Elon Musk has outlined an aggressive financial roadmap for SpaceX, projecting a staggering $1 trillion in revenue by 2030. This ambitious target relies on a massive expansion of the Starlink satellite internet service and a heavy pivot into artificial intelligence infrastructure. The strategy marks a significant evolution for the aerospace manufacturer, positioning it as a dominant force in global telecommunications and AI compute.
To reach these financial heights, SpaceX expects its high-speed internet division to generate $200 billion annually. The company plans to scale its Starlink customer base from the current 12 million users to 400 million. While the service is currently adding 1 million new users per month, projections indicate this will accelerate to 1.5 million monthly additions within the next two to four months.
The growth curve requires exponential scaling to hit these targets. The company must double its annual user acquisition repeatedly - from 18 million users per year currently, to 36 million, then 72 million, and ultimately reaching 144 million additions per year. Additionally, the direct-to-cellphone satellite service is targeting an even larger demographic of 1 to 2 billion customers, which is expected to bring in another $200 billion in annual revenue.
Beyond telecommunications, the company is rapidly building out its SPACEXAI infrastructure, which spans five distinct data center environments. Three of these are massive terrestrial sites, while the fourth leverages distributed AI computing in partnership with Tesla. The fifth, and most futuristic, involves deploying AI data centers directly in space.
The terrestrial AI operations are already generating a $26 billion annual runrate. The core of this operation is the Colossus cluster located across Memphis, Tennessee, and Southaven, Mississippi. Colossus 1 is currently operational with approximately 230,000 GPUs leased to Anthropic, consuming between 250 and 500 megawatts (MW) of power.
SpaceX is targeting 1 to 1.3 gigawatts (GW) for Colossus 2 to support the ramp-up of Blackwell and GB300 architectures. Renting out the entirety of Colossus 2 could generate up to $65 billion per year. A third facility, dubbed MACROHARDRR, will contribute an additional 500 MW, bringing the combined near-term capacity to 2 GW and paving the way for a 1-million GPU cluster by the end of 2027.
Powering this infrastructure requires massive energy investments, with the company planning to add 2 to 4 GW of capacity annually, scaling from 2 GW today to between 10 and 20 GW by 2030. To meet immediate needs, SpaceX has ordered $2 to $3 billion worth of mobile gas turbines over the next three years. This includes several 388 MW units from South Korea, with two arriving this year and another ten expected over the next few years.
The Orbital Compute Gamble
While scaling Starlink to 400 million users is a monumental logistical challenge, the true wildcard in Musk's $1 trillion projection is the deployment of AI data centers in space. Terrestrial data centers are currently bottlenecked by grid power availability and thermal management. Moving compute to orbit theoretically solves the cooling problem via the vacuum of space, but it introduces unprecedented challenges in power generation and data latency.
If SpaceX can successfully integrate its heavy-lift Starship capabilities with orbital server farms, it will bypass the terrestrial energy grid entirely. This would transition the company from a traditional aerospace and telecom provider into the foundational backbone of the global AI economy. However, relying on $3 billion in mobile gas turbines for near-term terrestrial power highlights just how energy-starved the AI industry remains while waiting for these futuristic solutions to materialize.
