SpaceX has moved more than 300 engineers from its rocket and Starlink divisions into its AI data center operations, a staffing surge that signals just how seriously the company is treating its computing infrastructure ambitions. The reassignment followed an internal call for volunteers that attracted over 1,300 applicants, roughly four times the number ultimately selected.
The engineering influx is part of a broader management shakeup that began in late June 2026, when the company replaced data center team lead Dan Rowland with Wesley Salandro, a veteran of SpaceX’s rocket production operations.
Fixing what fast growth broke
SpaceXAI currently operates roughly 1.4 GW of computing capacity, with plans to push past 2 GW before the end of 2026.
That attention, it turns out, was lacking during the initial build-out phase. The Memphis-area Colossus campus, SpaceXAI’s flagship facility, has faced persistent reliability challenges. Reports indicate that critical sites were operating without essential backup systems during the company’s most aggressive construction period.
The newly assigned engineers are focused on three priorities: building redundancy into power and cooling systems, establishing more rigorous testing protocols, and transitioning away from temporary infrastructure setups that were acceptable during the sprint phase but unsuitable for long-term operations.
Expansion plans outside the Memphis region have been deliberately slowed. The company is choosing to stabilize what it has before scaling further.
A $2.6 billion quarterly business
The urgency behind these changes becomes obvious when you look at the revenue numbers. SpaceXAI generated $2.6 billion in external compute rental revenue during the second quarter of 2026 alone.
Companies like Google and Anthropic have been among the entities working with SpaceX’s computing infrastructure, and the reliability improvements are partly aimed at retaining and expanding those relationships.
From ground to orbit
The data center push also connects to SpaceX’s longer-term vision for orbital computing infrastructure. The company is targeting an initial ramp-up of satellite production for planned orbital data centers by 2027.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.

1 week ago
77





English (US) ·