Musk is really good at calculating.

On February 2nd, SpaceX pulled off a major move.

Put simply, Elon Musk finally integrated his trump card. SpaceX officially acquired his AI company, xAI.

It's like Minister Musk welded the "brain" and the "muscle" together, forming a global private corporate behemoth worth a staggering $1.5 trillion.

Why did he do this? Musk has long seen a harsh reality: Developing AI on Earth is hitting a physical ceiling.

01

First of all, there aren't enough power outlets on Earth, which is Musk's biggest headache.

Think about it. All the world's AI giants are doing the same thing: scouring the globe for electricity. Sam Altman is begging everywhere to build nuclear power plants. Jeff Bezos, Mark Zuckerberg, and others are even considering resurrecting old nuclear reactors that have been abandoned for decades.

Why? Because today's AI is extremely "thirsty." It not only guzzles electricity but also requires water.

Building a computing center with tens of thousands or hundreds of thousands of graphics cards on Earth is now a "trilemma." The power grid can't handle it. If you suddenly draw dozens of gigawatts in one place, the entire power grid will collapse.

It's like when you're renovating your house and want to install 100 high - power air conditioners in the living room. The property management will surely stop you because the cables in the whole building are only so thick. There's a physical limit.

Then there's the issue of heat dissipation. The heat generated by running chips requires a massive amount of water for circulation cooling. This means competing with local residents for water resources. Many places are already protesting, saying, "Why should you guys in AI dry up our groundwater?"

Even if you have the money, land, and technology, the bureaucratic red tape of environmental approvals and land planning can drag on for three to five years, and you'll never get it done.

So, Musk has figured it out. Developing AI on Earth has hit both "sociological" and "physical" ceilings. Therefore, his idea is quite "radical":

Since Earth won't provide electricity, why not just plug in next to the sun?

In high orbits, the sun is available 24/7, with no clouds blocking it. The efficiency of solar panels is several times higher than on the ground. It's like welding AI directly to the sun's "generator."

Moreover, the temperature in space is close to absolute zero. Although heat dissipation relies on technology (radiation heat sinks), at least you don't have to worry about competing with urban residents for water, and no one will sue you in court for being too noisy or too hot.

To put it bluntly, Musk thinks developing AI on Earth is like running a chemical plant in a downtown area, full of constraints.

He wants to move this whole setup to the "uninhabited area." The formula he calculated in the statement actually tells everyone: As long as my Starship has enough carrying capacity, I can create countless "perpetual - motion machine computer rooms" in the sky.

So, this is the first underlying driving force for him to merge xAI: Since there's no electricity on the ground, go to the sky to get free solar power.

02

Now that the power - outlet logic is clear, what's the second reason?

Musk said that the Starship is his only and most powerful "hauler." He even calls it the "computing - power transport aircraft."

To send one million satellites into space, each carrying heavy - duty computing chips like NVIDIA's yet - to - be - released next - generation cards and huge solar panels, how heavy would that be?

According to Musk, without the Starship, AI can only compete on the ground. With the Starship, computing power can be "wholesaled by the ton."

You can imagine:

In the past, when the Falcon rockets sent satellites into space, it was all about "meticulous calculation." You had to calculate the weight for each satellite launch, and every launch was carefully planned.

The Starship V3, which officially entered service this year, is a monster. Its payload has jumped from the original 35 tons to over 100 tons. Musk also said that he wants to achieve "one launch per hour," with each launch carrying 200 tons. This means that SpaceX can send millions of tons of stuff into space in a year.

This has completely brought down the cost.

If you build a computer room on the ground, buying land, construction, and getting electricity connected can take several years for a project. But Musk thinks he can just use his Starship to scatter "computing - power satellites" into orbit like sprinkling pepper.

It is also estimated that within two to three years, space will become the place with the lowest cost for generating AI computing power. So, merging xAI into SpaceX is like putting the "brain" directly on the "truck chassis."

In the past, xAI had to beg NVIDIA for supplies and ask Altman for some interfaces to get computing power. Now, Musk's message is clear: I'll use my own Starship to move the computing power to the place closest to the power source.

Okay, we've got the "power outlet" and the "computing - power transport aircraft" sorted out. Here comes the third reason. Musk said that there's an even more subtle and harsh factor when developing AI on Earth: The scale ceiling.

You might think that as long as there's electricity and chips, AI models can grow infinitely large. But Musk doesn't think so. He sees two very harsh limitations.

The first is the "illegal construction" in physical space.

You know the rumored Grok - 5 with six trillion parameters. Do you know how crazy it is to train this thing? To train it on the ground, you'd have to build a super - computer room as big as a small city.

The problem is that this kind of "stacking" is almost reaching its limit on Earth. It's like trying to build a thousand - story skyscraper in a downtown area. The foundation can't support it, and the property management won't allow it.

Space is completely different. There's infinite space and no gravitational pressure. You can piece together tens of thousands of computing modules like Lego. Musk even wrote a formula in the statement, which means:

Intelligence ∝ Mass to Orbit × Compute Density.

Simply put, as long as the Starship can keep delivering goods into space, there's no upper limit to the scale of AI. The more you send, the greater the computing power, and in theory, it can expand infinitely.

The second is the "dimensionality reduction strike" in data dimensions.

This is where Musk is really shrewd and the core reason for merging xAI into SpaceX. In the past, xAI only had a "mouth," knowing what people around the world were saying through X (Twitter). This was social corpus. Now, it directly has "eyes" and "ears."

After merging into SpaceX, xAI is connected to Starlink. What is Starlink? It's hundreds of thousands of sensors around the world, real - time satellite images, and all kinds of global communication signals.

That is to say, Musk is integrating social data and physical sensor data. This scale is beyond the dreams of other AI companies.

OpenAI's data comes from static Internet documents, while Musk's AI can monitor every pixel on Earth and every signal in real - time. By merging these two companies, he wants AI to not just be an "armchair critic" but to have a "God's - eye view" of the global physical pulse.

So, Musk thinks developing AI on Earth is like learning to swim in a bathtub. No matter how big your movements are, there are limitations.

He wants to throw AI into the "vast expanse of the universe" and let it evolve by devouring real - time data from the entire Earth. He wants to create a real "omniscient orbital brain."

03

The fourth reason is more practical. Musk is doing this to find a "super money - printing machine" for his money - guzzling project and lock in a $1.5 - trillion, century - level IPO.

Although xAI is currently valued at $250 billion, it's essentially a "money - shredder." To compete with OpenAI and Google in algorithms and buy chips, it burns money at an astonishing rate every day.

In the past, when raising funds, it had to kowtow to venture - capital tycoons. You had to beg firms like Sequoia Capital and Andreessen Horowitz to invest.

Now that it's merged into SpaceX, the situation is completely different. SpaceX holds Starlink, which is a cash cow. Tens of millions of users around the world pay monthly subscriptions, and the cash flow is incredibly stable.

Merging xAI into SpaceX is like connecting a "powerful heart" that never stops to this "brain."

Moreover, have you noticed?

This acquisition also provides an "escape pod" for investors. In the past, investors in xAI might have been worried about when it would go public. Now, SpaceX plans to hold a $50 - billion - scale super IPO on June 2026, which is Musk's 55th birthday.

Wall Street is also interesting. The valuation of a pure rocket company has a ceiling, but if you add in the "future operating system for all of humanity," there's no upper limit to the valuation.

So, when you combine SpaceX (the muscle) and xAI (the brain), the $1.5 - trillion valuation makes sense. It's a super - digital colonizer that controls "space energy + computing power + transportation + data."

Peter Diamandis, the founder of the XPrize Foundation, put it very well:

Musk is breaking down corporate boundaries and piecing together Tesla, SpaceX, X, and xAI into a single entity. He even said that there might eventually be a real "Musk Corporation."

In the past, each company was separate. Tesla provided the money, SpaceX provided the vehicles, X provided the data, and xAI provided the brain, and they had to go through all kinds of cumbersome related - party transactions. Now, he's welded all these pieces together to form a "vertically integrated innovation engine."

To put it simply, Musk not only controls the AI software but also the hardware (satellites) that runs the software, the vehicles (Starship) that transport the hardware, and the energy (solar power) that supports it all.

To put it bluntly, this is the "ultimate form" of a company in his mind: A super - corporate empire that truly holds the brain, muscle, and energy in its own hands.

Okay, Musk is happy, but what about the investors and funds?

You should remember in 2016 when he used Tesla to acquire the money - losing SolarCity. He was severely criticized for "related - party transactions." Now, the merger of xAI and SpaceX is, in the eyes of many funds, the 2026 enhanced version of SolarCity.

xAI is valued at $250 billion, which seems impressive. But in the private - equity market, who would dare to take over this stake?

VCs like Sequoia Capital and a16z have invested huge amounts of money. If xAI can't outcompete OpenAI, this investment might take ten years to pay off.

Now that it's merged into SpaceX, the private - company options held by these funds have instantly become shares in a "quasi - listed behemoth." SpaceX's IPO in June this year is like a "super express pass" that Musk is giving to venture capitalists.

Also, look at Cathie Wood (ARK Fund) and Ron Baron (Baron Capital). They've long been singing the praises of SpaceX.

Now that xAI is merged in, not only do they hold rockets, but they also own the most powerful AI computing - power assets. This will directly drive up the net value of their funds before the IPO.

Even more interesting is Tesla. It invested $2 billion in xAI last month, and before the money had a chance to settle, xAI became part of SpaceX.

Retail investors in the secondary market don't really care about these controversies. They think that with the merger of Musk's "package," Tesla's AI dreams (FSD, Optimus) will have the support of space computing power.

From another perspective, Musk is playing with the "Musk Index." Wall Street is now saying: In the future, you don't need to buy the Nasdaq index. Just buy the "Musk Index."

To put it bluntly, he's doing three things:

Internal arbitrage. He's shuffling money between Tesla, xAI, and SpaceX. As long as the valuation grows faster than the money - burning rate, this bubble is sustainable.

Rejecting spin - offs. Starlink's cash flow feeds xAI, and xAI's potential boosts SpaceX's IPO.

Creating uniqueness. He's created the only integrated "land - sea - air - space - electricity + digital life" target in the capital market. If you want to invest in AI, space, or energy, you have to buy into it.

According to his statement on February 2nd, if these companies aren't integrated, humanity might not reach the "Kardashev Type II civilization" before 2030. Who among fund managers would dare to oppose this when he makes such a claim?

You can almost hear the clatter of Musk's abacus for this "asset unification" hundreds of kilometers away.

This article is from the WeChat official account "Wang Zhiyuan," written by Wang Zhiyuan and published by 36Kr with permission.