TeXla: Why Tesla, SpaceX, and xAI Could Merge Into One Company

Elon Musk effectively operates three strategic stacks that increasingly want to be one: vehicles/robots (Tesla), launch + satellite infrastructure (SpaceX/Starlink), and frontier model development (xAI). The corporate boundaries made sense historically. The product roadmap no longer respects them.

If these stacks were consolidated into a single entity - call it TeXla - the upside case is straightforward: tighter data/compute integration, faster robotics iteration, and a private infrastructure layer (Starlink + launch) that can scale deployments globally. The downside case is also straightforward: execution, regulation, and competition (especially China) compress the dream into a very expensive science project.

Let me explain the logic, the drivers, and what could break.

Why They Can't Stay Separate

Musk runs all three companies. SpaceX (~42% equity, ~79% voting control), Tesla (~13% equity, ~20% voting control), and xAI (majority owner). They grew up as separate companies because he started them at different times: SpaceX in 2002, Tesla in 2003, xAI in 2023. But they've stopped being separate businesses in any meaningful way. The corporate structure is a relic, like a fax machine in a room full of iPhones.

Here's the problem with keeping them apart:

Tesla needs xAI's brain. Tesla is building Optimus, a humanoid robot that Musk says could be worth more than every other Tesla business combined. But a humanoid robot is really just hardware wrapped around an AI brain. If that brain comes from a separate company (xAI), Tesla loses the vertical integration advantage that makes it Tesla. Tesla already put $2 billion into xAI. If xAI is building the brain for Tesla's robots, why are these separate companies?

xAI needs Tesla's data and SpaceX's infrastructure. Training frontier AI models requires insane amounts of data and compute power. Tesla has the largest real-world sensor network on earth (5 million vehicles collecting driving data every day). SpaceX has Starlink, a global satellite network that could become the backbone for distributing AI. Keeping them separate means xAI has to negotiate arms-length deals for resources its sister companies already control.

SpaceX needs revenue diversification. SpaceX makes money launching rockets and selling satellite internet. Great business, but limited ceiling. If SpaceX merges with Tesla and xAI, Starlink becomes the infrastructure layer for a global AI and robotics empire. You're not a rocket company anymore. You're the logistics backbone of the automation economy.

These companies need each other to win their respective markets, but corporate boundaries prevent full integration. The merger removes the friction.

Where The Pieces Are Today

Before I project where this goes, here's what we're working with right now.

50% EBITDA margins on $15.5 billion in revenue. SpaceX prints money. It might actually be the most profitable company per employee on the planet, but we can't confirm that because Musk doesn't have to tell us. Perks of being private.

SpaceX bought a company that's still deep in the red. But 38x revenue growth tells you where this is going. The losses are infrastructure spend, not operational failure. Big difference. Amazon lost money for years too, and people who understood that retired early.

Tesla had a rough year. Net income got cut almost in half. The New Model Y refresh, Cybertruck production scaling, and price cuts all hit margins. Wall Street threw a tantrum. But the stock still trades at 353x earnings because nobody is buying Tesla for 2025 earnings. They're buying for Optimus, for full self-driving, for the energy business. The car stuff is the opening band, not the headliner.

The Combined Entity: TeXla

Entity	Revenue	Net Income	Valuation
Tesla	$94.8B	$3.8B	$1.34T
SpaceX	~$15.5B	~$2B*	$1.0T
xAI	~$3.8B	~-$5.5B	$250B
TeXla	$114.1B	$0.3B	$2.59T

*SpaceX reported ~$8B EBITDA. Net income after depreciation on rockets, satellites, and Starship development is significantly lower. I'm estimating ~$2B but SpaceX is private and doesn't disclose GAAP net income. Convenient.

That combined P/E is meaningless because xAI's losses eat almost all of Tesla and SpaceX's earnings. The market isn't pricing current net income. It's pricing what these businesses look like in 3-5 years when xAI flips profitable and Optimus starts shipping. If you're the type of investor who needs a P/E ratio that makes sense today, this isn't for you. Go buy Coca-Cola.

Where The Pieces Fit

Tesla + xAI = The Robot Play

Tesla's Optimus Gen 3 production starts before the end of 2026. Planned capacity: 1 million units per year.

Goldman Sachs says the humanoid robot market hits $38B by 2035. Their bull case is $5 trillion total addressable market. I think even Goldman is being too conservative because they're thinking linearly. Which is kind of Goldman's whole thing, but I digress.

A general-purpose humanoid robot that can work in your warehouse, your kitchen, your construction site, that's not a product category. That's the entire labor economy. The global labor market is roughly $90 trillion a year. If robots capture even 5% of that, you're looking at $4.5 trillion annually. And robots don't sleep, don't get hurt, don't quit, don't need health insurance, and every unit learns from every other unit simultaneously. Your HR department just got a lot simpler.

At mass production, each Optimus probably costs under $20,000 to build. Early units will sell around $30,000, but prices will come down fast as Chinese competitors ramp up and Tesla scales production. The real margin isn't in the hardware anyway. It's the intelligence layer: leasing models, robot-hours, AI subscriptions on top of the physical unit. That's where the software-like margins live. Sell the razor, print money on the blades. Oldest trick in the book.

But here's what makes the merger essential: the one variable I think most people are getting wrong is China. Unitree, Fourier Intelligence, UBTECH. They're moving fast and they're cheap. The Unitree G1 shipped at $16,000. If Chinese companies can build something 80% as good at half the price, Tesla can't win on hardware cost alone. The moat is the AI brain. Whoever has the best AI wins. That's why combining Tesla with xAI isn't some nice-to-have synergy. It's existential. Hardware will commoditize. It always does. The intelligence layer won't. At least not yet. And that intelligence layer is xAI's Grok.

Starlink = The Infrastructure Backbone

If you have 10 million humanoid robots deployed globally by the early 2030s, you need a private network to coordinate them, push software updates, and collect data. You cannot rely on third-party internet service providers for that. Starlink is the backbone.

Orbital Compute

This is the angle I think the market is sleeping on completely.

OpenAI, Google, Anthropic, Meta, they all have the same problem. They need insane amounts of electricity to train and run AI models, and Earth is running out of good places to put data centers. Virginia's grid is strained. Ireland is rejecting new builds. The energy bottleneck for AI is real and getting worse. Turns out, "we need all the electricity" is a hard pitch to local governments that also need to keep the lights on for regular people.

SpaceX can put compute in orbit.

Solar power in space is basically unlimited. No weather, no night cycle if you position the orbit right, no dealing with local utilities or grid politics. Cooling is free courtesy of the vacuum of space. Nature's server room, if you will.

What makes this actually viable instead of just a cool idea: Starship is reusable. Full, rapid reusability. That brings launch costs down toward $10-20 per kilogram to orbit. Nobody else can do that. Not ULA, not Arianespace, not even Blue Origin (though Bezos is certainly trying, bless his heart). Amazon's Project Kuiper is years behind Starlink and doesn't have a reusable heavy-lift vehicle at all.

So every major AI company on Earth either becomes a SpaceX customer for orbital compute, or they stay on the ground and deal with energy constraints their competitors don't have. That's not a product. That's a platform monopoly.

SpaceX is already testing compute payloads on Starlink satellites. Going from test to dedicated AI compute constellation is a matter of when, not if.

The Moon

SpaceX shifted its near-term focus to the Moon. Artemis III is contracted, Starship is the lander. If you're building a permanent presence on the Moon, you need robots. A lot of them.

Robots to build habitats in vacuum, mine regolith for water and building materials, maintain life support in an environment that kills humans in minutes. Tesla's Optimus is the lunar labor force. Nobody's filing workers' comp claims on the Moon.

And once you're mining the Moon, you're talking about lunar mass drivers. Electromagnetic railguns that accelerate payloads along a track and launch them off the surface into orbit. No rocket fuel needed. The Moon has 1/6th Earth's gravity and no atmosphere, so this is physically feasible in a way it isn't on Earth. Mine raw materials, process them locally, shoot them into orbit at a tiny fraction of what it costs to launch from Earth's surface. That changes the economics of building things in space completely.

A successful manned Moon landing followed by real base construction reframes the entire company from "conglomerate" to "civilization-scale infrastructure company." Markets price those very differently.

The Numbers

Optimus at real scale: Tesla already builds 2 million cars a year. Robot manufacturing is cheaper per unit because robots are far smaller, use less material, don't need crash safety engineering or interior fitouts. By 2030 I think production hits 5M units a year. By 2035, 50M+.

At that scale, build cost drops well under $15K per unit. 50M units at $30K average annual value = $1.5 trillion in robot revenue alone. Margins of 40%+ because the intelligence layer is pure software with zero marginal cost.

Does 50 million robots sound insane? There are 3.5 billion workers on Earth. 50 million is 1.4% penetration. That's not the end of the story. That's the table of contents.

Orbital compute: Every AI company becomes a SpaceX customer because they can't get enough power on Earth. I think orbital compute revenue could hit $200-500B by 2035. For reference, AWS does about $100B today. Orbital compute has better margins and a much harder moat because the barrier to entry is literally "build a reusable rocket."

SpaceX milestones keep the growth premium alive. Every 2-3 years there's a new catalyst:

• 2026-2027: IPO + Tesla merger talks
• 2027-2028: Optimus Gen 4 mass production
• 2028-2030: Manned moon landing (Artemis III)
• 2030-2032: Orbital compute constellation goes live, lunar base construction
• 2032-2035: Lunar mining, mass driver construction, Optimus at tens of millions of units

Traditional valuation says high P/E ratios compress as companies mature. But what if the company never runs out of next acts? Each milestone unlocks something bigger. It's like watching someone climb a staircase that keeps building new steps above them.

The math:

• - Combined 2035 revenue: $4-5T
• - Net income margin 30% (software-heavy mix) = $1.2-1.5T
• - P/E 20-25x (still a growth company)
• - Market cap: $24-37.5T. Call it ~$30T.

Is it wild to say one company could be worth over half the current US stock market (~$55T)? Yeah. A little. But by 2035 total market cap will be much higher. And if humanoid robots are generating trillions in economic output that literally didn't exist before, that's new GDP. Not redistribution. The pie gets bigger.

What Could Break

I'm not pretending this is guaranteed. If investing were certain, we'd all be billionaires.

Scale: 50M robots a year by 2035 is aggressive. Global auto production is ~90M and took a century. Scaling any physical product to 50M units is a massive engineering and supply chain challenge. Murphy's Law charges rent.

Orbital compute: Latency, bandwidth, thermal management, radiation hardening. All solvable but each one costs money and time. Hinges on Starship reuse cadence. Space is unforgiving. It doesn't care about your investor presentation.

Materials: 50M robots a year means 50-100 GWh of batteries on top of vehicle production. Lithium, nickel, and cobalt are finite. You can't mine what isn't there.

China: The real threat. If Chinese manufacturers produce $10-15K humanoids that are 80% as good, Tesla's hardware margins collapse. The moat has to be the AI, not the metal. This is the keep-you-up-at-night risk.

Regulation: A $30T entity will attract attention no matter what. But the US government is more likely to see this entity as strategically essential for competing with China than to break it up. Washington can barely agree on a lunch order, but they can agree they don't want to lose to Beijing.

How This Actually Happens

A thesis is useless if you don't understand the mechanics.

SpaceX already acquired xAI. That's done. The next move is SpaceX going public, expected mid-2026 at a $1.5T+ valuation.

Once SpaceX is public, you have two public companies (Tesla and SpaceX-xAI) with the same controlling shareholder. At that point, a merger requires both sets of shareholders to vote on it. Musk doesn't control that vote unilaterally. He owns ~13% of Tesla and ~42% of SpaceX. If Tesla shareholders think the price is too high, they vote no. If SpaceX shareholders think the price is too low, they vote no.

The negotiation comes down to the exchange ratio: how many shares of the new entity does each shareholder get? Tesla is worth $1.34T. SpaceX (with xAI) will be worth $1.5T+. Splitting that fairly is the hard part, and both shareholder bases will have strong opinions.

Then there's regulatory approval. DOJ and FTC have to bless it. These aren't horizontal competitors (nobody's buying a rival), so the antitrust case is weaker. But the sheer scale will attract scrutiny.

Timeline estimate: SpaceX IPO mid-2026. Merger talks within 12 months after. Combined entity by late 2027 or early 2028.

Bottom Line

Musk controls three companies worth $2.6 trillion. They share technology, talent, and vision. He just proved he's willing to consolidate by buying xAI. Tesla is the obvious next move.

At that point you're not buying Tesla stock. You're buying shares in the most vertically integrated entity in human history. Electric vehicles, humanoid robots, batteries, solar, energy infrastructure, custom AI chips, lithium refining, rockets, satellites, orbital data centers, and the AI that runs all of it.

Conservative case: $5-8T by 2035. Exponential case: ~$30T. Worst case: ~$2.25T, roughly where you started.

I think the exponentials are real. Optimus isn't a product, it's a labor revolution. Orbital compute isn't a feature, it's an infrastructure monopoly backed by reusable rockets that nobody else has.

I know which side I'm on.