Anthropic becomes SpaceX's largest AI services customer in latest quarter
Demand for computing power is becoming a meaningful revenue driver inside SpaceX as its AI unit expands beyond the company's traditional satellite and launch businesses. Anthropic is the company's biggest single customer for those services in the third quarter, representing more than one-third of revenue for the division, according to a person familiar with the matter.
Highlights
- Anthropic accounted for about 34.5% of SpaceX's Super Intelligence division revenue in the third quarter, becoming its largest AI services customer.
- SpaceX's SEC filing reveals significant customer concentration in its AI unit, with one client (Anthropic) responsible for over a third of quarterly revenue.
- Anthropic has surpassed the U.S. government as SpaceX's biggest single AI client, accelerating AI infrastructure as a commercial segment for SpaceX.
Customer concentration in SpaceX's AI unit
As first reported by Bloomberg, Anthropic is identified as the customer behind a major revenue concentration disclosed by SpaceX in a filing last week. The filing says about 34.5% of the company's revenue in the last quarter came from a single client paying its AI division, now called Super Intelligence, which supplies computing power.That customer was listed only as "Customer B" in the filing. A person familiar with the matter says the client is Anthropic, declining to be identified because the details are private.
Implications for SpaceX's broader business mix
The disclosure highlights how rapidly AI infrastructure is emerging as a significant commercial activity for SpaceX alongside its satellite and rocket operations. It also points to a high degree of customer concentration in the unit during the quarter, with one client accounting for more than a third of revenue.Anthropic's position means it has surpassed the U.S. government as SpaceX's largest single customer for AI services, based on the information described in the report. The figures refer specifically to the third quarter of this year.
In our earlier article on the U.S. AI data-center buildout, we examined how rising financing costs, construction delays, and power constraints are slowing the pace at which new computing capacity can come online. We also noted that only a minority of planned projects look truly viable, putting investor confidence under pressure even as demand for AI compute remains strong.