The SpaceX IPO, Fable 5, AI Capex Update & Market Check w/ Gavin Baker, Andrew Fox & Clark Tang
SpaceX can stand up AI data centers in 122 days versus the industry standard of 3+ years, enabling them to monetize compute at premium rates ($50B/gigawatt for Google vs. industry average of $14B). This speed advantage, combined with their exclusive Cursor coding data and ability to secure scarce GP
1h 20mKey Takeaway
SpaceX can stand up AI data centers in 122 days versus the industry standard of 3+ years, enabling them to monetize compute at premium rates ($50B/gigawatt for Google vs. industry average of $14B). This speed advantage, combined with their exclusive Cursor coding data and ability to secure scarce GPUs, positions them as the potential AWS of AI infrastructure—a massive revenue stream that didn't exist in forecasts six months ago.
Episode Overview
A deep dive into SpaceX's upcoming IPO, exploring three major business segments: launch/Starlink connectivity (growing from $10B to $50B by 2028), terrestrial AI compute infrastructure (where they've become the #4 hyperscaler in 30 days), and the future potential of orbital data centers. The discussion reveals how SpaceX's unique engineering capabilities and the Cursor acquisition have created an unexpected AI infrastructure powerhouse.
Key Insights
SpaceX Became the #4 AI Hyperscaler in 30 Days
Through deals with Anthropic and Google, SpaceX went from not being considered an AI compute player to the fourth-largest hyperscaler globally, surpassing Oracle. Their ability to stand up 100,000 GPU clusters in 122 days (versus 3+ years industry standard) creates a massive competitive moat. They're monetizing at $22-50 billion per gigawatt versus industry expectations of $14 billion.
The Cursor Acquisition Changes the AI Model Game
The acquisition of Cursor (700-800 people, projecting $10B revenue run rate) brings proprietary coding data that exceeds what exists on the public internet. Their Composer 2.5 model achieved Pareto dominance with minimal training; when combined with SpaceX's massive compute in the Colossus 2 cluster, this could create a frontier-level coding model that scales revenue rapidly once on the Pareto curve.
Rapid Reusability Unlocks Everything
Two-stage rapid reusability of Starship is the foundational technology enabling all other business segments. It reduces launch costs from $1,500/kg to potentially under $250/kg, making orbital compute economically viable at $30B/gigawatt versus $60B terrestrially. The goal: launching 2-3 Starships per day within 3 years (from ~165 launches last year to thousands annually).
Orbital Compute Economics Are Compelling But Not Required
Space data centers could cost $5B/gigawatt in infrastructure versus $25B terrestrially (5x reduction), with free power, cooling, and 'land.' Combined with GPU costs, total orbital deployment would be $30B/gigawatt versus $60B+ terrestrially. However, investors can get excited about the IPO based purely on terrestrial AI compute and Starlink growth, with orbital as a massive call option.
Starlink Has Barely Scratched the Surface
Despite being a major revenue driver, Starlink has less than 1% penetration of global households. At only 0.3% penetration of the $15+ trillion global telecom market, the path to $50B in connectivity revenue by 2028 (from ~$10B today) requires minimal market share gains. The service is consistently the fastest, lowest latency connection globally, creating a 'better, faster, cheaper' winning formula.
Notable Quotes
"What they achieved is is singular. Never been done before. Just to put in perspective, 100,000 GPUs, that's you know, easily the fastest supercomputer on the planet as one cluster. Um, a supercomputer uh, that you would build would take normally 3 years to plan. And then they deliver the equipment, and it takes 1 year to get it all working. Yes. We're talking about 19 days."
"I'm not the most sophisticated thinker, but that math maths. You know, if you can borrow money at 6, 7, 8% and invest in something with a 55% ARR, I'm not the most sophisticated thinker, but that math maths."
"I think we're all pretty AI pilled. And if you're AI pilled, that means we got to build a lot more compute than the world thinks. And that these models are going to be a lot more valuable than people think. You combine that with their core business, I don't know another entrepreneur or another business that's a better bet on the future, right, than SpaceX."
"Speed is literally cost because every day you're paying electricians and plumbers. That's cost. And they're now monetizing them at arguably the highest rate."
"I think coding is going to continue to be very important. John Massad posted something very interesting. The founder of Replit. It he called it bitter lesson adjacent that coding may be the fastest path to AGI and ASI because if you really go to coding, you can write code if a model's good at coding to do anything."
Action Items
-
1
Evaluate Your Compute Strategy Through the SpaceX Lens
If you're building AI products, assess whether you're optimizing for speed-to-market with your compute infrastructure. SpaceX's 122-day deployment cycle versus industry 3+ year timelines shows that speed literally equals cost savings and competitive advantage. Consider whether your current cloud provider can scale at the pace your model development requires.
-
2
Identify Your 'Proprietary Data Moats' Like Cursor Did
Cursor's value came from having more proprietary coding tokens than exist on the public internet. Audit what unique data your company generates that could become a training advantage. Focus on collecting and organizing data that competitors cannot easily replicate, especially interaction data and domain-specific patterns.
-
3
Apply First Principles Thinking to Legacy Industries
SpaceX approached rockets, electric cars, and data centers by questioning fundamental assumptions rather than incrementally improving existing designs. Identify an industry problem you're working on and ask: 'If we were designing this from scratch today, knowing what we know, what would we build?' Challenge every constraint that seems 'just the way things are done.'
-
4
Build Optionality Into Your Strategic Bets
SpaceX's IPO valuation works even without orbital compute, but orbital represents massive upside optionality. When making major investments or strategic decisions, structure them so the base case is viable but there are clear paths to 5-10x outcomes if key assumptions prove true. Don't bet everything on the moonshot, but ensure the moonshot is possible.