Elon Musk Sets His Sights on Orbital Data Centers!

Last Friday, when SpaceX submitted plans to the Federal Communications Commission (FCC) for a million-satellite data network, it might have seemed like Elon Musk was joking. However, just a week later, it is evident that he is quite serious.

The most significant step is the formal merger between SpaceX and xAI, which was finalized on Monday. This merger aligns Musk’s space and AI initiatives, suggesting that a joint infrastructure project may be on the horizon.

Beyond the merger, the concept of orbital AI data clusters—essentially networks of computers operating in space—is beginning to solidify into a concrete plan. On Wednesday, the FCC accepted the filing and initiated a public comment schedule. While this is a standard procedure, FCC Chairman Brendan Carr took the unusual step of sharing the filing on social media. Throughout his tenure, Carr has demonstrated a willingness to support allies and challenge adversaries, indicating that as long as Musk maintains favorable relations with influential figures, the proposal is likely to progress smoothly.

In tandem with this, Elon Musk has begun to advocate for orbital data centers publicly. In a recent episode of the podcast “Cheeky Pint,” hosted by Patrick Collison, co-founder of Stripe, Musk articulated the rationale for relocating much of our AI computing capabilities to space. His argument centers on the efficiency of solar panels, which generate more power in space, thereby potentially reducing one of the primary operational costs associated with data centers.

“Scaling is more challenging on Earth than in space,” Musk mentioned during the podcast. “Any solar panel will produce about five times more power in space than on the ground, making it significantly more cost-effective.”

However, attentive listeners may recognize a flaw in this reasoning. While solar panels do yield more energy in space, energy is not the sole expense linked to operating a data center, and there are multiple energy sources available, as Patel pointed out during the discussion. He also expressed concerns regarding the maintenance of GPUs that may malfunction during AI model training, but for further details, I recommend listening to the complete episode.

Despite these challenges, Musk remained optimistic, identifying 2028 as a pivotal year for orbital data centers. “In 36 months, possibly closer to 30, the most economically viable location for AI will be in space,” he asserted.

Musk did not stop there; he also predicted, “In five years, we will launch and operate more AI in space each year than the cumulative total currently on Earth.”

For perspective, by 2030, global data center capacity is expected to reach 200 GW, representing approximately a trillion dollars’ worth of infrastructure when installed on Earth.

It’s worth noting that SpaceX generates revenue by launching payloads into orbit, making this scenario particularly advantageous for Musk—especially with the recent acquisition of an AI company. As the new SpaceX-xAI conglomerate approaches its IPO in the coming months, we can anticipate increased discourse surrounding orbital data centers. With technology firms continuously investing substantial sums into data center infrastructure each year, it’s highly likely that some of that capital will find its way beyond our planet.