SpaceX: More X than Space in Its Mission Focus

I used to wonder why SpaceX, the crown jewel of Elon Musk’s ventures, decided to scoop up xAI, which has been struggling. Now, I’m left questioning why we’re still labeling the whole operation as SpaceX.

When you break it down by revenue, SpaceX is more of a telecom and cloud computing company than a space exploration firm, as highlighted in its first quarterly earnings report since going public. The space division didn’t even hit a billion dollars this quarter, contributing just over 10% to the overall revenue. In fact, SpaceX is still its own biggest customer, as there aren’t enough clients lining up for its rockets. If rockets were the main focus, perhaps SpaceX wouldn’t be at risk of creating a new crater on the Moon with its space debris.

The telecom side, dubbed “connectivity” by SpaceX, is primarily driven by Starlink, the satellite internet service that raked in $4.2 billion in revenue and was the only segment not operating at a loss. During the earnings call, Gwynne Shotwell discussed plans for a phone service to rival major carriers, but the bulk of the investment and excitement seems to be centered around the so-called “AI” initiatives, likely to highlight any contributions from Grok. Analyst Alexander Potter predicts that spending on the neocloud sector, which leases data center capacity to AI firms, will soar to $65 billion next year, a significant increase from earlier estimates.

It’s clear that leasing data center space is crucial for SpaceX. Not only does it generate more revenue than the rocket side of the business, but it also drives substantial spending—$15.8 billion on AI alone in the second quarter, while the space and connectivity sectors barely surpassed a billion each. This positions SpaceX as a competitor to neocloud companies like CoreWeave and Nebius.

However, this wasn’t the original vision. Musk built the Colossus 1 data center in Memphis for Grok, his AI project that has had its share of controversies. xAI struggled to manage the facility and opted to lease it out instead. They faced latency issues that hindered in-house model training, and the mix of old and new chips created bottlenecks. Musk mentioned that only 10% of the compute resources SpaceX develops will be allocated to Grok.

Now, SpaceX has partnerships with Google, Anthropic, Reflection AI, and Cursor, the AI company Musk eventually decided to acquire. During the earnings call, SpaceX’s CFO Bret Johnsen stated that these partnerships could put the company “on a trajectory, including contribution from Cursor, to reach $100 billion of ARR,” a metric used to project annual revenue based on shorter-term estimates. Musk was even more optimistic, asserting that reaching $100 billion ARR by December is a certainty, with the actual figure potentially being higher.

That’s all well and good, but revenue doesn’t equate to profit, and building data centers is a costly endeavor.

There are inherent challenges in the bare-metal compute rental business—like inevitable obsolescence, construction hurdles, and the fact that compute is essentially a commodity, leading to price competition. As more data centers are built, the supply of compute increases, driving prices down.

Musk has claimed that he took SpaceX public to fund the construction of data centers in space. One might think that if he struggled to manage a data center on Earth, he’d want to master that before attempting something exponentially more complex—something unprecedented, even. But that’s not how Musk operates, it seems.

SpaceX has even pitched an orbital data center concept involving up to a million satellites to the Federal Communications Commission. The application lacks technical specifics—like satellite size or deployment timelines—suggesting it might be more about public relations than practicality. Musk has shared some sketches of the satellites, envisioning a chip manufacturer called Terafab producing one terawatt of chips annually, with a billion Optimus robots handling the labor, assuming Musk figures out how to give them hands.

This all sounds like a hefty investment, doesn’t it? Musk claims the ultimate goal is to construct a mass accelerator on the Moon.

I’m skeptical about these grand plans, just as I am about Musk’s assertion that Starlink will provide “a majority of the world’s internet,” as he mentioned during the earnings call. This rhetoric fits into Musk’s tendency to discuss a sci-fi future inspired by mid-20th-century writers. Sure, many scientists argue that Musk’s vision for space data centers is flawed. You could debate the feasibility of such projects, but let’s not forget: did the Hyperloop ever materialize? It’s all vaporware until it actually launches.

Stripping away the grandiose claims, we see a company that primarily launches rockets for its own use, operates a successful satellite internet service, and is navigating the risky, capital-heavy bare-metal compute business due to its in-house AI setbacks. That’s not quite as exciting as reaching Mars in six years, is it?

However, it does provide Tesla with a significant customer, one that purchases $295 million in Tesla Megapack battery storage. SpaceX has also been acquiring a number of Cybertrucks, Musk’s less-than-stellar vehicle. As of now, Tesla’s stock has dropped 25% since the start of the year.

Musk might be able to charge a premium for his compute services due to his political connections—after all, he doesn’t seem to be planning to spend $100 million on midterm elections for nothing. Perhaps those connections will ease the path for Musk to build terrestrial data centers, although those plans seem unpopular across the political spectrum. Who knows, maybe the strategy is simply, “That’s a nice AI business you have there. It would be a shame if something happened to it. You should consider buying my compute to keep it safe.”

But maybe I’m just thinking too far ahead. After all, SpaceX’s insider lockups are set to expire soon. If insiders sell, as short-sellers anticipate, the already declining stock could take an even bigger hit. So perhaps reminding everyone about the mass accelerator on the Moon is a way to rally support. Good thing Nasdaq adjusted its rules for SpaceX, so those losses will impact everyone holding an index fund!

Actually, I have a sci-fi thought about that. Hey Elon! Why don’t you take a flying leap at a rolling donut? Why don’t you take a flying leap at the Mooooooon?