SpaceX is framed as a compute-and-energy platform whose valuation could explode from today to $100T+ within 10 years, driven by rapid GPU cluster buildout, Starlink, launch cadence, and in-space/terrestrial data centers. The core claim is that the bottleneck is not money but physical deployment: power, cooling, racks, wiring, connectivity, and transmission, where SpaceX claims 9-12 month data-center ROI versus years for rivals. The note set also leans into merger speculation around SpaceX, TSLA, and XAI, while tying the whole bull case to 2026-2031 compute ramp and next week’s earnings/TSLA Wednesday catalysts.
Elon quote: “Unless there’s a major world war, SpaceX probably gets to more than 100 trillion dollars in less than 10 years.”
Ron Baron is cited at $50T for SpaceX in 10 years, and Starlink is assigned $14T-$15T of that total.
SpaceX S-1 TAM is cited at $28.5T; Elon’s response is quoted as: “We are wrong. It’s vastly higher than that.”
A Morgan Stanley note is cited as ignoring the inference engine across in-space plus terrestrial data centers, calling it close to breaking out “gradually then suddenly,” with Musk replying, “This is a very big deal.”
A Morgan Stanley room is said to have had one answer to “Does anybody own SpaceX?”: not one hand went up.
A Brownsville, Texas site near Starbase is said to span 444.45 acres, with 400 MW of wind turbines on site, and to house many small data-center buildings at once.
The Brownsville build is said to spin up all these clusters in 20 weeks; six buildings are projected at $132B-$180B of annual revenue, roughly twice Tesla’s revenue after nearly two decades.
The roadmap dated October 1 shows Colossus at 150,000 H100s, 50,000 H200s, and 30,000 GB200s; then 210,000 GB200s and 440,000 GB300s; another 220,000 GB300s is said to be fully operational next week.
Another 220,000 GB300s is claimed for November, with a further 220,000 possible before late-December ramp.
The clip claims nobody else has strung together more than 30,000 GPUs, while SpaceX repeatedly does 220,000 GPUs; competitors are said to need 2-4 years for comparable clusters while SpaceX can build a facility in 20 weeks and 100,000 GPUs in 90 days.
The Brownsville site is said to generate at least $22B a year on 220,000 GB300s and $30B on 220,000 Vera Rubins, with Vera Rubin hardware arriving very soon.
Starship is said to be moving 100-150 ton payloads, able to carry about 60 satellites per flight, and wider than a 787 or 747; the FAA is said to be reviewing Starbase expansion for up to 456 launches and 912 landings per year.
SpaceX’s compute push is used to rebut CoreWeave bankruptcy fears: the money is the easy part, while the hard part is building power, cooling, racks, wiring, and connectivity; data centers are said to take 2.5-3 years to build after the money arrives.
The SpaceX CFO is cited saying full ROI on a data center can be reached in 9-12 months, versus about 2.5x as long for others just to get stood up.
North America compute is said to need 33-35 GW next year, short by 13-15 GW; China is said to plan 40-45 GW with only a 2 GW shortfall, implying compute scarcity could send prices sharply higher and lift SpaceX revenue per GW to $50B-$60B.
Barclays is cited modeling TSLA acquisition by SpaceX as the likeliest all-stock structure; Tesla’s Dec. 3 annual meeting is said to have no merger mention in the ballot, with no 8-Ks or press releases, and the deal ‘isn’t going to happen that soon.’
Prediction-market odds for a SpaceX-TSLA merger are said to have spiked to nearly 60% around spring 2027; Polymarket and Kalshi are said to be showing the same thing, with ‘something is cooking.’
Rumors say SpaceX has locked up 30%-40% of NVDA’s Vera Rubin allocation, including some for space next year; a radiation-capable Vera Rubin variant for ‘all high CU’ is mentioned, plus $40B of NVDA financing in talks with Apollo.
Arc is cited at 66-67 GW for Tesla + SpaceX by 2031, while Musk is said to be disappointed if that is all they get.
Compute-only valuation math gives a base case of 66.5 GW online by 2031 and an $825 per-share target, or 5.65x; the bull case doubles the GW and reaches nearly $4,000 per share, or 21.7x.
The ‘retire on bag’ target is 655 shares; at $825 that is about $540,000, and at nearly $4,000 it is about $2.4M; the target is tied to reaching $1M of value in 2032.
Two polls are cited: one where 1 vote chose Bitcoin over SpaceX and 78% chose 100 shares of SpaceX, and another where 0.18 BTC got 26% versus 100 shares of SpaceX at 74% for the next 4 years.
The close frames AI as the faster horse and Bitcoin as the fallback if the world ‘goes to poop,’ with earnings this week and Tesla on Wednesday flagged as possible catalysts for nice surprises.