When the most dominant aerospace operation on Earth finally filed the paperwork to go public, Wall Street expected a manifesto on the mechanics of multi-planetary life. Instead, it received a prospectus for a data processing firm.
To read the regulatory S-1 filing for SpaceX's recent initial public offering is to experience a distinct form of corporate cognitive dissonance. Other aerospace conglomerates file under standard industrial codes for spacecraft or defense. SpaceX chose code 7370, the classification reserved for computer programming and data processing.
The market responded with immediate, staggering capital inflows. Within hours of its listing, SpaceX eclipsed a $2.3 trillion valuation, instantly transforming into the seventh most valuable corporation on the globe. It is a market capitalization roughly equivalent to the entire gross domestic product of Canada. To its vocal base of retail backers, the valuation represents a reasonable premium for a business that dismantled the traditional economics of spaceflight. To veteran market observers, it looks like an alarming sign of an asset bubble seeking its next host.
Code 7370
The core of the issue lies in a single metric buried on page 11 of the company's prospectus: the total addressable market. SpaceX values its ultimate revenue opportunity at a massive $28.5 trillion. Yet a mere 15 percent of that figure belongs to rockets, orbital payloads, and satellite communications.
The remaining 85 percent belongs entirely to artificial intelligence and data infrastructure. By the company's own framing of its total addressable market, AI and data infrastructure account for 85 percent of the opportunity, while space and communications make up just 15 percent.
The pivot was executed quietly in February 2026, when SpaceX merged with xAI, the artificial intelligence startup founded by Elon Musk. The transaction fundamentally rewrote the financial identity of the aerospace firm. Before the merger, SpaceX operated as a capital-intensive but fundamentally viable enterprise. In 2024, it cleared $791 million in net income. Its satellite internet division, Starlink, had scaled to over 9,000 active satellites, generating $11.4 billion in revenue and a clean $4.4 billion in operating profit by 2025.
The xAI acquisition changed those figures completely. Under retroactive accounting rules, the combined entity had to declare a net loss of $5 billion for 2025. The satellite network's steady profits were instantly swallowed by a business model that burns through capital at a historic pace.
The Cost of an AI Shadow
Artificial intelligence is an expensive pursuit. In late 2025, xAI was losing two dollars for every single dollar it brought in, burning roughly $28 million every 24 hours. By the first quarter of 2026, those losses hardened into structural deficits. The division's capital expenditure alone reached $7.7 billion for the quarter, an annualized pace of more than $30 billion.
Rockets are no longer the primary driver of SpaceX's expenses. Capital expenditure, which previously sat at a sustainable 42 percent of revenue in 2023, expanded to 215 percent of total revenue following the merger. The company is spending far more than it earns, using public equity markets to bankroll a chatbot called Grok and the computational infrastructure required to keep it competitive.
This financial strain complicated the company's path to the public markets. SpaceX initially attempted to fast-track its entry into the S&P 500 immediately upon listing. Had the index committee agreed, it would have triggered roughly $14 billion in automatic, passive inflows from retirement portfolios and institutional index funds worldwide.
The committee declined. The rules of the S&P 500 require a firm to report net profitability over its most recent four quarters. The $5 billion deficit introduced by xAI disqualified the aerospace giant from entry. The tech-heavy NASDAQ index proved more accommodating, fast-tracking the stock and forcing institutional managers to absorb the asset into their portfolios regardless of their risk tolerance.
Circular Capital and Empty Desks
Outside the trading floor, the operational reality of SpaceX's AI arm raises serious questions about enterprise viability. Grok commands less than 1 percent, roughly 0.4 percent by some industry estimates, of the enterprise AI market, where corporate contracts actually generate recurring yield.
The internal leadership structure has dissolved. Of the 11 co-founders who launched xAI alongside Musk in 2023, not a single one remains at the firm today.
To compensate for low commercial adoption, the company has relied on intricate financing arrangements. A highly publicized agreement with Google to lease GPU compute for $920 million a month features an unusual caveat: either party can dissolve the contract with a mere 90 days' notice. Furthermore, Google maintains a 6 percent equity stake in xAI, an arrangement that bears the distinct characteristics of circular financing designed to inflate revenue metrics during a crucial IPO window.
Even the infrastructure assets have suffered from basic design missteps. The company's premier ground-based mega data center was mistakenly built using three mismatched variations of graphics processing units. Because modern machine learning models require uniform hardware synchronization, the faster chips sat idle waiting for slower legacy silicon to finish calculations. The facility operated at just 11 percent of its theoretical capacity. Incapable of utilizing the site for its own training needs, the company ultimately leased the flawed facility to competitor Anthropic while moving its own model training elsewhere.
The Physics of Cold Silicon
The long-term justification for this corporate restructuring rests on an ambitious regulatory application filed with the Federal Communications Commission: a proposal to construct an orbital "space cloud" consisting of up to one million satellites designed for space-based AI data centers.
The engineering hurdles are formidable.
Data centers are notorious heat engines. On Earth, they are cooled by massive water chilling loops and ambient atmospheric airflow. In the vacuum of space, there is no air, no water, and no natural convection. Heat can only escape through thermal radiation, a passive process that is highly inefficient when applied to high-density, power-hungry graphics processors.
The environment itself is hostile to modern silicon. High-energy cosmic rays and solar protons passing through unshielded hardware cause "bit flipping," spontaneous alterations in binary memory that corrupt complex computational models. Energetic solar flares can destroy unshielded microarchitecture entirely, while ambient electrons within the Van Allen belts systematically degrade solar arrays over time.
Preventing this degradation requires massive, heavy radiation shielding and redundant circuit architecture. Every kilogram of shielding added to a satellite increases the launch cost and reduces the economic competitiveness of the asset. Compounding the risk is the rapid depreciation cycle of hardware; ground-based GPUs face obsolescence within three years. Replacing a constellation of one million orbital servers every few years presents a logistical and financial challenge that defies traditional venture capital timelines.
The structural reality of the SpaceX IPO is clear. Investors who believe they are buying into a high-margin, uncompetitive orbital launch monopoly are instead funding an unproven, capital-intensive artificial intelligence startup. The rockets are real, but they are increasingly serving as the outer shell for a very different kind of machine.
