Recently, a new application submitted by SpaceX to the Federal Communications Commission (FCC) was exposed, and then exploded in the technology circle

SpaceX is planning to build a giant satellite network called the Orbital Data Center System — it intends to launch and operate up to 1 million satellites deployed on multiple orbital layers 500 kilometers to 2,000 kilometers above the ground, according to the documents.

Note that this is not more broadband satellites, but moving data centers to the sky and using space computing power to run AI reasoning.

|SpaceX wants to build a "space AI brain"

The scale of this plan far exceeds the existing "Starlink" (Starlink) - Starlink currently has about 7,000 satellites, and this time SpaceX applied to deploy up to 1 million in one go!

This figure immediately attracted widespread attention in the market: where does the money come from for such a huge plan? How does supervision work? Can the technology be implemented?

In the eight-page document, SpaceX calls the new system an "orbital data center" and gives a clear goal: "This is a network of satellites with unprecedented computing power dedicated to running advanced artificial intelligence (AI) models and related applications."

Why do you need so many satellites?

SpaceX explained: To provide AI services to billions of people around the world (such as real-time translation, autonomous driving, large model reasoning, etc.), ground data centers are almost "unable to hold on" - they face bottlenecks such as power shortages, heat dissipation difficulties, and expensive land.

Space has natural advantages: solar energy is almost infinite and can continuously supply satellites; There is no air resistance, and heat dissipation is easier; Through laser communication between satellites, computing power can be interconnected at high speed and seamlessly connected to the existing Starlink network to quickly transmit the results back to Earth.

The document also specifically mentions that the satellites will operate in a narrow orbital band no more than 50 kilometers wide, with the aim of avoiding "collisions" with other companies' space projects and leaving a safe space.

Finally, SpaceX asserted: "Moving data centers to the sky is the most efficient way to meet future AI computing power needs." ——This is not only to expand the network, but also to create an "AI brain" around the earth.

It's not a cake, but there is a "hard support"

This seemingly crazy million-satellite program is not a fantasy, but is built on SpaceX's two core capabilities: Starship. To send 1 million satellites into the sky, they must be launched frequently and at low cost.

The "Starship" that SpaceX is building is the key: it has a large capacity and low cost, and is currently the only rocket that can support such a large deployment mission.

The capital market burns money on it. According to media reports, SpaceX is preparing for an initial public offering (IPO), one of the important purposes of which is to raise huge funds for this "space data center" project.

To put it simply: only with a starship can it be sent; Only with an IPO can you afford to burn money. These two legs support SpaceX's "orbital AI dream".

|Regulatory problem: 15,000 has just been approved, how easy is it to talk about 1 million?

On January 10, the U.S. Federal Communications Commission (FCC) just approved SpaceX to launch another 7,500 second-generation Starlink satellites, bringing the total number of second-generation satellites approved to 15,000 - mainly used to achieve "mobile phone direct connection to satellites" and "gigabit Internet access".

But compared with the 1 million applied for this time, 15,000 is simply a "fraction". The order of magnitude difference is nearly 70 times, and the difficulty of supervision has also increased exponentially.

The FCC has previously put forward strict requirements for SpaceX: half of the deployment of second-generation satellites must be completed by 2028; At the same time, it is necessary to seriously deal with risks such as space debris and orbital safety.

Now, if millions of satellites are really stuffed into the fortress, the problem will be even more serious:

Will the track be too crowded?

How to control the risk of collision?

Will other countries (such as China, the European Union) strongly oppose it?

It is widely believed in the industry that a plan of this magnitude will face extremely severe scrutiny and may even lead to international disputes.

However, SpaceX also revealed a little strategy in the application document: "We will customize different versions of satellite hardware for different orbital layers to optimize operational efficiency." To put it bluntly, they know that the challenge is huge and have begun to find ways to "manage in layers". But whether it can pass the supervision is still the biggest suspense.

Don't chase the concept, keep an eye on the "cashable" link

Opportunities related to SpaceX (e.g., potential IPO, supply chain):

Short-term speculation needs to be cautious: 1 million is a long-term vision, not a short-term executable plan; Real beneficiaries: Starship industry chain (such as engines, composite materials), laser communication module suppliers; If the IPO lands, the initial valuation may be high, but it is necessary to pay attention to the actual deployment rhythm and regulatory progress.

For AI and cloud computing tracks:

Ground data centers still face bottlenecks in energy consumption, land, and heat dissipation. "Edge + low Earth orbit" hybrid computing power architecture or long-term direction; You can pay attention to enterprises in the intersection of low-orbit satellite communication + AI edge computing.

For A-share/Hong Kong stock investors:

Domestic commercial aerospace is still in its early stages, but there is policy support in satellite manufacturing, measurement and control, laser communication and other links; In the long run, if the "orbital computing power" comes true, the space-ground integrated network will become the focus of new infrastructure.