The Mystery of the Simultaneous AI Blackouts

By Lily Hay Newman | Security | September 3, 2026, 5:56 PM

On Thursday morning, a rare series of outages struck the frontier models of Anthropic, OpenAI, and xAI, leaving users of ChatGPT, Claude, and Grok in the dark. While the timing was nearly identical, the reasons behind the downtime remain murky.

The Breakdown by Provider

While the events coincided, the explanations provided by the companies varied significantly.

1. xAI (Grok)

SpaceX, the parent organization of xAI, attributed the Grok failures to a specific physical location.

“An outage at our Memphis compute center this morning,” the company stated on Thursday afternoon.

Interestingly, SpaceX also issued a broader apology: “We’d also like to apologize to our impacted compute partners.” This is notable because Anthropic and xAI had previously entered into a “compute partnership” with SpaceX back in May.

2. OpenAI (ChatGPT)

OpenAI's spokesperson, Kathleen Chaykowski, identified a technical glitch rather than a hardware failure.

“A routing error starting around 7:43 am PT on Thursday, September 3, made ChatGPT and Codex unavailable for some users across platforms.”

3. Anthropic (Claude)

Anthropic was less communicative, declining to provide a detailed comment to WIRED, though they did update their status page. They reported a “partial outage” involving several specific models.

---

Timeline of Events

The following table summarizes the chaos of Thursday morning:

Visualizing the Overlap

Technical Nuances & Unanswered Questions

The coincidence of these outages suggests a shared dependency. Mathematically, the probability of three independent major AI providers failing within a narrow window is low: $P(\text{Simultaneous Failure}) \approx P(\text{OpenAI}) \times P(\text{Anthropic}) \times P(\text{xAI})$

Usually, such a pattern points to a failure in a Content Delivery Network (CDN) or a Cloud Service Provider. However:

• Cloudflare reported no issues.
• Amazon Web Services (AWS) reported no issues.
• Microsoft Azure reported no issues.

Furthermore, neither OpenAI nor Anthropic mentioned an external vendor.

Other Observations

• Google Gemini: There were scattered reports of Gemini failing, but Google did not confirm this, and their status dashboard remained clear.
• Claude Specifics: The outage primarily hit Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5. Claude Sonnet 5 experienced brief issues shortly after 9:00 am PT.

Current Status Checklist

• Grok services restored
• ChatGPT/Codex routing fixed
• Claude Mythos/Fable/Opus resolved
• Definitive shared cause identified

---

About the Author

Lily Hay Newman is a senior writer at WIRED specializing in hacking, digital privacy, and information security. Her previous experience includes roles at Slate and Future Tense.

---

Related Stories

Other Headlines:

• AI slop backlash is actually having an impact
• The cop who took on Flock
• The rise of the 1am job interview
• Silicon Valley’s ‘Hot Girl Renaissance’