The night before the listing, Musk detailed the "space data center plan": This is not very difficult for SpaceX.

```

On the eve of SpaceX’s IPO, Elon Musk for the first time disclosed in detail the company’s plans for orbital AI data centers, positioning them as the core growth engine and aiming to break the restrictions on AI industry development imposed by Earth’s power supply by shifting massive computing power to low Earth orbit.

In a video released on Monday, Musk publicly revealed for the first time the design sketch and key technical parameters of the first-generation AI satellite, “AI1.”

Musk made it clear that establishing data centers in space does not require “magic” yet to be invented. The technological challenges are even less than those faced by the existing Starlink business, and for SpaceX, these are not extremely difficult engineering problems. The company will strive to achieve an annualized deployment rate of 1 gigawatt (GW) of space AI computing power by the end of 2027.

This latest statement sends a clear signal to the capital markets: SpaceX is attempting to turn its absolute advantage in satellite mass production and launches into a moat for next-generation AI computing infrastructure.

Although industry competitors remain cautious about the economic viability of space computing power, SpaceX has applied to the U.S. Federal Communications Commission (FCC) to launch up to one million AI satellites. To overcome the high barrier of computing and launch costs, SpaceX is advancing in-house chip factory projects with partners such as Tesla and Intel, and plans to use Starship heavy rockets to disrupt existing launch economics, thus opening a strategic gap with its competitors.

Aiming at Computing Power Bottlenecks and Huge Potential Markets

In its IPO application documents, SpaceX pointed out that the total potential AI market could reach as high as $26.5 trillion, but will be severely constrained by “Earth’s inability to rapidly expand power generation capacity.” For this reason, orbital AI data centers powered by solar energy are viewed by Musk and aerospace executives as the key technology to meet the growing energy demands of AI companies.

Regarding the construction timeline, Musk gave an ambitious prediction. He said SpaceX will strive to achieve an annualized deployment rate of 1 gigawatt (GW) of space AI computing power by the end of 2027, and seek to expand at an exponential rate each year, ultimately reaching a computing power scale of 1 terawatt (TW).

However, he also cautioned investors to “maintain reservations” about this aggressive timeline, while the official expectation in the IPO documents is more conservative, with commercialization gradually advancing from 2028 onward.

Revealing AI1 Satellite: Nvidia Cabinet in Orbit

To address misconceptions that space data centers are simply “launching ground-based server rooms into space,” SpaceX clarified the true hardware form. The core engineering challenge is not the physical migration of buildings, but rather acquiring power in a vacuum environment and efficiently radiating waste heat generated by high-power computing.

The video showed, for the first time, quantitative specifications of the AI1 computation satellite. The satellite’s peak power consumption is 150 kilowatts, and its sustained average computational power consumption is 120 kilowatts.

Musk stated that these specs precisely match the operational power envelope of an Nvidia GB300 compute rack (containing 72 GPUs) used in ground-based data centers, essentially launching an entire Nvidia AI computation module into space.

To meet the high power and heat dissipation demands, the AI1 satellite is equipped with a pair of giant wings, each with a wingspan of 70 meters. Solar array generation density is set at 250W/m², and bilateral cooling panel dissipation density reaches 1400 W/m². In orbit, the satellite will orient in a “blade” posture to the sun for maximum heat dissipation.

Simplified Architecture and Technology Reuse to Build Manufacturing Moat

In terms of hardware structure, the engineering design of the AI1 satellite is even more streamlined than traditional Starlink satellites.

Existing Starlink satellites need to be equipped with extremely complex giant phased-array antennas and parabolic antennas. The AI satellite, by contrast, is essentially a purely large hardware apparatus: mainly composed of large solar arrays, huge radiator panels, and basic laser links, omitting the complex ground communication antennas.

Musk and the engineering team emphasized that the vast majority of the AI satellite’s manufacturing directly reuses SpaceX’s already developed and validated Starlink V3 satellite platform technology.

This means SpaceX does not need breakthroughs at the basic science level to directly transfer its existing experience in satellite mass production, launch, and operation. As the IPO approaches, this high level of technology reuse and engineering scalability forms the company’s unique competitive advantage presented to investors.

Ecosystem Synergy Tackles Cost and Latency Challenges

SpaceX has provided clear solutions to concerns about network latency in space data centers.

AI satellites will be deployed in low Earth orbit (LEO), 600 to 800 km above the ground, with one-way network latency of only about 3 milliseconds. Satellites will integrate inter-satellite laser links with up to 1 terabit per second (Tbps) bandwidth, and will use Starlink’s existing KA and KU band antenna networks, or directly use laser uplinks and downlinks for high-speed data transmission to the ground.

However, there are still internal industry disagreements regarding commercial viability.

Blue Origin, Amazon founder Jeff Bezos, and researchers such as Andrew McCalip have pointed out that expensive AI chips and high launch costs are currently industry obstacles, and the present economic model is not reasonable.

To this end, SpaceX is attempting to build a vertically integrated supply chain to break cost barriers: on one hand, leveraging Starship heavy rockets to sharply reduce launch costs; on the other, developing its own AI chips in collaboration with Tesla and Intel at a planned factory called Terafab. By controlling both launches and the underlying computing hardware, SpaceX is accelerating the commercial rollout of space-based computing.

Risk Warning and DisclaimerThe market has risks; investment needs caution. This article does not constitute personal investment advice and does not take into account individual users’ specific investment goals, financial situation, or needs. Users should consider whether any opinions, views, or conclusions in this article suit their specific circumstances. Any investments made based on this article are at your own risk. ```