Elon Musk's Ambitious Vision for Space AI

In recent months, Elon Musk has been making a flurry of moves that are truly dazzling.

In February, SpaceX officially announced its acquisition of xAI at a valuation of $1.25 trillion, bundling rockets, satellites, and AI models into an "interstellar tech behemoth." Immediately afterwards, Tesla announced the discontinuation of the iconic flagship models, the Model S and Model X, and directly transformed the Fremont factory into a production line for the Optimus robot. In interviews, he even dropped bombshells such as "space computing power will exceed the world's total," "the AI5 chip is crucial for Tesla's survival," and "Optimus will have a production capacity of over 10 million units in four years."

Musk's series of moves still make me can't help but say, "This isn't just running a company; it's clearly about rewriting the rules of the game."

When you connect the dots, you'll find that there's a complete set of logic behind this: to integrate the entire chain of "space energy - computing chips - physical terminals," liberate AI from the power bottlenecks and production limitations on the ground, and push it into space and the physical world.

The year 2026 is set to be a super year for tech IPOs. Both OpenAI and Anthropic are planning to go public in droves. Musk, however, has taken the lead in integrating assets and rewritten the valuation logic with the narrative of "space AI." This move not only saves the cash - burning xAI but also adds an "irreplaceable" aura to SpaceX's IPO.

More importantly, he is completely detaching Tesla from the label of an "automobile company" and betting on robots and autonomous driving, wagering on the next $25 - trillion market.

1. Space Is Not a Gimmick

Many people think that Musk's idea of "moving data centers into space" is like something out of a science - fiction novel, but those in the industry know that it's an inevitable choice out of necessity.

Currently, the biggest bottleneck in the global AI development is not chips but electricity.

Except for China, the electricity production in major countries around the world has basically stagnated, while the demand for computing power is growing exponentially. At this rate, by the end of this year, there will be an embarrassing situation where "chips are produced but can't be powered on": the chips in large - scale computing clusters can't be started and will only pile up in warehouses.

Musk has it all figured out: Ground - based computing power is limited by various factors such as electricity, land, and environmental protection, like trying to perform a ceremony in a snail's shell. Space, on the other hand, is "unclaimed territory." As long as the rocket's carrying capacity is sufficient, it can expand infinitely.

He plans to launch rockets 30,000 times a year, and in five years, the computing power in space will exceed the total on the ground worldwide.

2. A Narrative Reconstruction

The year 2026 will definitely be a super year for tech IPOs. Both OpenAI and Anthropic are planning to go public in large numbers. Musk, however, has taken the lead by having SpaceX acquire xAI, pushing the valuation to $1.25 trillion. The brilliance of this move lies in "attaching an AI label to traditional businesses," completely rewriting the narrative logic in the capital market.

For a simple rocket company or communication company, there is a clear ceiling on valuation. Both telecommunications operations and logistics transportation are traditional industries with low price - to - earnings ratios, which can't support high expectations. However, with the story of "space AI + orbital data centers," SpaceX instantly transforms from a "space logistics provider" into an "interstellar tech giant," and the valuation logic directly switches to the high - growth potential of the AI track.

Wall Street eats this up: When you talk to them about the profit from rocket launches, they talk to you about the trillion - dollar space computing power market; when you tell them that xAI burns $1 billion a month, they talk to you about the commercial value after the implementation of AGI.

This acquisition is essentially a dual operation of "financial rescue + narrative upgrade." On the one hand, xAI is a real "money - guzzler." It burned through $7.8 billion in cash in the first nine months of 2025, with a monthly burn rate of up to $1 billion. Meanwhile, SpaceX has become a "cash cow" thanks to its Starlink business, with a profit of about $8 billion in 2025, which can just provide a financial boost to xAI.

On the other hand, SpaceX originally planned to go public in June 2026, aiming for a valuation of $1.5 trillion and a fundraising of $50 billion. With the AGI story of xAI, this valuation becomes more convincing - after all, the combination of "space + AI" is unique in the capital market.

This isn't the first time Musk has pulled off such a move. In 2016, he had Tesla acquire the nearly bankrupt SolarCity and bundled the solar energy business into Tesla's energy division.

Now, SpaceX's cash flow supports xAI, and xAI's potential boosts SpaceX's IPO valuation. One has to admit that he really understands Wall Street's mindset.

3. Tesla's Discontinuation Is Not a Desperate Move

Some people are sad about the retirement of the Model S and Model X, thinking that Tesla has lost its original vision. In my opinion, however, this is Musk's most decisive strategic shift. Tesla has never been just a "car - manufacturing company." Car - manufacturing was just a "practice stage" for it to accumulate AI and manufacturing capabilities. Now, it's finally time to reap the rewards.

Let's first look at the current situation: The Model S and Model X have become a "burden." Why do I say so? In 2025, the total sales of the "other models" category, which includes these two models, were only 50,800 units, a year - on - year decline of 40.2%, accounting for less than 3% of Tesla's annual sales.

In 2025, Tesla's annual revenue decreased by 3% year - on - year, and its net profit plummeted by 46%. It also lost the title of the world's top - selling electric vehicle brand to BYD. Continuing to invest resources in the declining high - end models is clearly not cost - effective.

More importantly, Musk has his eyes on a bigger prize: the robot market.

He has long said that the future market value of robots is expected to exceed that of the automotive business. By 2050, the relevant market is estimated to be worth $25 trillion, accounting for 80% of Tesla's future overall value.

To seize this market, resources must be concentrated. The production lines at the Fremont factory that originally produced the Model S and Model X are being transformed into production lines for the Optimus robot, with a designed annual production capacity of 1 million units. This has a much higher input - output ratio than continuing to produce niche high - end cars.

Moreover, Tesla has enough confidence to make this transformation: As of the end of the fourth quarter of 2025, the company's total cash and investments reached $44.1 billion, which can support capital expenditures of over $20 billion in 2026.

More importantly, the autonomous driving technology, sensors, and supply - chain management experience that Tesla has accumulated in the process of car - manufacturing can be directly transferred to the Optimus robot, reducing the technical and cost barriers to mass - producing robots.

Musk put it bluntly: "In the end, no one will remember that Tesla used to make cars." What he wants is not the title of an automotive giant but the dominant position in the "AI + robot" ecosystem.

4. Optimus: The Allure and Pitfalls of a "Money - Making Perpetual Motion Machine"

Musk calls Optimus a "money - making perpetual motion machine." This statement is not an exaggeration, but only if the two major problems of "mass production + technology" can be solved.

The value of humanoid robots lies in their ability to translate the digital capabilities of AI into the physical world. In the past, AI could only work inside computers. With Optimus, it can enter factories and do household chores, which is like giving AI "hands and feet."

Musk says that currently, no other robot has hands as flexible as Optimus. In terms of production capacity planning, the goal is to reach a production capacity of 1 million units in three years and 10 million units in four years. Once the recursive growth of "robots manufacturing robots" is achieved, the cost will drop sharply, and the profitability will skyrocket.

However, Optimus still faces a host of problems: The dexterous hand has insufficient grip strength, the lifespan of the core joints is only one year (the industrial - grade requirement is more than five years), and the cost during the trial - production period is $40,000 - $50,000, which is far from the mass - production target of $20,000.

Moreover, China already dominates the global humanoid robot market. Of the 13,000 humanoid robots shipped globally in 2025, more than 80% were from Chinese companies. The shipment volumes of Unitree Robotics and Zhipu Robotics far exceeded that of Tesla. It will be extremely difficult for Optimus to catch up.

Musk himself doesn't shy away from it and bluntly says, "If the United States doesn't innovate, China will dominate AI and robots." This is not an alarmist statement. China's advantages in large - scale ground - based manufacturing have formed an insurmountable barrier.

5. Beyond the Survival Line: In - House Chip Development Is the Ultimate Moat

"If Tesla can't get the AI5 right, it's over."

The AI5 is not an ordinary in - vehicle chip. It is a "universal platform" that supports FSD, Robotaxi, Optimus, and data centers. Its inference performance is 40 times that of HW4, its memory is nine times larger, and it has removed the GPU and ISP, focusing on running end - to - end neural networks. Yet, its cost is much lower than that of NVIDIA.

To ensure production capacity, Musk has adopted a strategy of "dual - outsourcing + self - built factory": The AI5 is being produced simultaneously by TSMC (at its Arizona factory using the 3nm N3E process) and Samsung (at its Texas factory using the 3nm GAA process) to avoid putting all eggs in one basket. He has also secured all the available wafer - fab production capacity he can get.

However, he doesn't think this is enough. He plans to build the "TeraFab," integrating logic chips, memory, and packaging. The goal is to start with a monthly production of 100,000 wafers and eventually reach 1 million wafers, completely getting rid of the dependence on contract manufacturers.

More importantly, the AI5 is the key to "unifying the in - vehicle and robot brains." It not only needs to be installed in cars but also needs to be compatible with the Optimus robot. In the future, the AI6 chip will achieve "integration of training and inference," which can be used for both terminal inference and data - center training.

The AI7 is specifically designed for space scenarios, forming a comprehensive chip layout covering the ground, space, terminals, and the cloud.

This means that Tesla's chips not only have to compete with those of NVIDIA and AMD but also need to support the operation of the entire ecosystem. If it fails, the foundation of the entire empire will be shaken.

In conclusion, the core of all of Musk's operations is to build a self - cycling and self - strengthening ecological closed - loop: SpaceX is responsible for "space energy + computing power transportation," xAI is in charge of the "AI brain," Tesla is responsible for the "physical terminals (cars + robots)," and Starlink is responsible for "global communication," forming a complete chain from "obtaining energy from space, to AI processing data, and then to robots for practical applications."

I think Musk's ambition is worthy of respect, but we also need to be clear - eyed: Technological revolutions are never built on just stories. In the end, it all comes down to products, production capacity, and profitability.

Anyway, Musk has raised the level of technological competition from the ground to space. In the next few years, we'll witness an "interstellar - level" business war with our own eyes.

This article is from the WeChat official account "Intelligent Machine Island." Author: Huo Rujun. Republished by 36Kr with permission.