SpaceX to Build Super Satellite Factory: Mass Production of AI Computing Power Satellites Next Year, Targeting 100 Gigawatts by 2030
```
SpaceX is fully shifting its satellite mass production advantage toward the space-based AI computing power track.
Elon Musk's aerospace company SpaceX has announced plans to build a super satellite factory, "Gigasat," spanning 11 million square feet in Bastrop, Texas, dedicated to producing orbital AI data center satellites. The goal is to achieve an annualized deployment rate of 1 gigawatt (GW) of space AI computing power by the end of 2027, and to push this figure to 100 GW by 2030.
In an internal interview video released by Musk on the X platform on June 8, he publicly revealed the design sketch and core technical parameters of the first-generation AI satellite "AI1", and stated clearly that the technical challenges of building a data center in space are even lower than those faced by Starlink, saying "this is not an extremely difficult engineering problem for SpaceX."
Analysis points out that this statement sends a clear signal to the capital markets: SpaceX is attempting to convert its scale advantages in satellite production and launch into a core moat for next-generation AI infrastructure.
According to a WallstreetCN article, the above disclosure comes on the eve of SpaceX's IPO, with Musk positioning orbital AI data centers as the company's core growth engine, trying to break the constraints that Earth's electricity supply places on the development of the AI industry by moving massive computing power to low Earth orbit.
In its IPO filing, SpaceX stated that the estimated total potential market for AI, as high as $26.5 trillion, will be severely limited by "the inability for Earth to rapidly expand power generation capability." Orbital AI data centers powered by solar energy are seen as the key technical path to satisfy the increasing energy demands of AI companies.
Gigasat Factory: Ten Times Larger than the Biggest Existing Aerospace Manufacturing Facility
The Gigasat factory covers 1,000 acres, with a building area of 11 million square feet, more than ten times the size of SpaceX's current largest aerospace manufacturing complex, Starfactory. This facility will achieve highly vertically integrated AI1 satellite supply chains within a single campus, covering solar ingots and wafers, solar cells, printed circuit boards (PCBs), silicon-based electronic components, user terminals, ground gateways, and the entire manufacturing process for the AI1 satellite as a whole.
The campus will also feature satellite R&D and testing facilities, warehousing and logistics infrastructure, and large-scale AI satellite mass production lines. Musk revealed that solar manufacturing facilities are under construction, and the AI satellite production plant will soon break ground, with "substantial" mass production and shipments expected by the end of 2027.
AI1 Satellite: An NVIDIA Server Cabinet in Orbit
The AI1 satellite represents SpaceX's concrete implementation of the orbital data center concept. The satellite has a wingspan of about 70 meters and is equipped with large solar arrays, power density of 250W/m², peak computing power consumption of 150 kilowatts, and sustained average computing power consumption of 120 kilowatts.
According to a WallstreetCN article, Musk says this power consumption metric exactly matches the power envelope of the NVIDIA GB300 server cabinet used in ground-based data centers (containing 72 GPUs), which is equivalent to sending an entire NVIDIA AI compute module into space.
To tackle extremely high heat dissipation demands, the AI1 satellite uses vertically arranged dual-sided heat plates, with a dissipation density of 1,400W/m². In orbit, it will face the sun in a "blade" orientation to maximize cooling, with the computational payload deployed in the central structure.
In terms of architecture, the AI1 satellite is more streamlined than traditional Starlink satellites. Existing Starlink satellites need complex giant phased array and parabolic antennas, while the AI1 satellite is essentially a large hardware assemblage: massive solar arrays, ultra-large heat plates, and basic laser inter-satellite links, omitting complicated ground communication antenna systems.
Technical Reuse Creates Manufacturing Barriers
SpaceX's core competitive logic lies in technical reuse. Musk and the engineering team emphasize that most of the technology needed to manufacture the AI1 satellite is directly reused from SpaceX's already developed and verified Starlink V3 satellite platform. This means the company doesn’t need breakthroughs in basic science, but can transfer its mass satellite production, launch, and operational expertise directly to the AI satellite business.
To achieve 1 GW annual space AI computing power deployment by the end of 2027, SpaceX will need to launch more than 6,000 AI1 satellites in a single year. As reference, as of June 2026, Starlink has about 10,500 active satellites in orbit.
The Scale of the 100 GW Target: Equivalent to Building 20 Meta Super Data Centers a Year
Musk has set an ambitious expansion path for this plan: 1 GW annualized computing power by the end of 2027, expanding at orders-of-magnitude rates every year thereafter, 10 GW by 2029, 100 GW by 2030, with terawatt-level space computing power as the ultimate goal, provided chip manufacturing technology can keep pace.
What does 100 GW mean? The largest currently announced AI data center project is Meta’s Hyperion in Louisiana, with a designed maximum scale of 5 GW, over $100 billion investment, and its first 2 GW phase won't be completed until 2030.
xAI’s Colossus 2 in Memphis has just expanded to nearly 2 GW, equipped with 555,000 GPUs, costing about $18 billion, making it the largest single-site AI facility worldwide. 100 GW is about equivalent to building 20 Hyperions or 50 Colossus 2s each year.
Despite the massive scale, the commercial viability of space-based computing power remains debated within the industry.
According to a WallstreetCN article, Blue Origin, Amazon founder Jeff Bezos, and researcher Andrew McCalip among others point out that expensive AI chips and high launch costs form the main barriers at present, and current economic models are not yet feasible.
In response, SpaceX is attempting to break through cost barriers via vertically integrated supply chains: on one hand, relying on the Starship heavy rocket to sharply lower per-launch costs; on the other, advancing a so-called Terafab plan factory, collaborating with Tesla and Intel to develop and manufacture proprietary AI chips, targeting chip production sufficient for 1 terawatt of computing power annually in a 1 million square foot factory, about 100 million to 200 million advanced-process chips. However, the Terafab plan faces widespread skepticism—it’s aimed directly at 2nm advanced process nodes, while none of the three partners has chip manufacturing experience.
On latency, SpaceX has provided a clear technical solution: AI1 satellites will be deployed in low Earth orbit 600-800 km above ground, with one-way network latency of only about 3 milliseconds, and integrated inter-satellite laser links with bandwidth up to 1 terabit/second, enabling high-speed downlink via Starlink’s existing KA and KU band antenna network or space-to-ground laser link.
Musk himself also cautioned investors to "be skeptical" about the radical timeline, with SpaceX’s IPO filing giving an official projection that commercialization will advance gradually starting from 2028.
Risk Warnings and DisclaimerThe market has risks, and investment needs to be cautious. This article does not constitute personal investment advice, nor does it take into account the unique investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions herein fit their specific circumstances. Invest at your own risk. ```