SpaceX's upcoming initial public offering (IPO) reveals that the company's massive $1.5 trillion valuation depends almost entirely on whether it can execute an unconventional strategy: building artificial intelligence data centers in orbit and manufacturing its own chips. The company filed its S-1 prospectus on May 24, 2026, exposing the financial reality behind Elon Musk's sprawling space, connectivity, and AI empire.

Right now, only one part of SpaceX's business is actually profitable. The Connectivity segment, which includes Starlink broadband services, generated $11.4 billion in revenue during 2025 with $4.4 billion in operating income. That's impressive, but it's not enough to justify a $1.5 trillion valuation on its own. The Space segment lost $657 million in 2025, while the AI segment lost a staggering $6.4 billion despite generating $3.2 billion in revenue.

Why Is SpaceX's AI Business Losing So Much Money?

The AI segment includes xAI's Grok language models, X's digital advertising business, and rental agreements with competitors. The problem is clear: xAI's models haven't gained meaningful traction against established competitors like Anthropic, which is reporting a $30 billion annualized revenue run rate as of April 2026. To fill its expensive computing capacity, xAI is renting out $1.25 billion per month to Anthropic, essentially admitting that its own models aren't competitive enough to use all the infrastructure it built.

This creates a paradox. SpaceX estimates that 93 percent of its total addressable market, roughly $26.5 trillion out of $28.5 trillion, comes from AI applications. Yet the company's own AI models are struggling to capture that opportunity. The rental agreement with Anthropic helps the numbers look better on paper, but it's a tacit acknowledgment that xAI needs to improve its competitive position.

How Could SpaceX Gain an AI Advantage Over Competitors?

SpaceX's S-1 filing outlines a differentiated strategy with three key components designed to lower the cost of running AI models:

• Space-Based Data Centers: By placing data centers in orbit, SpaceX can power them with solar arrays that receive perpetual sunlight, eliminating the utility costs that terrestrial data centers must pay. This could provide a significant cost advantage in the expensive business of running large language models.
• Custom Chip Manufacturing: SpaceX is partnering with Tesla on a joint venture called Terafab to design and manufacture their own AI chips. This would eliminate the massive margins paid to Nvidia and Taiwan Semiconductor Manufacturing (TSMC), which currently dominate the market.
• Vertical Integration: By controlling the underlying model, the computing chips, and the energy supply, SpaceX could theoretically undercut competitors on the cost per token, a key metric for AI profitability.

The challenge is that both initiatives sound implausible by today's standards. Sending data centers to space has never been done before, and semiconductor fabs typically take years to go from conception to production. However, SpaceX has a track record of achieving what seemed impossible, such as landing and reusing rockets on a launchpad.

What Should Investors Watch After SpaceX Goes Public?

The space-based AI data center strategy is the critical variable that will determine whether SpaceX is undervalued or overvalued at $1.5 trillion. If the company can execute on this vision, it could gain a genuine competitive advantage in the AI infrastructure market. If it cannot, investors will be paying for Starlink's profitability plus a collection of unprofitable businesses with uncertain futures.

The first-quarter 2026 results offer a cautionary note. The Space segment's revenue dropped to $619 million, well below the $1 billion quarterly run rate from 2025, though this could reflect seasonal launch timing. The AI segment's revenue of $818 million in Q1 2026 shows essentially flat growth compared to the 2025 average, suggesting that the business is not accelerating.

For now, SpaceX's near-term business remains centered on Starlink, which is genuinely profitable and growing. The company also outlines ambitious long-term plans to mine minerals on the moon and colonize Mars. But the medium-term upside, and where SpaceX must show meaningful progress, is in AI. The company's ability to make space-based data centers and custom chips a reality will likely determine whether its IPO valuation proves justified or becomes a cautionary tale about betting on unproven technology at an unprecedented scale.