InvestAnswers
InvestAnswersJul 30
Finance

Who Wins the CAPEX Wars in Age of AGI Answer will surprise you!

22 min video5 key momentsWatch original
TL;DR

Tesla and SpaceX are winning the AI infrastructure race by spending 4% of hyperscaler CAPEX but achieving superior returns through vertical integration and physical asset building.

Key Insights

1

2x annual return on CAPEXTesla's $7 billion gigafactory spend generates $15 billion annual profit — delivering a 2x return on capital in a single year, versus hyperscalers cycling money between each other in circular deals.

2

122 days vs 2-3 yearsElon can build a data center in 122 days while Microsoft, Google, and Amazon take 2-3 years — time compression translates directly to cost advantage in the physical AI race.

3

11 million Optimus units annuallyTesla is building 1 million humanoid Optimus units per year capacity in Fremont and 10 million annually in Austin, with production ramping September 2025.

4

Vertically integrated lithium supplyTesla secured North America's largest lithium refinery (30 GWh) in Corpus Christi, eliminating foreign supply chain risk — a move Musk had to fund himself after begging the industry for years.

5

20x and 12x returns achievedRon Baron's Tesla investment turned $400M into $8B+ in profits; his SpaceX $2B stake is now worth $25B with expected 5-10x upside over ten years.

Want this for every new video InvestAnswers posts? Brevyd summarizes each upload automatically, the morning it drops.

Deep Dive

Why CAPEX matters now more than ever

CAPEX has become the new moat in the age of AGI because scale requires physical infrastructure across compute, cooling, energy, data centers, and real estate. The creator lays out four reasons CAPEX is vital: compute demands massive power and cooling systems; manufacturing at scale (humanoid robots, Cybercabs) requires enormous physical footprints; vertical integration from supply chain to finished product matters; and this integration is rare. Hyperscalers are in an arms race spending $750 billion in 2026 combined, but critically, they're mostly buying cloud server farms from each other — a circular flow of capital. Tesla and SpaceX, by contrast, are locking down primary physical assets. The creator emphasizes the distinction: there's CAPEX that funds external players for chips, and CAPEX that builds real tangible stuff in a lean, efficient manner. This distinction is the core of why the upcoming ten years will look radically different.

Tesla's sprawling physical buildout across the globe

Tesla is in the middle of the most ambitious manufacturing and energy build-out in corporate history. In Fremont, they ripped out Model S/X assembly lines in 42 days and converted that space to produce 1 million humanoid Optimus units annually, employing 12,000 people. A second, much larger Optimus factory in Austin can produce 10 million humanoids per year. Separately, Tesla is constructing an AI chip fabrication plant in Austin (Terafab), a 100-gigawatt solar panel factory in Texas with full vertical integration from ingots to modules, and a 30 GWh lithium refinery in Corpus Christi — the largest in North America. They're expanding Giga Berlin for both cars and 4680 battery production, building a mega factory in China that's twice the size of their Lathrop facility, and launching another mega factory in Houston. They've also built two Cortex supercomputer clusters in Texas with hundreds of thousands of Nvidia chips for Full Self-Driving and Optimus neural networks. The 4680 battery expansion in Sparks, Nevada, now produces cells cheaper than Panasonic at scale. Additionally, Giga Satoshi (1,000 acres) will manufacture solar cells for space, AI satellites, and data centers that will operate in orbit. Cybercab factories are producing 120 units daily with 2,000+ already built, and Tesla Semi factories are ramping with hundreds already on roads.

SpaceX and the Terafab convergence

SpaceX is building massive infrastructure at Starbase to support rapid reuse, orbital launches, reusable Starship rockets, Starlink V3 deployments, and orbital AI data center clusters. SpaceX plans to manufacture multiple Starships simultaneously and eventually launch multiple per day. The massive collaboration between Tesla and SpaceX (which the creator notes should be viewed as nearly synonymous for the next 1-2 years, potentially merging before the next presidential election) is the Terafab project — a trillion-watt compute facility per year. Because the US domestic grid is constrained, much of this compute will be deployed in space via SpaceX's orbital infrastructure. This is where the synergy emerges: Tesla will vertically integrate logic, memory, packaging, and testing under one roof, then ship the output via SpaceX's rockets into orbit where it can operate as data centers. This convergence creates an unprecedented physical intelligence backbone that no competitor possesses. The scale is genuinely unheard of in corporate history.

Why $25B CAPEX beats $750B in returns

Ron Baron highlighted on CNBC that legacy automakers lose $50,000 per EV sold while Tesla profits $9,000 per car. Tesla's $7 billion annual gigafactory capital spend generates $15 billion in annual profit — a 2x return in a single year. This matters because it reveals the capital efficiency gap: Tesla's $25 billion total CAPEX (only 4% of what hyperscalers spend) generates asymmetric returns because every dollar targets massive trillion-dollar TAM opportunities: Optimus robotics, Cybercab robotaxis, and the Terafab. By contrast, hyperscalers like Microsoft, Google, Amazon, and Meta spend $750 billion cycling capital through cloud infrastructure, intermediate software, connectivity, and chips — circular flows where one hyperscaler's CAPEX becomes another's revenue. Nvidia sits at the center of these circular deals. The creator's key insight: absolute CAPEX figures are misleading. What matters is return on CAPEX and what physical moat that spending creates. Elon builds data centers in 122 days versus 2-3 years for competitors, compressing time-to-productivity and cost. That execution advantage is the edge.

The trillion-dollar bet on physical robotics

The creator and Ron Baron both emphasize that Optimus humanoid robots represent a larger opportunity than anything Tesla accomplished in automotive. Musk targets 100 million Optimus units eventually. Jensen Huang called Tesla's stack (Grok, Full Self-Driving, Optimus) world-class and revolutionary, noting physical robotics represents the next multi-trillion-dollar industry arriving imminently. The payoff potential is staggering: Musk calls Optimus a $30 billion business, Cybercab robotaxis $10 billion, and Terafab $10 billion minimum. Ron Baron and Peter Diamandis both believe the combined Tesla-SpaceX entity could become the first $100 billion market-cap company. Even CITIC Securities, China's largest asset manager, publicly stated Tesla possesses an unmatched lead in both AI large-scale modeling and large-scale manufacturing, calling Tesla the top tier of the global intelligent industrial supply chain. Meanwhile Cathie Wood is aggressively buying both Tesla and SpaceX on every market dip, viewing current weakness as a classic wall of worry and accumulating for conviction plays.

Takeaways

  • Stop evaluating Tesla's value based on quarterly earnings — track where CAPEX is deployed and what physical moats it builds (Optimus capacity, lithium supply, chip fabs, Terafab).
  • The return-on-CAPEX metric beats absolute CAPEX size; Tesla's $25B generates more long-term value than hyperscalers' $750B because it targets tangible assets with defensible moats.
  • Mark September 2025 on your calendar for Optimus production ramp — this is the inflection point where physical AI moves from prototype to manufacturing, not years away.
  • Watch Tesla and SpaceX as a combined entity rather than separate companies; the Terafab and orbital AI data center convergence between them is the structural backbone of AGI infrastructure.

Key moments

1:30CAPEX is the new moat

I believe CAPEX is the new moat. You got to pay it forward. You got to pay to play. If you don't, you're toast.

5:25Ron Baron's return on capital thesis

Tesla made a 15 billion a year on a 7 billion gigafactory capital spend. In the space for year, they get a 2x return on capital from the investment they make.

15:15Optimus production ramp date

They will be able to build a million of these per year at this factory and they'll be able to build them starting in September in earnest. That is a month and a half away, not far.

18:00Execution speed advantage

It takes them two or three years to build a data center that Elon Musk builds in 122 days. This is the edge. This is the physical AI edge that the other players don't have.

21:30First 100 billion company prediction

Many people were asked with a gun pointed to the head, which company will be the first 100 billion dollar company? Everybody is saying the same thing.

You just read one. Brevyd does this for every upload.

Follow InvestAnswers and every new video comes back as a summary like this, in your morning briefing. No watching required.