Elon Musk said SpaceX will build its AI services exclusively on Nvidia's chips, declaring "So we're exclusive to Nvidia" in remarks reported August 4, 2026. The announcement came just one day after SpaceX released its first-ever financial results, which showed how much Musk is spending on AI infrastructure.

Nvidia makes the specialized computer chips, called GPUs, that power most of today's artificial intelligence. Think of a GPU as the engine inside a very fast car — it does the heavy mathematical work that lets AI systems learn from enormous amounts of data. Nvidia currently dominates this market the way Intel once dominated the chips in personal computers.

SpaceX's first financial results, reported August 3, revealed compute contracts that could exceed $25 billion per year, according to Reuters. That figure puts Musk's AI spending in a different league from anything SpaceX has done before. The same results included warnings to investors about chip supply risks and high capital expenditures — the money a company spends on physical infrastructure like data centers and equipment.

The commitment to Nvidia is not really a new direction. In March 2026, Musk said SpaceX AI and Tesla expect to keep ordering Nvidia chips at large scale. What the August statement adds is an explicit closing of the door to other chip suppliers for SpaceX's AI work. The phrasing leaves no ambiguity: exclusive means exclusive.

The relationship between Nvidia and Musk's companies has been growing for some time. Jensen Huang, Nvidia's CEO, hand-delivered a device called the DGX Spark — described as the world's smallest AI supercomputer — to Musk at SpaceX on October 13, 2025. Musk, who also leads the AI company xAI, later praised Nvidia's Rubin platform, calling it "a rocket engine for AI." Nvidia's networking technology, called Spectrum-X, is already used to speed up xAI's workloads, including a supercomputer called Colossus. Another Nvidia component, the NVLink Switch, connects chips together at very high speeds and is part of the same system tying Musk's AI projects to Nvidia's product roadmap.

What makes the exclusivity declaration interesting is the tension it creates with SpaceX's own plans. In April 2026, SpaceX told investors it is working toward building its own GPUs through a project called Terafab, while also warning about chip supply risks and high spending. So SpaceX is publicly committing to Nvidia-only AI systems while simultaneously trying to develop its own chips. That is a near-term bet on Nvidia to get AI work done quickly, paired with a long-term plan to reduce reliance on that same company.

The $25 billion annual spending figure is the number that matters for anyone trying to understand SpaceX's costs. At that level, Nvidia's chips are not just one expense among many — they are the single biggest category of spending. Going exclusive means SpaceX cannot shop around for a cheaper second supplier, at least while the commitment holds. It also means all the risk of a supply shortage sits with one company, at a time when SpaceX itself is warning investors about exactly that risk.

The broader context here is that multiple Musk companies — SpaceX AI, xAI, and Tesla — are all buying Nvidia products across different layers of their systems. Musk's public praise of Nvidia's Rubin platform signals alignment with Nvidia's future products, not just the ones available today.

For anyone watching the markets, the key unanswered questions are how long the exclusivity lasts and how broad it is. Musk did not specify either in his remarks. A permanent commitment to Nvidia at $25 billion a year is very different from a short-term or limited arrangement. The Terafab project adds another unknown: if SpaceX successfully builds its own chips, the Nvidia exclusivity would presumably need to change, or the in-house chips would need to be used for something other than AI.

SpaceX's own warning to investors about chip supply costs suggests the company sees the current setup as expensive. Buying only from Nvidia, which charges premium prices for its data-center chips, does not ease that financial pressure in the short term. What it does provide is simplicity: using one vendor for everything means the pieces are designed to work together, which has real value when you are building at the scale Musk is attempting.

The financial results and the exclusivity statement, arriving within 48 hours of each other, frame the central tension: SpaceX is making a very large bet on Nvidia chips while also telling investors it needs to build its own chips to manage costs and supply risk. Both things can be true at the same time, and neither one solves the other.