Breaking

SpaceX Engineers Rework xAI Data Centers After Reliability Problems

September 15, 2026 at 08:31 EDT

Infra & Chips
Foundation Models
Sales & Business

SpaceX has moved hundreds of rocket engineers into xAI's data-center build-out and is redesigning parts of existing facilities, shifting from a speed-first construction push to a reliability-first plan. The change could slow expansion of one of the largest AI-compute operations in the U.S. just as SpaceX is selling capacity to outside AI customers.

Sep. 15, 2026 | SpaceXAI

From speed build to uptime build

SpaceX rocket engineers are reworking xAI data centers with more redundancy, earlier testing and tighter construction practices as AI-compute contracts grow.

300+

SpaceX engineers on the data-center project

200K

H100 GPUs in the Colossus build-out

$1.11B

monthly value of a new hosting deal

$16B

Q2 AI infrastructure spending

The speed benchmark

24 months

expected

4 months

delivered

Each block represents four months. The old advantage was speed; the new challenge is keeping that capacity stable.

What changes in the build

Leadership reset

Facility redesign

Backup power and cooling

Earlier testing

Why it helps

Paid AI workloads need dependable power, cooling and commissioning. More redundancy can reduce costly interruptions.

What it risks

Tighter testing and redesign work may slow new GPU capacity while demand for AI compute remains constrained.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.