Grok went offline for three and a half hours on September 3, 2026, after an outage at a SpaceXAI data center in Memphis. The disruption began at 6:30AM PT and affected Grok along with other compute partners connected to the facility.
SpaceXAI issued a brief apology to Grok users and the compute partners affected by the incident. The company said the Memphis compute center outage caused the issues that Grok users experienced that morning, while all systems have now been restored and are functioning nominally.
What happened at the Memphis facility
The outage centered on a Memphis data center run by SpaceXAI. Grok remained unavailable for three and a half hours after the incident began at 6:30AM PT on September 3, 2026. That placed the disruption among the day’s first major service interruptions involving AI systems.
SpaceXAI addressed the incident directly in its apology: “We are sorry for the issues you may have experienced with Grok following an outage at our Memphis compute center this morning.” The company also said, “We’d also like to apologize to our impacted compute partners.”
The wording matters because Grok was not the only system affected by the event at the Memphis facility. SpaceXAI described the outage as affecting other compute partners as well, although it did not provide a longer list of those partners in its apology.
Elon Musk said SpaceXAI was “taking corrective action to ensure this does not happen again.” SpaceXAI also said its systems had been restored and were functioning nominally, giving users and affected partners a clear status after the three-and-a-half-hour interruption.
Other AI services faced problems that morning
The same morning included service problems involving Anthropic and OpenAI. Anthropic’s Claude experienced issues starting around 6:30AM PT, the same time Grok’s outage began, and those Claude problems were resolved by 9:16AM PT.
The affected Claude models were Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5. These model names were part of the affected service details for Anthropic’s incident on September 3, 2026.
OpenAI also experienced an outage that began around 7:30AM PT. The interruption affected ChatGPT and Codex, and OpenAI resolved the issues by 9:55AM PT.
That means the day included separate service problems involving Grok, Claude, ChatGPT, and Codex. Grok’s outage began first at 6:30AM PT and lasted three and a half hours, while Anthropic resolved its Claude issues by 9:16AM PT and OpenAI resolved its problems by 9:55AM PT.
Why the timeline matters
The timing provides a clear record of how the disruptions unfolded. SpaceXAI’s Memphis data center outage began at 6:30AM PT, Claude problems started around the same time, and OpenAI’s outage began around 7:30AM PT. The affected services were restored at different times.
For Grok, the key detail is the length of the interruption: three and a half hours. For Anthropic, the listed end time was 9:16AM PT, and for OpenAI, the listed end time was 9:55AM PT. Each service had its own outage window on September 3, 2026.
SpaceXAI’s apology and Musk’s statement also establish the company’s response. SpaceXAI apologized to Grok users and impacted compute partners, confirmed that systems were functioning nominally, and said corrective action was underway to prevent another incident.
For now, the confirmed outcome is restoration. Grok, the affected compute partners, Claude, ChatGPT, and Codex are tied to a morning marked by service interruptions, with the Memphis outage serving as the source of Grok’s three-and-a-half-hour downtime.
Based on
