Thursday morning brought an unsettling silence to the bustling world of Artificial Intelligence. Leading frontier models from OpenAI, Anthropic, and xAI — the powerhouses behind ChatGPT, Claude, and Grok — all experienced rare and concurrent outages. For users relying on these advanced AI chatbots, it meant unexpected downtime and a nagging question: What happened?
The Unraveling of a Simultaneous Blackout
While the timing suggested a common culprit, the explanations offered by the AI giants varied, adding a layer of mystery to the situation.
- OpenAI’s ChatGPT and Codex: An OpenAI spokesperson confirmed a “routing error” that began around 7:43 am PT, affecting some users across platforms. By 8:17 am PT, a solution was implemented, and services were being monitored.
- Anthropic’s Claude: Anthropic reported a “partial outage” starting at 6:23 am PT, specifically impacting Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5 with “elevated errors.” Though they declined to comment on specifics to WIRED, the company swiftly identified the cause and deployed a fix, resolving the issue by 9:16 am PT. Claude Sonnet 5 also faced brief issues.
- xAI’s Grok: The earliest reports came from xAI, with Grok experiencing outages across all platforms from 6:30 am PT. SpaceX, xAI’s parent company, later attributed the issues to “an outage at our Memphis compute center this morning” and remarkably apologized to their “impacted compute partners.” This apology raised eyebrows, especially given Anthropic and xAI’s announced “compute partnership” with SpaceX in May.
A Coincidence or a Connected Thread?
In the tech world, simultaneous outages often point to a single point of failure – perhaps a major cloud provider, a content delivery network, or another third-party vendor affecting multiple customers. However, in this instance, prominent internet infrastructure providers like Cloudflare, Amazon Web Services (AWS), and Microsoft Azure reported no major issues on Thursday. Neither OpenAI nor Anthropic explicitly cited an external source.
The plot thickens with xAI’s statement. Their apology to “impacted compute partners” in the wake of a SpaceX compute center outage subtly hints at a potential shared infrastructure dependency, particularly considering Anthropic’s previously announced partnership with SpaceX. While not a definitive answer for all outages, it provides a compelling piece of the puzzle for a significant portion of the downtime.
Scattered reports of a possible Google Gemini outage also surfaced, though unconfirmed by Google itself.
The Importance of Resilience in the AI Era
These incidents underscore the critical need for robust infrastructure and transparent communication in the rapidly evolving AI landscape. As our reliance on these frontier models grows, understanding the causes of downtime—and ensuring swift, effective resolutions—becomes paramount for both developers and users. The mystery of the simultaneous outages may not be fully solved for all parties, but it serves as a stark reminder that even the most advanced AI systems are subject to the real-world vulnerabilities of the infrastructure that powers them.
Read more: AI’s Dual Leap: Anthropic Dominates Science Benchmarks, OpenAI Fortifies Cyber Defenses
Read more: OpenAI Unleashes GPT-6 Astra: The Next Evolution for ChatGPT and Codex
Read more: The Green Paradox: Is AI’s Boom Undermining Tech’s Climate Pledges?
