SpaceX is now a public company, but its first quarterly earnings statement tells a story that feels more like a telecommunications and cloud-computing operation than a rocket maker. By revenue, the company is mostly Starlink and AI compute rental, with actual spaceflight contributing only a small slice of the total.

Headline: SpaceX is barely Space and mostly X

Elon Musk's exploding rockets are a front for the real moneymakers, according to the new financials. Here are the key facts from the first quarterly report as a public company:

• Starlink connectivity revenue reached $4.2 billion in the quarter and was the only segment without an operating loss.
• The space sector contributed just over 10 percent of total revenue and did not break $1 billion.
• SpaceX remains its own biggest rocket customer.
• AI data center leasing brought in more money than rockets and drove $15.8 billion in spending in the second quarter.
• Analysts expect neocloud spending to climb to $65 billion next year.
• SpaceX has deals with Google, Anthropic, Reflection AI, and Cursor, which Musk later chose to acquire.
• CFO Bret Johnsen said the company is on a trajectory to reach $100 billion in annualized revenue run rate, including Cursor's contribution.
• Musk said only 10 percent of the compute SpaceX builds will go to Grok, its in-house AI.
• SpaceX has proposed an orbital data center with up to 1 million satellites to the FCC.
• Insider lockups begin expiring August 6, putting pressure on the stock.

Rockets no longer drive revenue

The old narrative around SpaceX was simple: a private company pushing the boundaries of rocketry, lowering launch costs, and preparing humanity to become multiplanetary. That story has been complicated by the first earnings report since the company went public. The space segment of the business did not break a billion dollars this quarter and contributed only a touch over 10 percent of total revenue. SpaceX remains its own biggest customer for rocket launches, which suggests there simply are not enough outside customers willing to buy rides to orbit.

Meanwhile, the company's connectivity business, otherwise known as Starlink, generated $4.2 billion in revenue and was the only operating segment that did not post a loss. Starlink has become the financial anchor of the entire enterprise, and it is also the platform for a planned phone service that would compete directly with traditional wireless carriers. On the earnings call, Gwynne Shotwell outlined ambitions to take on AT&T, Verizon, and T-Mobile, signaling a broader shift from selling internet terminals to becoming a direct consumer wireless provider.

The AI compute business

But Starlink was not the biggest story in the earnings report. That honor belongs to SpaceX's rapidly growing AI compute rental business. The company has essentially become a neocloud provider, leasing data center capacity to AI companies that need massive clusters of GPUs. In the second quarter alone, SpaceX spent $15.8 billion on AI-related infrastructure. That spending dwarfs the roughly $1 billion each spent on the space and connectivity sectors.

The compute rental business is also generating more revenue than rockets, and it is expected to grow even larger. Analyst Alexander Potter projects that spending on the neocloud business will climb to $65 billion next year, a $17 billion increase from his previous estimate. SpaceX has signed deals with Google, Anthropic, Reflection AI, and Cursor, the AI company Musk eventually chose to acquire. CFO Bret Johnsen said those deals put the company on a trajectory, including Cursor's contribution, to reach $100 billion in annualized revenue run rate. Musk was even more bullish, claiming that hitting $100 billion in ARR by December is not a question mark.

The mess behind the AI pivot

This pivot to AI was not the original plan. Musk built SpaceX's Colossus 1 data center in Memphis for Grok, his in-house AI model that has been beset by embarrassing behavior and technical struggles. But xAI had trouble running the complex, encountering latency issues that made it difficult to train models. The center also used a mix of newer and older chips that created bottlenecks. As a result, xAI decided to rent out the capacity instead. Musk now says only 10 percent of the compute SpaceX builds will go to Grok.

That admission effectively turns SpaceX into a commodity infrastructure provider. The bare-metal business of renting compute is capital-intensive and full of risks. Chips become obsolete quickly, construction timelines slip, and compute prices fall as more data centers come online. If the market is flooded with AI data center capacity, companies will not be able to charge as much for their hardware. Revenue is not profit, and building data centers is expensive, no matter how many billion-dollar deals the company signs.

A science-fiction vision

Musk has claimed that he took SpaceX public because he wanted to build data centers in space. To that end, SpaceX has proposed an orbital data center consisting of as many as 1 million satellites to the Federal Communications Commission. The application is light on technical details, with no clear explanation of satellite size or deployment schedule. That lack of specificity makes it easy to dismiss the filing as a public relations exercise rather than a serious engineering plan.

Musk has also offered a few drawings of the proposed satellites and an accompanying vision that involves a Musk-owned chip producer called Terafab, which would manufacture one terawatt of chips every year. A billion Optimus robots would supposedly do the work, once Musk figures out how to make useful humanoid hands. The end goal, he claims, is to build a mass accelerator on the Moon. This all sounds expensive, and it is far from clear that any of it will ever ship.

Musk's habit of promising a science-fiction future is well documented. The Hyperloop was never built. Full self-driving has been perpetually one year away. Claims about making Mars colonization affordable have not matched reality. So when Musk says Starlink will deliver a majority of the world's internet, or that space data centers are just around the corner, there is little reason to take it at face value. It is all vaporware until it ships.

Tesla, politics, and pressure

Stripping away the grand vision leaves a company that launches rockets mostly for itself, runs a successful satellite internet business, and operates a risky data center rental arm born out of its AI failures. SpaceX also provides an important lifeline to Tesla. The company is buying $295 million in Tesla Megapack battery storage and has been purchasing Cybertrucks, which have been a historic flop in the automobile market. Tesla stock is down 25 percent since January, and those purchases help prop up one of Musk's other troubled ventures.

Musk may be able to charge a premium for his compute because of his political connections. He has signaled plans to spend heavily on the midterms, and those connections could make it easier to get terrestrial data centers approved. But those projects are wildly unpopular across the political spectrum. There is also an implicit threat in Musk's position: if regulators or politicians cross him, access to his AI infrastructure might become less reliable. That is a dangerous dynamic for any industry built on trust and open markets.

Lockups are coming

All of this context matters because SpaceX's insider lockups start expiring on August 6, which is the day after this article was published. If insiders decide to sell, as short-sellers expect them to, the already slumping stock could fall even further. The sudden push for Moon accelerators and orbital data centers may be an attempt to rally investors and employees before the lockup expiration becomes a new wave of sell pressure.

Nasdaq changed its rules to allow SpaceX to be included in index funds, meaning those losses will hit ordinary investors who hold broad stock market funds. The company's financial reality is now much more mundane than the science-fiction headlines suggest: a satellite internet provider, a compute landlord, and a rocket company that is its own best customer. That may not be the dream of reaching Mars in six years, but it is the actual business of the company now called SpaceX.

Source: The Verge News