Like most of you, I watched the Terafab livestream last Saturday night.
Musk was standing in a decommissioned power plant in Austin — the kind of building that hasn’t powered anything since the 90s — and he told the crowd he was going to build the biggest chip factory in history. $25 billion. Two-nanometer technology. 200 billion chips per year. Everything from design to packaging under one roof.
Tesla dropped 17% the next morning.
I’m going to be honest with you: as someone who’s been invested in Tesla for over ten years, my reaction was the opposite. I wasn’t panicking. I was actually pretty excited. Not because I think Musk can compete with TSMC — I don’t think that’s what he’s doing. But because for the first time, I felt like the full picture of what he’s building finally clicked into place.
Let me explain.
I’ve spent a long time — honestly, years — trying to answer one question: what is Musk’s actual plan?
Not the surface-level plan. Not “electric cars” or “rockets” or “robots.” The real plan. The one that connects all the pieces together and makes sense of decisions that look irrational in isolation.
Because if you just look at the headlines, Musk’s moves over the last two years make no sense. Why would the CEO of the world’s most valuable automaker spend his time running a social media platform? Why would he merge his AI company with his rocket company? Why would he announce a $25 billion chip factory when Tesla’s auto business is declining for the second year in a row? And why would 80% of that factory’s output go to space instead of to Tesla cars?
None of this makes sense if you think Musk runs a car company. All of it makes sense if you understand the real thesis.
And I think the real thesis starts with a book.
I’m going to make a detour here, but stay with me — it’s the most important part of this article.
When Musk was 14, he was going through what he later called an “existential crisis.” He was reading Nietzsche and Schopenhauer — which, as he put it, “you should not read at age 14. It is bad. It’s really negative.” Then he picked up The Hitchhiker’s Guide to the Galaxy by Douglas Adams.
He’s never shut up about this book since. He calls Douglas Adams his “favorite philosopher.” He named SpaceX’s drone ships after ships from Iain M. Banks’ sci-fi novels. He named his AI chatbot “Grok” after a term from Heinlein. He said he’d name the first SpaceX ship to Mars “Heart of Gold” — the ship from Hitchhiker’s Guide.
But here’s the thing most people miss about how this book shaped Musk. It’s not the space travel part. It’s a philosophical idea buried in a joke.
In the story, a supercomputer called Deep Thought spends millions of years computing “the answer to the ultimate question of life, the universe, and everything.” The answer it finds: 42. Completely meaningless. Because nobody had ever understood the question well enough.
Musk has referenced this lesson dozens of times: “A lot of times the question is harder than the answer. If you can properly phrase the question, then the answer is the easy part.”
I’ve been investing in Tesla for over a decade, and for most of that time I was trying to answer the wrong question. I was asking “is Tesla a good car company?” or “can Musk deliver on his timelines?” Those are the questions Wall Street asks. They’re not the question Musk is asking.
I think the question Musk has been asking himself for twenty years — the one that explains every decision he makes — is something much bigger:
What if humanity’s future doesn’t come from Earth? What if the energy we consume, the infrastructure that powers every product and service we use, the compute behind our AI, the connectivity that links everything together — what if all of that, long-term, comes from space?
Not going to space. Getting resources from space.
Think about that for a second. Today, we’re hitting walls everywhere on Earth. Data centers can’t get enough power — Bloomberg found electricity costs near data centers spiked up to 267%. Grids are maxed out. Communities fight every new facility. Water for cooling is running out in places like Arizona. And every year, the demand for AI compute grows faster than our ability to supply it from the ground.
Now look up. In space, solar energy is near-continuous. There’s no grid to overload. No land to buy. No water to consume. No community to fight. No electricity bill. And it scales infinitely — you just launch more satellites.
Musk understood this before anyone in tech. And I think he’s been executing on it for twenty years without ever saying it clearly, because if he did, the market would call him even crazier than it already does.
Asimov’s Foundation series — another book Musk calls one of the all-time greats — is about a collapsing galactic civilization and a small group of people who build the technological infrastructure to preserve knowledge and restart it. Musk has explicitly said: “Given that this is the first time in 4.5 billion years where it’s been possible for humanity to extend life beyond Earth, it seems like we’d be wise to act while the window was open.”
He’s not being poetic. He’s being literal.
And here’s the historical parallel that made this click for me.
We’re not in the age of Christopher Columbus anymore. There are still things to discover on Earth. But in space, there’s more. There’s always been more. The pattern of human civilization is that the biggest explosions of wealth creation come from accessing new sources of resources — not from optimizing existing ones.
In the 15th century, Columbus understood that the real value was across the ocean. The monarchs who financed ships captured the wealth of the next three centuries. They didn’t build better farms in Spain. They built ships.
In the 19th century, Rockefeller understood that the real value was underground. He didn’t sell petroleum — he built the refineries and the pipelines. The infrastructure. Standard Oil became the most valuable company in history.
In the 21st century, I believe Musk understands that the real value is above our heads. Unlimited solar energy. Unlimited compute. Global connectivity. Maybe, eventually, raw materials and waste processing. And he’s not building rockets to go there — he’s building the refineries and pipelines of space. The infrastructure that brings those resources back to Earth in the form of products and services that no terrestrial competitor can match.
That’s the thesis. Everything else — the cars, the robots, the chips — is downstream.
The infrastructure-first playbook
Once I saw this, I started looking at every Musk company through one lens: he never bets on products. He bets on the infrastructure that makes his products impossible to compete with.
The product is what you and I see. The infrastructure is the invisible layer underneath that actually determines who wins. And every single time Musk invests in infrastructure, the market panics — because it looks like reckless spending on something that isn’t generating revenue. Then the cost curve crosses a threshold, and suddenly he owns a monopoly.
Let me show you the pattern with real numbers, because this is what convinced me.
The Gigafactory bet. In 2014, Musk announced he was building a battery factory in Nevada that would produce more lithium-ion cells than the entire rest of the world combined. At the time, Tesla sold 35,000 cars a year and was burning cash. Everyone said it was suicide. The stock was at $15 (split-adjusted). But the Gigafactory wasn’t about making batteries — it was about owning the cost curve of the energy layer underneath electric vehicles. Once Tesla controlled battery production at scale, nobody could compete on cost. The stock went from $15 to $900. That’s a 60x return — for people who understood that the infrastructure was the bet.
The Falcon 9 reusability bet. SpaceX spent 13 years and billions of dollars figuring out how to land a rocket. Thirteen years. The internet was full of compilation videos called “SpaceX rocket crashes.” The aerospace industry thought it was a physics stunt. But Musk wasn’t building a rocket — he was building the infrastructure layer of cheap access to orbit. Once you can reuse a booster 40 times instead of throwing it away, launch costs drop by 95%. SpaceX’s valuation went from $12 billion in 2015 to $1.25 trillion today. That’s 100x in 11 years. For the people who understood that the explosions were R&D, not failure.
The Starlink bet. This is the one that really nails the pattern, because the cost curve is documented quarter by quarter. In 2020, Starlink launched with 10,000 beta users and zero revenue. Each user terminal cost $2,400 to make but was sold at $499 — SpaceX was losing $1,900 on every single customer. The market saw a money pit.
Here’s what actually happened. Terminal cost: $2,400 in 2020, $1,300 in 2021, $500 in 2023, profitable at $349 in 2025. Revenue: $0 in 2020, $1.4 billion in 2022, $4.2 billion in 2023, $8.2 billion in 2024, roughly $15 billion in 2025. Subscribers: 10,000 in 2020, 9 million+ today. From zero to potentially $15 billion in five years. The investors who looked at Starlink in 2020 and saw “unsustainable losses” were right about the present and catastrophically wrong about the future.
The Tesla Energy bet. Battery Day in September 2020 promised 4680 cells, 3 TWh by 2030, 50% cost cuts. Five and a half years later, the 4680 is years behind schedule. Tesla’s own top battery supplier publicly said “Elon doesn’t know how to make battery cells.” The market called it a failure.
Meanwhile, while everyone was tracking Battery Day promises, Tesla was building the fastest-growing energy storage business on the planet. Deployments in 2025: 46.7 GWh, more than double the prior year. Q4 2025 energy revenue: $3.84 billion, up 25% year over year. Full year: around $12.6 billion. That’s bigger than most utilities. The market spent five years calling Battery Day a failure. Tesla spent five years building a $12.6 billion business nobody was watching.
Every single time: infrastructure investment → market panic → cost curve inflection → monopoly.
Now I can tell you what I think Terafab is really about. And it’s not what Jensen Huang thinks.
Jensen said at a TSMC event in November: “Building advanced chip manufacturing is extremely hard. It’s not just build the plant, but the engineering, the science and the artistry of doing what TSMC does for a living is extremely hard.” He called matching TSMC “virtually impossible.”
And he’s right. If Terafab was trying to compete with TSMC on the open market — selling chips to Apple, Qualcomm, NVIDIA — it would fail. No question. TSMC has 30 years of process engineering. Intel has been trying for a decade with $100 billion and still can’t catch up.
But here’s the number almost nobody reported from the Terafab event: 80% of Terafab’s production goes to space. Only 20% is for Earth.
Musk isn’t building a chip factory to compete with TSMC. He’s building a chip factory to supply his own orbital data center constellation — a market that doesn’t exist yet, and that only he can serve.
On January 30, SpaceX filed with the FCC to launch up to one million satellites as orbital AI data centers. One million. The filing says: “Orbital data centers are the most efficient way to meet the accelerating demand for AI computing power.” The satellites would orbit at 500-2,000km in sun-synchronous orbit — meaning near-continuous sunlight. Near-continuous solar power. No electricity bill. No water for cooling. No land. No NIMBY fights.
At the Terafab event, Musk showed the “AI Sat Mini” — the first-generation satellite. Each one: 100 kilowatts of onboard AI compute. Over 170 meters long at scale. Future versions: one megawatt. The chip inside? The D3 — a radiation-hardened processor designed to run hotter than terrestrial chips, optimized for the hostile environment of space.
That’s what 80% of Terafab produces. Not chips to compete with TSMC on Earth. Chips for a market in space that doesn’t exist yet — and where Musk is the only person who owns the rockets to launch them, the satellite network to connect them, and the AI models to run on them.
Jensen is looking at the wrong market. He’s measuring Terafab against TSMC’s terrestrial business. But Musk isn’t playing on TSMC’s field. He’s building infrastructure for a field that nobody else can access.
It’s the Starlink terminal playbook again. SpaceX didn’t buy terminals from an existing manufacturer — it built the world’s largest phased-array antenna factory and drove the cost down 85% in five years. Terafab is the same approach applied to semiconductors. Not competing with the incumbents. Controlling the supply chain for a market only you can serve.
The space thesis that connects everything
Here’s what I’ve come to understand, and I think this is the single most important insight for anyone investing in any Musk company.
Every project Musk runs converges on space. Space is the infrastructure layer. The products on Earth are downstream.
Think about it. The three biggest constraints every technology company faces today:
Energy — on Earth, energy is scarce, expensive, and every new data center triggers community opposition. Bloomberg found that areas near data centers saw electricity costs spike by up to 267%. In space? Continuous solar power. No grid. No bill.
Compute — on Earth, AI data centers are limited by power, water, land, regulations. In space? Solar-powered processors with heat dissipated by radiation. No water. No real estate constraint. Musk said at Terafab: “Increasing power on Earth becomes harder over time and more expensive. In space it becomes cheaper and easier over time.”
Connectivity — on Earth, fiber optic is expensive and doesn’t reach everywhere. In space, Starlink already covers 140+ countries with 9,500 satellites and laser links at 200Gbps, soon 1Tbps+.
Musk understood something before anyone else in tech: space has more resources to offer humanity than we imagine.Not as a destination to visit. As a platform to build on. A platform for unlimited energy, unlimited compute, and global connectivity — that feeds back down to Earth in the form of products nobody can compete with.
And here’s the complete stack he’s been assembling for twenty years:
SpaceX → the transport layer (cheapest access to orbit, 134 launches in 2024 alone) Starlink → the network layer (9,500+ satellites, global coverage) Terafab → the hardware layer (custom chips for space and Earth) Orbital data centers → the compute layer (1M satellites, solar-powered AI) xAI → the intelligence layer (Grok, models running on orbital compute) Tesla/Optimus/Cybercab → the product layer (autonomous vehicles and robots powered by all of the above)
No other person alive controls all of these layers simultaneously. That’s the moat.
And he’s not alone in seeing it. Jeff Bezos predicted “giant gigawatt data centers in space” within 20 years. Sam Altman said space might be “the long-term solution” for AI compute. Former Google CEO Eric Schmidt bought a rocket company as a play on orbital data centers. Blue Origin announced a competing satellite system. But none of them have the rockets, the satellite network, the AI company, the chip factory, the car fleet, and the robot program all in the same ecosystem.
Note before we continue: Macro Notes is now open to sponsors. You can promote your company to our audience of 65,000 professional and retail investors.
Our standard CPM is $100 (which typically translates to $2,000–$2,500 per placement).
However, for your first placement, we offer a fixed test price of $500.
This is a limited-time offer — feel free to reach out at pierre@altis.finance to discuss further.
What this means for investors
Let me bring this back to what matters: your money and mine.
The market is punishing Tesla because auto deliveries are declining. And yes, deliveries fell to 1.636 million in 2025, down 8.6%. Revenue dropped to $94.8 billion — first annual decline ever. BYD passed Tesla to become the world’s biggest EV maker.
If Tesla is a car company, the bears are right. At $360 a share, with 130x trailing earnings, it’s overvalued for a declining automaker.
But I’ve just walked you through what I believe is the real thesis. And the real thesis says Tesla isn’t a car company. It’s the publicly traded entry point to a vertically integrated space-AI-energy-robotics infrastructure conglomerate.
The auto business isn’t the thesis — it’s the cash engine funding the transition. The same way Amazon’s retail business funded AWS. The same way SpaceX’s launch business funded Starlink. Every one of Musk’s companies goes through a phase where the visible business declines while the invisible infrastructure grows underneath. That’s what’s happening right now.
The catalyst calendar has never been this loaded. Q1 delivery report in two days. SpaceX IPO potentially this summer at $1.5 trillion — which would be the largest IPO in history. Cybercab production starting in H1 2026. Wedbush predicting a Tesla-SpaceX merger in 2027. Terafab AI5 chip pilot production by late 2026.
Here’s the question I keep asking myself: in 10 years, will the market still be valuing Tesla on quarterly car deliveries? Or will it be valuing the entity that manufactures its own chips, launches its own satellites, operates orbital AI data centers powered by unlimited solar energy, and uses that intelligence to run billions of autonomous vehicles and robots on Earth?
I know which side I’m on.
My full Tesla study continues in the premium section of this edition. Here’s what’s inside:
How I value Tesla not as a car company but as a space infrastructure conglomerate — with a sum-of-the-parts model covering auto, energy, robotaxi, Optimus, the SpaceX stake, and Terafab optionality.
Three scenarios with specific price targets: bear ($200), base ($550), bull ($900+) — and the exact catalysts that trigger each.
My positioning strategy: buy now or wait for the Q1 report? My entry zone, my sizing rules for a stock this volatile, and the one discipline that’s kept me invested through every Musk chaos cycle for ten years.
The SpaceX IPO playbook: how to prepare, the indirect access vehicles available today (EchoStar, Alphabet, KraneShares AGIX), and why this IPO changes everything for Tesla shareholders.
The 5 metrics I track that will tell us 6-12 months early whether the plan is working or failing — before the market catches on.
And the “satellite positions” — companies that benefit from the Musk space buildout without the concentration risk of TSLA…