Musk carries the xAI rocket on his shoulders, the picture is generated by AI

Author: Su Yang

"Use rockets to help AI finance."Hu Yong, a professor at Peking University School of Journalism and Communication, commented on the merger of xAI and SpaceX, "Basically telling the story of capital, I call it 'science fiction financialization'".

On February 3, 2026, xAI and SpaceX completed their merger at lightning speed, less than a week after relevant rumors. The combined entity was valued at US$1.25 trillion, and the narrative of space computing power completed a closed loop.

Foreign media quoted sources familiar with the matter as saying that the transaction amount was US$250 billion.

Before the merger rumors, the most discussed topic in the market was the listing of SpaceX in 2026 - a planned IPO financing of up to US$50 billion, with a valuation of approximately US$1.5 trillion.

"A pure 'rocket' + 'satellite' company is very expensive." said Lu Yi, a partner at Brilliant Phoenix Al, a private equity firm that has invested in SpaceX, Neuralink, Anthropic and xAI.

Lu Yi believes that this merger is more sexy and imaginative in terms of narrative.

"Join xAI andGrok, it immediately became a super story of "space + AI + global communication + universe understanding"

As the two most important private companies owned by Musk, xAI is responsible for the artificial intelligence business, and SpaceX has planned the vision of immigrating to Mars. Both companies have ambitious goals of exploring the future of mankind. However, at this stage, they are both facing the problem that a single narrative is unsustainable.

In the fierce competition between the artificial intelligence market and unicorns such as OpenAI, Anthropic, and giants such as Google, xAI needs a steady stream of capital transfusions, Acquire computing power to maintain training iterations of advanced models. However, due to this fierce competition, xAI is difficult to continuously obtain funding supplements from the primary market.

xAI's difficulty in financing is most obviously reflected in OpenAI - currently OpenAI plans to attract US$100 billion in investment from giants such as Microsoft, SoftBank, NVIDIA, and Amazon with a valuation of 830 billion. In order to accelerate the implementation of financing, it has even released news of an IPO.

For SpaceX, Starlink's revenue has stabilized. Reuters quoted two people familiar with the matter as saying that Elon Musk's SpaceX achieved revenue of 15 billion to 16 billion US dollars last year. EBITDA profit is approximately US$8 billion,But the key problem is the inability to obtain incremental orders and the poor progress in immigration to Mars.

Combining the two businesses of xAI and SpaceX achieves a "dilemma solution" - xAI relies on the SpaceX platform to obtain continuous blood transfusion, and SpaceX can rely on xAI's "space computing power" territory to obtain incremental commercial launch orders.

Lu Yi specifically added that SpaceX’s acquisition of xAI, since both companies are privately controlled by Musk and unlisted, has the least resistance and the smoothest integration.

Many industry insiders emphasized that blood transfusion for xAI is a key goal of the merger.

"Solving the extreme money-burning problem of xAI should be the most direct and urgent motivation," Lu Yi said, "xAI consumes approximately US$1 billion in cash per month (training Grok, building Colossus supercomputer clusters, etc.), and although its valuation has soared to US$200-230 billion, it is still essentially a company in the pure cash-burning stage."

Lu Yi emphasized that SpaceX has healthy cash flow (Starlink has begun to make large profits + stable launch business) and has a higher valuation (800 billion to 1.5 trillion US dollars), becoming a natural "blood transfusion platform".

However, Musk deliberately avoided the logic of using SpaceX to provide blood transfusion for xAI in his open letter, instead emphasizing the unlimited and sustainable space computing power.

"The current development of artificial intelligence is highly dependent on large-scale ground-based data centers. If we rely solely on ground-based solutions, the power demand of the global artificial intelligence industry will not be met at all. Forcible advancement will bring a heavy burden to human society and the natural environment. In the long run, space-based artificial intelligence is obviously the only way to achieve large-scale development." Musk said.

Although he avoided xAI blood transfusion, Musk did not shy away from highlighting in the open letterThe demand for massive "space computing power" empowers SpaceX's business.

"If 1 million tons of satellites are launched every year, and each ton of satellites can provide 100 kilowatts of computing power." Musk believes thatThe number of satellites required for space-based data centers will be even larger, which will push "starships" to a higher technological level.

According to public data, Falcon 9 has a low-Earth orbit transport capacity of 17.5 tons based on recycling, with a single quote of about US$70 million.The annual launch demand of 1 million tons is expected to bring an increase of US$4 trillion in revenue to SpaceX (estimated data for reference).

The incremental revenue of US$4 trillion is expected to accelerate the listing process of SpaceX with a valuation of US$1.5 trillion. "It is a familiar capital market operation." Lu Yi emphasized.

As a typical capital operation, this merger is equivalent to the strongest cash cow + launch capability to secure the most expensive but most imaginative AI brain for the future, while creating an epic narrative for the SpaceX IPO. However, "space computing power" engineering is the primary issue that needs to be resolved before this huge new story can be implemented.

In space,The vacuum environment has no convection and conduction and relies only on thermal radiation to dissipate heat. This is equivalent to "barbecue mode" for satellites on the sun-facing side. Coupled withspace rays and high-energy particles, it is possible to disrupt the high-performance computing process.

"If the computing chip you use on the ground is in the sky, it will most likely not work." A researcher in the aerospace industry previously emphasized.

As for heat dissipation, although it is possible to use the low temperature of deep space to alleviate it, this also raises a problem - the distance is too far, and another more efficient energy source, such as nuclear energy, is needed to replace solar energy.

As for the radiation problem, although it can be achieved through anti-radiation reinforcement, this adds additional technical challenges and costs beyond conventional GPUs.

"Radiation-resistant reinforcement, the materials and processes themselves are very expensive, and there is an upper limit for further reinforcement. It is not as good as the radiation-resistant effect of the atmosphere." said the aforementioned researcher.

In addition to heat dissipation and cosmic rays, the first issue that promoters of the "space computing power" narrative have to consider is rocket explosions and accidents - the costs of such small-probability events will far exceed the value of increased transportation capacity. After all, 1GW of computing power on earth is as high as 50 billion US dollars.

In other words, originally"space computing power" was an economical solution with unlimited resources and low cost. However, in the face of engineering and technical challenges and launch risks, the uncertainty of this economical solution has been increased.