On the eve of its IPO, 马斯克 unveiled a comprehensive strategy for orbital AI data centers, positioning them as the primary engine for SpaceX's future growth. The core objective is to circumvent the severe limitations of Earth's power grid on AI development by relocating massive computing infrastructure to low Earth orbit. In a video released on Monday, 马斯克 presented design sketches and technical specifications for the first-generation AI satellite, designated AI1. He asserted that constructing data centers in space relies on existing engineering capabilities rather than hypothetical technologies, noting that the technical hurdles are actually lower than those encountered in the current Starlink project. Data compiled by Woofun AI indicates that SpaceX aims to achieve an annual deployment rate of 1 gigawatt of space-based AI computing power by the end of 2027, signaling a strategic pivot to leverage satellite mass production advantages for next-generation AI infrastructure. Despite skepticism from industry competitors regarding economic viability, SpaceX has filed applications with the Federal Communications Commission to launch up to 1 million AI satellites, underscoring its commitment to this trajectory.

The fundamental driver behind this initiative is the physical constraint of terrestrial power generation. IPO application documents highlight that the global AI market, valued at up to $26.5 trillion, faces a critical bottleneck because Earth cannot rapidly expand its power generation capacity to meet demand. Consequently, solar-powered orbital data centers are viewed by 马斯克 and aerospace experts as the essential solution for scaling AI energy consumption. While 马斯克 set an ambitious target of reaching 1 terawatt of capacity through exponential growth, he advised investors to exercise caution regarding these aggressive timelines. Official IPO documents reflect a more conservative outlook, suggesting a gradual commercialization process beginning in 2028. This divergence between public ambition and regulatory filings highlights the complex risk profile associated with the venture.

Addressing misconceptions that space data centers merely involve transporting terrestrial facilities into orbit, SpaceX clarified that the hardware design focuses on power acquisition and thermal management in a vacuum. The AI1 satellite is engineered with a peak power consumption of 150 kilowatts and an average continuous computing load of 120 kilowatts. Woofun AI notes that these figures align closely with the power draw of NVIDIA GB300 compute racks containing 72 GPUs, effectively placing an entire high-performance AI module into space. To support these demands, the satellite features wings spanning 70 meters, a solar array with a generation density of 250 watts per square meter, and double-sided heat sinks capable of dissipating 1,400 watts per square meter. In orbit, the satellite will orient itself directly toward the sun to maximize thermal efficiency.

The hardware architecture of the AI1 satellite represents a significant simplification compared to traditional communication satellites. Unlike current Starlink satellites that require complex phased array and parabolic antennas, the AI satellite functions as a streamlined hardware component consisting primarily of solar panels, massive heat sinks, and basic laser communication links. This design eliminates the need for intricate ground-based communication antennas, reducing manufacturing complexity. 马斯克 and his engineering team emphasized that the production of AI satellites will extensively utilize technologies already proven in the Starlink V3 platform. This approach allows SpaceX to leverage its existing expertise in mass production, launch operations, and orbital management without necessitating breakthroughs in fundamental scientific research, creating a distinct competitive moat as the IPO approaches.

Network latency concerns are addressed through strategic orbital placement and advanced connectivity solutions. AI satellites will be deployed in low Earth orbit at altitudes ranging from 600 to 800 kilometers, resulting in a round-trip network latency of approximately 3 milliseconds. The constellation will be equipped with inter-satellite laser links offering bandwidths up to 1 terabit per second, utilizing SpaceX's existing KA and KU band antenna networks or direct ground-to-space laser links for high-speed data transmission.

However, significant disagreements persist within the industry regarding the commercial sustainability of this model. Blue Origin, Jeff Bezos of Amazon, and researcher Andrew McCalip have argued that the high costs of AI chips and launch services currently render the economic model unsustainable.

To counter these economic headwinds, SpaceX is pursuing a vertically integrated supply chain strategy designed to drastically reduce operational expenditures. The company plans to leverage the Starship heavy rocket to revolutionize launch economics while collaborating with partners such as Tesla and Intel to develop proprietary AI chips using the Terafab manufacturing facility. By controlling both the launch vehicle and the underlying computing hardware, SpaceX aims to accelerate the commercialization of space-based computing. Woofun AI analysis suggests that this vertical integration is critical for overcoming the cost barriers identified by critics and establishing a long-term strategic advantage over competitors who rely on fragmented supply chains.