SpaceX is pursuing a sci-fi future of data centers in space. But in the short term, Wall Street is raising questions about the company’s higher-than-expected spending to build them here on Earth.

In the near term, the company's AI division, formerly known as xAI, is competing with AI-model builders like OpenAI and Anthropic and AI computing providers like CoreWeave. In the longer term, Elon Musk hasn't been shy about touting his vision of the firm becoming a provider of space-based AI data centers — effectively combining SpaceX’s capabilities in spaceflight, space-based internet and AI infrastructure.

Tuesday’s earnings report, SpaceX’s first since becoming a public company, highlighted a growing tension in the firm’s pitch to investors, as Wall Street weighs Musk’s lofty goals — and track record — against short-term performance that is raising red flags.

For many investors, backing Musk’s companies means betting on his ability to turn improbable-sounding technological ambitions into reality. In the case of SpaceX, this means buying into the likelihood that the firm will launch vast constellations of AI computing satellites into orbit, even if many of the details remain unclear.

It is this faith in Musk’s ability to execute on grand, long-term visions that underpinned SpaceX’s record IPO and have driven its valuation close to $1.6B.

At the same time, Wall Street is now evaluating SpaceX through the same lens it applies to other AI infrastructure giants like Amazon, Google, Microsoft and Meta, with investors demanding that unprecedented capital expenditures be matched by a credible path to AI profitability.

Just as markets punished Meta and Google last month when AI infrastructure spending outpaced near-term returns, investors sent SpaceX shares plummeting close to 13% after the company reported higher-than-expected capex that continues to outpace earnings from its AI business.

"The relationship between capex and revenue is unsustainable, so capex has to fall or revenue has to grow tremendously, and that is where faith in Musk's vision, engineering leadership and execution track record separates the ​bulls from the bears," Drew Cupps, portfolio manager at Polen Capital, told Reuters.

SpaceX’s spending skyrocketed to $28.5B in the first half of 2026, a more than 300% year-over-year increase, with the vast majority of that spending going toward AI infrastructure. The same level of spending is likely to persist through the next two quarters, SpaceX executives told analysts Tuesday.

SpaceX’s AI business, which operated independently as xAI until a February merger, generates revenue through two segments. The company is an AI-model developer: its Grok AI products are competitors with OpenAI and Anthropic, and the firm acquired AI startup Cursor this year to strengthen its offerings. At the same time, SpaceX sells access to its AI computing capacity to other companies, including Google, Anthropic and Apple.

While SpaceX's AI revenues are rising, reaching nearly $3.4B for the first half of the year, they aren't growing as fast as the firm’s spending.

The company reported a net loss of $4.9B last year, largely due to hefty investments in artificial intelligence infrastructure. Its only source of profit was its connectivity segment, anchored by the Starlink satellite internet service.

Still, Musk wasn't shy about touting the scale of the company’s digital infrastructure investment, telling analysts that SpaceX is building AI computing capacity faster than any other company. Much of SpaceX’s AI infrastructure is concentrated in its Colossus data center cluster in the suburbs of Memphis, Tennessee.

Musk said he expects the firm to end 2026 with more than 2 gigawatts of total computing capacity, and that will expand rapidly in the coming months. He told analysts the firm’s capacity will approach 10 GW by the end of 2027.

Such aggressive capex plans are making many investors nervous. Analysts said this anxiety will continue until the firm can demonstrate a pathway to economic sustainability.

“We've watched the same scrutiny land on Big Tech this earnings season, where investors have questioned open-ended wallets and started demanding a visible return on ​them,” said Josh Gilbert, lead analyst at trading ​platform eToro, according to Reuters. “SpaceX faces that test with ⁠an added degree of difficulty because it's asking shareholders to bankroll data centres in orbit.”

As for the plans for those orbital data centers, SpaceX leaders worked to frame them as an imminent reality more than a sci-fi pipe dream.

SpaceX isn’t the only firm putting real resources into pursuing space-based data centers, driven by the promise of limitless free, uninterrupted solar power amid energy constraints and political blowback on Earth. Google is also exploring orbital data centers through a program called Project Suncatcher, as are Jeff Bezos’s Blue Origin and a host of well-funded startups like Aetherflux and Relativity Space.

Musk’s specific vision, outlined in a filing with the Federal Communications Commission earlier this year, involves a constellation of up to a million solar-powered satellites designed to function as interconnected AI data centers. These satellites have since been branded as Starmind.

Speaking with analysts Tuesday, Musk touted SpaceX’s partnership with graphics processing unit giant Nvidia, which announced this week that it is adapting its existing processors and systems for Starmind. And while he acknowledged the challenges of making orbital data centers a reality — calling terrestrial data centers a “trivial problem” compared to launching them into space — he also emphasized that he expects the first test launch of Nvidia-equipped Starmind satellites in the coming months.

“This is not some sort of far-future, distant thing,” Musk said. “We expect to start launching these next year.”

However, obstacles remain. Among the most significant hurdles are concerns about the development of SpaceX’s Starship rocket, which is critical to its space-based data center ambitions.

Launch costs — and the limited availability of rockets capable of carrying data center nodes into space at the scale envisioned — remain among the biggest obstacles to the feasibility of orbital computing. Starship’s significantly larger payload has the potential to shift the economics of spaceflight, lowering launch costs to a point where space AI deployments become commercially viable.

While a largely successful test flight was conducted in July, development of Starship has gone over budget and fallen behind schedule. And although Musk and his team expressed belief that Starship will begin making commercial flights in the near future, many analysts have voiced doubt, citing Musk’s tendency to give overly optimistic or misleading timelines for product development.

“I felt like we got sold the story again, which I appreciate,” Bloomberg Intelligence analyst George Ferguson said. “That’s what management teams get to do. But in my gut, it’s overly optimistic.”