How DeepStack settles a question · Part 02
SpaceX: exit or ten-bagger?
SpaceX after the IPO: exit liquidity, or the next ten-bagger?
From Part 02: SpaceX Is Funding an AI Bet Before the Proof. 6 voices on 2 sides, attributed and dated; the DeepStack Call with its history; 4 dated tests that settle it; the results as they come in.
The mechanism
SpaceX raised $75 billion on June 12 at a valuation near $1.75 trillion, and the price assumed an AI buildout the meter has not yet recorded: 0.4 gigawatts of compute is deployed against a site plan near 10 gigawatts by the end of 2027. A site-backed plan is meaningful, but it is not cash generation, and the $1.25 billion a month Anthropic pays through May 2029 is contracted demand that still has to prove durable without the capital cycle doing the work. Starlink's $4.4 billion of operating income carries the build, while the widening AI loss leaves shareholders little room for error. The stock therefore moved ahead of the operating evidence: the capacity is credible, and the proof arrives one deployed gigawatt at a time.
Approved by Marco Cauduro on October 10, 2026.
The DeepStack callSkeptical
SpaceX’s AI build is credible. The stock has already paid for 10 gigawatts it has not built.
Our Call takes the side “Exit liquidity”.
Call history
- v2Ahead of the evidenceConviction Moderate (3/5)Side: Exit liquiditySide named by the owner.
- v1Ahead of the evidenceConviction Moderate (3/5)First call, as published with the story. The conviction, the actions and the flips were set down for the site on October 7, 2026.
4 voices
Exit liquidity
Retail buyers of the SpaceX IPO are the exit liquidity: the fancy name for paying insiders, VCs and early investors at peak hype valuations.
The Google SpaceX GPU rental is bullish for Nvidia because there are still not enough GPUs to quench demand; Nvidia ramps production continuously but cannot follow demand.
SpaceX deployed 0.4 gigawatts of nameplate compute in the last quarter while Musk promises more than 10 gigawatts by end-2027 and describes close to 15 gigawatts at the PowerPoint level.
SpaceX's AI segment reveals the industry's bad accounting habits: how AI revenue is counted, what is hidden in construction-in-progress, and where the Nvidia GPUs everyone calls short-supplied actually are.
2 voices
The number is real
Selling frontier-model inference is extraordinarily profitable: on a GB300 cluster, OpenAI and Anthropic can generate over 100 billion dollars per gigawatt-year of revenue against roughly 12 billion of rental cost.
SpaceX is on track to build about 10 gigawatts of datacenter capacity by year-end 2027; independent site-by-site evaluation supports the guided 6 to 8 incremental gigawatts, implying a path to 300 billion dollars of ARR with only half the…
3 voices
Voices that cross sides
SpaceX at roughly 100 times trailing revenue would never qualify as a value investment, yet it is feasibly a ten-bagger: an extreme bull scenario could carry the company toward an 18 trillion dollar valuation or beyond over ten to twenty-five years.
Lenders back Starlink's cash and bet the AI spend never drains the company.
Long AI in the long run, doubtful the market absorbs three $1T cash-burning IPOs at once.
Where they cross
Who answers whom.
The specific points where one voice meets another: the same evidence read two ways, or the same mechanism with a different sign.
Kris contradicts Trey Henninger, on premise
Kakashii qualifies Jeremie Eliahou Ontiveros, on the evidence
Kakashii qualifies Kris, on the evidence
Alternative explanations
Three other readings.
What would have to be true for the Call to be wrong, as the other side argues it.
- The price is early, not wrong: at about 100 times trailing revenue SpaceX is no value investment, yet an extreme bull scenario carries it toward $18 trillion over ten to twenty-five years (the case Trey Henninger makes).
- The 10 gigawatts arrive on schedule and each one earns: over $100 billion of revenue per gigawatt-year against about $12 billion of rental cost, a path to $300 billion of ARR (the case Jeremie Eliahou Ontiveros makes).
- Lenders have already priced it: they back Starlink's cash and bet the AI spending never drains the company, so the bond market, not the stock, carries the proof (the case JunkBondInvestor makes).
What settles it
The dated tests.
The tests the story set, each with its specification where the owner has written one and the piece of the argument it verifies; on the Docket, resolved on the Results page.
Quarterly deployed gigawatts will test the roughly 10 GW site plan against 0.4 GW operating now.
Part 02: SpaceX Is Funding an AI Bet Before the Proof Pendingp02-t01Open questionChecks the deployed-gigawatts claim
Specification pending
Anthropic’s $1.25B monthly commitment expires; payment durability will test contracted AI demand.
Part 02: SpaceX Is Funding an AI Bet Before the Proof Pendingp02-t02 · monthOpen questionChecks the contracted-demand claim
Specification pending
- Next bond tranche
Bond pricing will show whether lenders still trust Starlink to carry AI spending.
Part 02: SpaceX Is Funding an AI Bet Before the Proof Pendingp02-t03Open questionChecks the lender-trust claim
Specification pending
- OpenAI IPO window
A third trillion-dollar cash-burning listing would test Kevin Gee’s market-capacity concern.
Part 02: SpaceX Is Funding an AI Bet Before the Proof Pendingp02-t04Open questionChecks the market-capacity claim
Specification pending
What changed
The record of Part 02.
The Part’s publication, every change to the Call, every result that came in and the next dated test, oldest first.
- PublishedPart 02 published; the Call: Ahead of the evidence.
- Call updatedThe Call: Ahead of the evidence. SpaceX’s AI build is credible. The stock has already paid for 10 gigawatts it has not built. Our Call takes the side “Exit liquidity”. by Marco Cauduro
- Next testNext: Jan 1, 2027, Quarterly deployed gigawatts will test the roughly 10 GW site plan against 0.4 GW operating now. Open question
Sources (3)
- Potential Multibaggers - June 8 SpaceX valuation warning
- JunkBondInvestor - SpaceX credit guide
- SemiAnalysis - SpaceX 10GW capacity analysis
Market data are as of the dates cited. Voices are paraphrased from public writing and attributed by name.
Analysis and opinion, not investment advice. Voices are paraphrased from public writing and attributed by name; DeepStack shows no score, ranking or accuracy for any person. Every question, one page each.