kW average with a 160 mยฒ radiator”. At 175 kilowatts per satellite, 6,000 racks come to 1.05 gigawatts, the figure FundaAI uses.
The note describes a space version of NVL72 with the codename Kessel, which it says “keeps the 72-GPU NVLink domain” and compresses the 48U terrestrial rack into “a ~9U flat enclosure”. It says the first generation “targets inference; large-model training stays on the ground”.
The 1 GW Is a Bull Case for 2028
The viral post on X said SpaceX plans to launch 1 GW into space. FundaAI frames that as a bull case: “~1 GW on orbit by end-2028”. It assumes about 220 Starship launches in 2028, of which about 154 go to compute satellites, at 40 satellites per launch.
By our arithmetic, 154 launches of 40 satellites is 6,160 satellites, slightly more than the 6,000 racks. FundaAI names two checkpoints: Starship Flight 15, which it expects no earlier than late November, and a test satellite on Starship in May 2027 that it says will deliver the first on-orbit thermal data.
Why the Note Ties It to the Midterms
FundaAI argues that “Constraints on the ground are pushing compute into orbit”, citing the 3 November midterms and the incoming Texas governor’s pledge to pause new data centers. It says orbital compute “needs no land, grid connection or water, and no state or local data-center permits”.
Our Numbers
By our arithmetic on FundaAI’s figures: $42 billion to $45 billion across 6,000 racks is $7.0 million to $7.5 million per rack; across 432,000 GPUs it is about $97,000 to $104,000 per GPU; and 6,000 of 30,000 racks is 20%.
Connecting the Coverage
SpaceX’s orbital plans run through our recent coverage: its filing to seek up to one million orbital data-center satellites (our piece), Musk’s claim that next-generation Starlink lasers could compete with subsea cables (our piece), his Baron conference remarks on SpaceX’s fourth-quarter revenue (our piece) and his accusation that SoftBank defaulted on turbine orders (our piece).
The Structural Read
As we read the note, the order figure comes from an analyst’s industry checks rather than from SpaceX.
As we read it, the 1 GW in the viral post is FundaAI’s bull case for the end of 2028, which depends on about 154 Starship launches carrying compute satellites that year.
As we read the thermal section, the design choice FundaAI describes is one rack per roughly 2-tonne satellite, which it says avoids on-orbit assembly.
FundaAI, 6 November 2026
“Constraints on the ground are pushing compute into orbit.”
Three Implications
FOR THE SERVER SUPPLY CHAIN FundaAI says a single unnamed server ODM is the sole joint design manufacturer, due to complete NPI in May 2027 and finish production in December.
FOR THE LAUNCH SCHEDULE FundaAI names Starship Flight 15 and a May 2027 test satellite as the two checkpoints for its estimate.
FOR READING THE CLAIM SpaceX has not confirmed the order, and the 1 GW figure is FundaAI’s end-2028 bull case.
The analysis tallies “~$1.04 trillion in 2026 compute capex, the first trillion-dollar year of compute capex in human history.”
As we read it, FundaAI’s $42 billion to $45 billion for 2027 space racks would be a new line on that map if confirmed; the analysis does not discuss orbital compute or SpaceX’s order.
What Is Not Established
The order is FundaAI’s report from its industry checks; SpaceX has not confirmed it, and the note does not name the manufacturer. The 1 GW figure is FundaAI’s bull case for the end of 2028, not a SpaceX plan. The hardware value uses terrestrial rack pricing, not a disclosed contract price.
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.
FundaAI reports that SpaceX has ordered 6,000 space-specific racks for 2027, equal to 432,000 GPUs and 1.05 GW, with about $42 billion to $45 billion of hardware value at terrestrial prices. That is $7.0 million to $7.5 million per rack and a fifth of the roughly 30,000-rack program FundaAI describes, by our arithmetic.
A note on sourcing. The order and program figures come from FundaAI’s 6 November research note, linked in the text, read in its public portion; SpaceX has not confirmed them. The per-rack and per-GPU values 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.
