← SpaceX
InvestAnswers
How To Allocate To AI: TSLA + SPCX + SOTP Valuation, Moat & Abundance🚀🤖💰
⏱ 84 min video · 4 min read21 May 2026Worth watching
TL;DR
InvestAnswers host and veteran tech developer Phil conduct a deep-dive into how to allocate to AI, covering Tesla's revenue ramp across FSD, Semi, RoboTaxi, and Optimus, plus a sum-of-the-parts valuation of SpaceX that reaches $2 trillion without even counting TerraFab. The core argument is that SpaceX will become the mandatory compute backbone for all AI because terrestrial power capacity will hit a hard ceiling within years.
Key points
1
SpaceX S1 revealed nearly $19B in core revenue; the Anthropic AI compute deal at $15B/year proves Elon Web Services is a real, high-margin standalone business.
2
Phil argues SpaceX controls 'the spice' — vertical integration of chip production (TerraFab), orbital transport (Starship), and space-based data centers — making it the only viable path to meet AI compute demand as terrestrial power caps out near 1 terawatt.
3
Sum-of-the-parts SpaceX valuation: Space Launch $250B, Starlink $500B, xAI/Cursor $400B, orbital data centers $200B, Elon Web Services $600B = $2T, with TerraFab adding an estimated $5T on top.
4
Optimus is currently running on AI4 chips (prototype-only status); AI5 is needed for customer-ready units and is expected in volume in Q1 2027, making 2027 the real Optimus production inflection year.
5
Cursor 2.5 benchmarks at 63.2% at $0.55 per task vs $11 for Claude Opus 4.7 and $4.37 for GPT-5.5, representing a step-change cost advantage that will drive rapid developer adoption.
Actionable insights
→
If you receive a SpaceX IPO allocation at a ~$1.75T valuation, the guest argues it is genuinely cheap given a defensible $2T+ SOTP — consider holding rather than flipping.
→
For a 3-5 asset AI portfolio, the guest prioritizes Tesla (bought on every red day via dollar-cost averaging) and SpaceX as the two highest-conviction positions, with Apple being trimmed to fund Tesla buys.
→
Tesla CyberCab/RoboTaxi revenue is expected to hit meaningfully in 2026, but the real scale inflection for both RoboTaxi and Optimus is 2027 — position sizing should reflect that timeline.
→
Cursor 2.5 is worth trialing now for AI-assisted development; the guest (40+ years coding experience) reports it anticipates architectural concerns unprompted, a meaningful edge over alternatives.
→
Do not factor Chinese humanoid robots into the competitive threat model for Western markets — regulatory and security barriers mean they will not be permitted on US soil.
Notable quotes
“If you believe in AI, if you sit back and you go, AI is the biggest thing I have ever seen, then you have to realize that AI gets delivered through SpaceX in the next couple of years. There is no other option.”
“Ignore all the other businesses that SpaceX is theoretically in. Think of it as nothing more than a company that controls the spice — and the spice is compute.”
“These tools obviate the need for entry level computer science people, but they don't obviate the need for advanced computer science people. It has turned upside down.”
Worth watching?
✅
Worth watching the full video?
Watch if you want the live reasoning behind the SpaceX SOTP valuation and the orbital compute thesis — the key numbers and arguments are captured here, but the back-and-forth between two technically experienced investors adds texture that is worth hearing directly.
Saved you some time? The creator still deserves a like.
Want this for your own channels?
Add the channels you follow. Every new video summarised and in your inbox the moment it drops. From £4/month.
Try it free