SpaceX and the Cost of Becoming a Compute Company

SpaceX reported third-quarter revenue of $14.6 billion, up 161% from $5.6 billion a year earlier, and net income of $4.4 billion compared with a net loss of $1.1 billion in the same quarter of 2025. The headline suggests a company whose rockets and satellite constellations have finally turned into a dependable profit engine. The far more consequential story sits inside the segment table. The Super Intelligence division, which leases high-performance artificial intelligence computing capacity, generated $7.6 billion in revenue during the period. For the first time, compute revenue surpassed the orbital connectivity business that established the modern operating profile of the firm. Equity investors are no longer evaluating a pure transportation network or an orbital telecommunications utility. They are financing a business whose largest ongoing capital expenditures and fastest-growing cash receipts both sit inside warehouse data centers rather than on coastal launch pads. Starlink continues to provide the operational cash foundation that funds these initiatives. The company concluded the third quarter with 13.6 million consumer broadband subscribers after adding 1.6 million net connections during the three-month period. Each active subscriber requires a constellation of satellites, an extensive terrestrial ground network, and an ongoing customer support apparatus. Because low-Earth-orbit satellites experience atmospheric drag and component degradation, each unit launched into orbit must eventually be replaced on a rolling schedule. That structural reality rewards massive consumer scale, spreading the heavy fixed costs of the orbital bus and ground stations over a wider base of recurring subscribers. Yet it also means that every upward tick in the subscriber base increases the mandatory replacement hardware bill, making connectivity margins depend as much on the internal pace and cost of launches as on aggregate retail demand. The compute business alters this financial arithmetic. The $7.6 billion generated by the Super Intelligence division sits alongside the single largest capital allocation commitment in the multi-year business plan, which is now dedicated to data center infrastructure rather than orbital rocketry. This structural shift matters because commercial compute buyers do not purchase services the way consumers buy home satellite internet. Compute capacity is acquired under massive, multi-year contracts that are inherently large, lumpy, and concentrated. An orbital rocket launch represents a discrete, ring-fenced transaction with a sovereign government, civil agency, or commercial satellite fleet operator. A long-term data center hosting agreement ties up a tenant's primary operational workflow for years, but it can also be renegotiated, pared back, or terminated when the customer encounters its own capital constraints. Extreme customer concentration has emerged as the central operational question facing the company. Bloomberg reported, citing a person familiar with the transaction, that artificial intelligence lab Anthropic represents the company's single largest commercial tenant. SpaceX's quarterly filing identifies this party only as Customer B, which alone accounted for 34.5% of total third-quarter revenue, or approximately $5.04 billion in business. While neither management team has publicly confirmed the counterparty identity on the record, allowing more than a third of total corporate turnover to flow through a single customer fundamentally shifts the risk profile of the entire balance sheet. Strategic capital allocation, balance-sheet liquidity, or corporate governance decisions made inside one external boardroom now carry systemic consequences comparable to a collapse across an entire legacy operating segment. Customer concentration also distorts what headline revenue figures actually represent. Commercial compute contracts routinely carry nominal lifetime maximum spending ceilings that far ex