SpaceX is now an 'infrastructure company' — and it is betting on xAI to get to Mars

SpaceX no longer sees itself primarily as a rocket company. "Fundamentally we are an infrastructure company at this point," said Kiko Dontchev, the company's vice president of launch, speaking at the UP.Summit conference in Florida last week.

The remark is the clearest signal yet that SpaceX's identity has shifted away from the business of building newer, larger and ever-more-reusable rockets — the work that made the company famous — toward data, artificial intelligence and orbital infrastructure. That shift began roughly a decade ago with the introduction of Starlink, the company's satellite internet constellation. According to Dontchev, it has accelerated sharply during the last 18 months as SpaceX has pivoted to data and AI.

"We're building infrastructure on the ground and low Earth orbit, soon on the Moon, and obviously eventually Mars," Dontchev said.

What did SpaceX say about its identity?

Dontchev's framing matters because it comes from the executive responsible for launch — the part of the business most closely associated with rockets. For a long time, SpaceX has been synonymous with the development of newer, larger and increasingly reusable launch vehicles. A launch VP describing the company as an infrastructure builder signals that rockets are now a means, not the end.

The infrastructure Dontchev described spans three domains: ground infrastructure on Earth, infrastructure in low Earth orbit (the region below roughly 2,000 kilometers where Starlink operates), and, eventually, infrastructure on the Moon and Mars. He offered no timeline for the lunar or Martian phases.

In plain words: the executive in charge of launching rockets now describes those rockets as one layer of a much broader build-out.

How is xAI changing how SpaceX works?

Earlier this year, SpaceX acquired xAI, the artificial intelligence company founded by Elon Musk, who also founded SpaceX. Dontchev said SpaceX engineers were already using xAI's tools before the deal. What changed is the pace.

"What's happened at SpaceX is we've really gone through a huge acceleration of adoption," Dontchev said. "Even just a year ago, we weren't adopting it that quickly, but really starting this year, we started to adopt it extremely rapidly. And the acceleration we have seen in our ability to do engineering is unlike anything I've ever seen at the company in my 16 years."

That is a strong claim from a 16-year veteran, and Dontchev backed it with a concrete metric: tasks that used to take SpaceX employees hours to days now take seconds to minutes. He did not specify which engineering tasks, how the figures were measured, or how the productivity gains were verified — a notable gap in an otherwise sweeping claim.

The takeaway: SpaceX says AI has compressed days of work into minutes, but the company has not published the evidence behind that number.

Where is AI already deployed?

The most specific example Dontchev gave was not engineering design but safety inspection. He said the work of identifying foreign object debris — known in the industry as FOD, stray material left inside aerospace hardware — has been largely automated using Grok, xAI's AI assistant.

FOD is not a marginal concern. This debris can cause catastrophic damage to engines and other spacecraft components, which is why detection has traditionally demanded careful human inspection. Automating that process at scale would affect both the speed and the reliability of SpaceX's production lines.

Dontchev did not say when the Grok-based FOD detection was deployed, what accuracy rate it achieves, or how much human oversight remains.

In plain words: the one concrete AI deployment SpaceX named is automated debris inspection, a safety-critical task — and the company provided no performance data for it.

What happens to the Falcon 9?

Dontchev also addressed the future of the Falcon 9, the rocket that has dominated global launch for a decade and remains, by the company's own account, the world's preeminent launch vehicle. His message: the Falcon 9 is firmly on the road to retirement.

He offered no retirement date and no successor transition schedule in the remarks reported from the summit. The stated direction — retiring the industry's leading rocket while reorienting the company around infrastructure and AI — frames the stakes: SpaceX is willing to sunset its most proven product to pursue a broader ambition.

The caveat is timing. "Firmly on the road to retirement" is a direction, not a schedule, and the Falcon 9's dominance means any replacement will be judged against a decade-long record of reliability.

In plain words: SpaceX's most successful rocket is being wound down in principle, with no announced deadline.

Why does this matter?

Two currents run through Dontchev's comments. The first is consolidation inside the Musk ecosystem: SpaceX acquired xAI earlier this year, and within months its tools — including Grok — are described as transforming engineering workflows and safety inspection. That vertical integration of an AI lab into a hardware manufacturer is a model other aerospace and defense companies are watching closely.

The second is the market signal. A company that has dominated global launch for a decade is telling investors, customers and regulators that its future lies in orbital and planetary infrastructure, with AI as the accelerant. If the claimed productivity gains are real — hours-to-days of work compressed to seconds-to-minutes — they would reshape how launch vehicles and satellites get built. If they are not, the rhetoric will face the same scrutiny as the timelines that have historically slipped.

Dontchev's own words carry the strongest evidence and the sharpest open question: the adoption acceleration is "unlike anything" he has seen in 16 years, and the company that just retired its own self-image as a rocket builder still has to prove the new one.