EVENT DOSSIER
On September 3, 2026, ChatGPT from OpenAI, Claude from Anthropic, and Grok from SpaceX all experienced failures at the same time. OpenAI encountered a problem at 7:43 AM ET on Thursday and resumed operations around 8:17; Claude was down for about three hours before returning to normal at 16:16 UTC; Grok was interrupted due to a failure at SpaceX’s Memphis data center. Although the simultaneous outages by the three services raised concerns about Cloudflare, their common supplier, Cloudflare confirmed that its systems were functioning normally, and there were no technical issues with AWS, Google Cloud, and Microsoft Azure. SpaceX subsequently issued an apology, stating that the problems were caused by failures at its internal data centers, and all services have now been restored.
OpenAI 表示其 ChatGPT 服务已恢复，此前因路由错误导致部分用户无法使用。与此同时，Anthropic 的 Claude 及 SpaceX 的 Grok 也均经历了故障并宣布修复。OpenAI 方面，问题始于周四上午 7:43 美东时间，约 8:17 美东时间解决方案实施成功；Claude 在经历三小时六分钟停机后于 UTC 16:16 恢复；SpaceX 则因孟菲斯计算中心故障导致 Grok 服务中断。尽管三家 AI 提供商同时出现可用性问题引发对共同供应商 Cloudflare 的质疑，但 Cloudflare 确认其系统运行正常，且 AWS、Google Cloud 和 Microsoft Azure 亦无相关技术困难迹象。SpaceX 随后道歉并指出问题源于其计算中心故障，现已全部恢复。