We’ve been hearing the same story for about a decade: everything Elon Musk builds is a step towards colonizing Mars.

For example:

  • SpaceX rockets will move people and cargo there.

  • Tesla car and battery sales will fund much of the operation—and its battery and solar tech will provide power once folks get there.

  • Starlink will also pay part of the tab, and its satellites—currently blanketing the Earth with high-speed internet connectivity—will do the same for Mars.

  • Optimus autonomous humanoid robots will handle things like building habitats and digging for resources—work that’s too difficult and dangerous for humans to do on an inhospitable alien world.

  • The Boring Company’s tunneling tech will carve out underground shelters to shield colonists from radiation, killer dust storms, and wild temperature swings.

  • xAI will provide the superintelligence to tie everything together—solving the hardest problems as they arise in an instant and running the whole operation.

In other words, every piece of Musk’s empire is either writing the checks, laying the infrastructure, or supplying the brains and brawn to get us to Mars and keep us alive once we’re there.

This story has always been a simplification. But it’s useful. It gives investors a framework for understanding why one person would spread himself across electric vehicles, rockets, robots, satellites, brain implants, tunnel boring, social media, and AI.

Colonizing Mars was the thread that tied everything together.

That thread is still there, but the tapestry has gotten much bigger and more complex, layered, and immediate. Understanding how it’s evolved is vital for growth investors because Musk’s long-term vision is directing the flow of hundreds of billions of dollars in capital today—and potentially trillions more over the next decade.

Where that capital goes, and why, will shape a big chunk of the world’s industrial activity in the coming years. Get the map right and you stand a much better chance of positioning your own money where the real value is being built.

Mars remains Musk’s North Star. It’s still the ultimate reason the whole empire exists—an insurance policy for human consciousness to protect against Earth’s next extinction level event.

But as the tapestry has evolved, so too have the practical plans and associated investment opportunities. The good news is that these opportunities are more immediate and more investable today.

AI supremacy and civilization-scale infrastructure have become critical near-term enablers. And the Moon has become the near-term proving ground and steppingstone.

The North Star hasn’t changed—but the roadmap has

As recently as January 2025, Musk posted on X that the Moon was “a distraction” and SpaceX was going “straight to Mars.”

Just over a year later, in February 2026, he publicly pivoted. SpaceX, he announced, had “already shifted focus to building a self-growing city on the Moon”—something that could be achieved in under 10 years, compared to 20+ for Mars colonization. And he pushed active Mars efforts out five to seven years.

Why the pivot?

One big reason is that the development of Starship—the enormous spacecraft designed to take astronauts and cargo to Mars—isn’t moving fast enough for a near-term trip to The Red Planet. Starship has had some successful test flights, but it has not yet reached orbit. And orbital refueling—critical for both lunar and Mars missions—remains undemonstrated as of today.

Pivoting to the moon should get Starship back on track because launch windows come around every couple weeks instead of every couple years like they do for Mars. And a much shorter travel distance allows faster iteration and problem fixing.

In other words, the Moon just makes sense as a practical proving ground for the technologies and logistics that Mars will eventually require.

Another reason for the pivot: raising capital. Now that SpaceX and xAI have merged and filed for an IPO that will seek to raise up to $75 billion, it’s a much more palpable investor pitch to say something like “we’ll generate massive revenue from Starlink subscribers, lunar project contracts, and orbital AI data centers” (more on that later) than something like “we’re spending billions on a Mars colony that won’t generate revenue for decades.”

Then there’s the practical operating reality that’s become clear: you can’t safely build a sustainable colony on Mars (or even a long-term Moon base) until you solve AI, energy abundance, robotics, and connectivity at a scale that works for an entire civilization.

So Mars hasn’t been abandoned. But it has gone from “the thing we’re doing next” to “the thing we’re building toward”—and this “building toward” is where big money will be made, regardless of when (or whether) the first settlement makes it to Mars.

Mapping Musk’s empire—how the pieces fit together

Viewing Musk’s ventures as a collection of separate entities is the wrong lens. He’s built a system. And the power isn’t in any single piece; it’s in how the pieces fit together and reinforce each other like parts of one big machine.

Each piece generates data, revenue, energy, connectivity, or hardware that feeds the others, creating self-reinforcing flywheels that accelerate innovation, lower costs, and scale faster than any standalone entity could.

Recent moves like Tesla’s $2 billion investment in xAI, xAI acquiring X (formerly Twitter), and SpaceX acquiring xAI have made these ties even tighter.

Think of the empire like a tech stack with multiple layers. Each layer solves specific problems and feeds the others—creating a vertically integrated civilization-scale infrastructure platform.

Tesla is like the autonomy, energy, and labor layers.

  • The vehicle and robotaxi side of the business is the near-term cash generator and autonomy platform. Tesla’s FSD (full-self-driving) tech—with more than 7 billion miles of real-world driving data powering the world’s most advanced autonomous driving system—isn’t just about selling rides without drivers. It’s also about providing the “brains” for other autonomous robots and systems that can operate anywhere and do anything a human can… and more.

  • The energy side of the business—solar, Powerwall, and Megapack (batteries)—is quietly becoming one of the most important pieces of the entire empire. You can’t run giant AI training clusters, charge fleets of autonomous robots, or power off-world bases without abundant reliable energy.

  • Tesla sold $430 million in Megapack batteries directly to xAI’s Colossus cluster in 2025. And its entire energy business generated revenue of $12.8 billion while growing 27% from the year before with a gross profit margin of about 30% (nearly double the gross margin of the car business). This is not some sort of side project. It’s foundational.

  • Optimus—Tesla’s autonomous humanoid robot powered by the same AI “brain” as FSD—is like the labor layer. This venture could ultimately dwarf everything else in terms of economic impact. An affordable general-purpose robot that can perform virtually any task you need (dangerous, repetitive, specialized, etc.) would transform the economics of manufacturing, construction, elder care, and off-world building.

    Goldman Sachs recently upped its humanoid robot addressable market estimate by 6X to $38 billion by 2035. And some analysts project over $4 trillion when you include household applications.

    A few weeks ago at Peter Diamandis’ Abundance Summit, Musk said Optimus Gen 3—the first version designed specifically for mass production—will be “by far the most advanced robot in the world” and begin production this summer. He’s acknowledged that production will be slow and painful at first, but Tesla’s $20 billion capex guidance for 2026 (more than double the prior year) will flow primarily to Optimus manufacturing, Cybercab production, and AI compute. So he’s putting serious money behind this.

SpaceX is like the logistics, connectivity, and intelligence layers.

  • Starship is the vehicle that can drive down launch costs so the Moon and Mars become practical.

  • Starlink is the cash engine that funds it—with a fast-growing subscriber base already over 10 million strong providing more than $10 billion in annual revenue for internet from space, and government contracts adding another $7 billion in 2026 alone—while also forming the connectivity layer that every other piece will eventually run on.

  • xAI, which is now part of SpaceX, is the intelligence layer providing the connective tissue that runs through everything. Its Colossus supercomputer is one of the world’s largest AI training clusters. Its Grok models have over 60 million monthly active users (including me) and a new $200 million Pentagon contract. X’s (formerly Twitter) 600 million monthly users generate real-time AI training data.

Neuralink, which has provided brain implants to at least five patients and received FDA Breakthrough Device Designation for both vision and speech restoration, is like the interface layer. It remains the longest-duration piece of the machine. But the bridge between human cognition and AI becomes more relevant as autonomous systems proliferate.

The Boring Company—relevant to urban transit and off-world tunneling—is like the infrastructure layer. It’s the lowest profile piece, but it’s already built the Vegas Loop and secured contracts in Nashville and Dubai.

I’ve hinted at the flywheels that tie the whole empire together, but to be a bit more explicit about some of that:

  • Tesla’s autonomous driving data helps train xAI’s Grok models. This makes its AI smarter, which improves Tesla’s autonomous driving and Optimus robot capabilities and generates even more useful data.

  • Tesla energy products (especially Megapack batteries) help power xAI’s data centers. This enables AI training for Grok models, which further enhances Tesla’s autonomy platform.

  • Tesla energy products also help power SpaceX operations. This helps Starlink and Starship scale and provides important real-world validation of these products in the process.

  • Starlink revenue helps fund Starship development. This makes space access cheaper and faster, which allows for even bigger Starlink constellations that generate more revenue.

  • Starlink provides Tesla vehicles with high-speed internet anywhere on Earth. This enables more reliable FSD/robotaxi operation in remote areas, faster over-the-air updates, and better infotainment, which makes Tesla vehicles more valuable to help drive sales.

This is just a taste of what’s going on under the hood of Elon’s empire. But you get the idea. We’re talking about deliberate, compounding flywheels. Success in one area (data, energy, revenue, compute, connectivity) directly juices others—turning the empire into a vertically integrated, self-accelerating engine.

And the most aggressive integration yet: the March 2026 announcement of “Terafab,” a joint Tesla-SpaceX semiconductor fabrication facility planned for the Austin area, with Intel as a partner. Terafab represents Musk’s realization that he needs to control chip supply itself to keep the whole system scaling.

This is worth digging into a bit, because Terafab isn’t just another factory. It reflects a core constraint that threatens to bottleneck everything Musk is building. Every satellite, every robot, every autonomous vehicle, every AI cluster runs on advanced chips—and global supply is already stretched thin.

Musk has been characteristically blunt about the problem: the choice is either hit the chip wall or build a fab. And if you’ve followed his career, you know the path he takes when faced with a supply chain constraint: He vertically integrates and removes that constraint by building what he needs himself.

The Terafab blueprint calls for modular production units, each capable of manufacturing millions of advanced chips per month, built in the US with Tesla’s manufacturing prowess on compressed timelines.

Traditional semiconductor fabs take four to five years from permitting to production. And while making chips is arguably the most difficult manufacturing there is, Musk has repeatedly shown he can compress that kind of schedule dramatically. Let’s hope he can do it again here because the demand justification is overwhelming: millions of Cybercabs, potentially tens of millions of Optimus robots over time, an expanding constellation of Starlink satellites, rapidly growing xAI training clusters.

And then there’s the project that makes all those numbers look modest by comparison…

See, Terafab isn’t just about Earth-based applications—it’s the supply chain for something far more audacious.

The orbital AI vision: why space will become the cheapest place to think

The newest and arguably most ambitious element of Elon’s endgame: a constellation of up to one million satellites forming a distributed orbital AI data center in low Earth orbit.

I know it sounds like science fiction. But the logic is straightforward.

AI data centers are enormous facilities that consume staggering amounts of electricity. A single large cluster can draw as much power as a small-to-medium city.

And the constraints these facilities face on Earth are becoming increasingly burdensome. I’m talking about things like permitting delays and other regulatory red tape, grid limitations, long wait times for supporting infrastructure and equipment, higher and higher cooling costs for denser and denser racks, and community opposition.

These bottlenecks are already slowing the pace of AI infrastructure expansion. As models grow larger and more compute-hungry and inference demand skyrockets, the problems only get worse.

Space changes that equation entirely. In orbit, solar arrays can generate 6X to 13X more power per unit than ground-based installations. And there’s no need for real estate, bulky and expensive cooling systems, or grid connections.

Yes, you have the radiation issue in space—which will fry electronics without extra protection—that you don’t have on Earth.

But you don’t need to spend a fortune hardening every chip in orbit for radiation, which would also significantly reduce performance by the way. Instead, you can build in redundancy and house lots of chips in larger radiation-shielding structures. That’s much cheaper and results in virtually no performance degradation.

Here’s what the architecture would look like: each satellite functions as a self-contained mini data center, equipped with advanced GPUs, CPUs, and interconnects optimized for AI workloads. They’ll “talk” to each other using laser beams (optical links) to constantly share data and work together like one giant distributed AI supercomputer in space.

Musk said he expects 80% of Terafab’s chip output to go to space, where SpaceX will do the AI computing the hyperscalers like Google and Microsoft currently do on Earth.

He argues that as SpaceX continues to drive down launch costs with Starship, running AI compute in orbit will become cheaper than doing it on the ground within two to three years. Whether you buy that timeline or not, the vision is extraordinary: SpaceX wouldn’t just be the company that launches things into space and provides satellite internet, it would own the compute layer of low-Earth orbit.

This was a big reason behind the SpaceX-xAI merger.

In 2026, capital expenditures across Musk’s empire are expected to approach $50 billion—and grow significantly from there as Terafab and the orbital AI data center visions begin to take shape.

The total addressable market across these verticals—EVs, energy storage, robotaxis, humanoid robots, AI infrastructure, satellite internet, space launch—conservatively combine to $3 to $4 trillion by 2030, with credible projections reaching $10 trillion or more by 2035.

Now, I realize that the natural reaction to all this is skepticism.

Orbital AI data centers… a million satellites… humanoid robots at scale… semiconductor fabs built at Tesla speed… a self-sustaining city on the Moon… any one of these would be the most ambitious industrial undertaking of the decade. Musk is pursuing all of them simultaneously.

For literally anyone else on the planet, this plan would be pure fantasy. But before you dismiss it outright, consider what he’s already accomplished—because his track record is the single best argument for taking the next chapter seriously.

Musk took a niche electric vehicle startup that most of Wall Street expected to go bankrupt and turned it into a global technology powerhouse, which revolutionized manufacturing and autonomous driving technology, worth $1.5 trillion today.

At the same time, he rebuilt the economics of space from scratch. SpaceX developed reusable rockets when the entire aerospace establishment said it couldn’t be done, slashing launch costs by more than 95% in the process. In less than 20 years, the company went from a startup struggling to get a single rocket to orbit into the dominant launch provider on the planet. Today, SpaceX accounts for more than 90% of all the mass launched into orbit globally.

Then came Starlink. Musk deployed thousands of satellites to deliver affordable high-speed internet to the most remote corners of the world—mountaintops, open ocean, aircraft in flight, active war zones. The system cut the cost structure of satellite internet by an estimated 97% compared to legacy providers. It now serves over 10 million subscribers across more than 100 countries and generates over $10 billion in annual recurring revenue.

And layered across all of this is artificial intelligence. Remember, Musk co-founded OpenAI, the company behind ChatGPT. When he left, he built his own AI company from the ground up—xAI—and developed Grok into a frontier model. And he did it years faster than industry experts said was possible.

Today, AI isn’t just one piece of the empire. It’s the connective tissue running through everything: the autonomous driving system in every Tesla, the intelligence behind Optimus, the optimization engine for Starlink, and the reasoning core behind a consumer product (Grok) with 60 million active monthly users.

That’s the resume of the person telling you he’s going to build AI data centers in space and put a city on the Moon.

We can debate timelines. We can acknowledge the high execution risk. But I think it would be extremely unwise to dismiss the vision outright. To do so would mean betting against the most prolific builder of the 21st century. And that bet has been wrong, consistently, for over 20 years.

That’s Elon’s endgame as it stands today—and how the pieces connect.

In Part 2, we’ll get actionable and tackle the questions that matter most for your portfolio: Where does the capital flow, over what time horizons? And how do you position your investments around all this?

Thank you for reading. And stay tuned.

Related: Stock Dip Can’t Stop Tesla From Dominating Future Tech