DeepSeek Suffers Longest-Ever Outage — 7 Hours Down as Pressure Mounts for Next Model
DeepSeek's AI chatbot went dark for 7 hours and 13 minutes on Monday morning — its longest consumer-facing outage since the model's viral breakout in January 2025. The disruption, which DeepSeek itself labeled a "major outage" on its status page, ran from the early hours of Monday until 10:33 AM local time (02:33 GMT). No cause was disclosed; the company's standard protocol is to offer no explanation. Speculation ranges from a server failure to a model update gone wrong, but the silence itself has become part of the story.
The timing couldn't be more loaded. DeepSeek still has not announced its next-generation model, and the global AI industry has been watching for any signal. Meanwhile, some third-party estimates put DeepSeek's user base at roughly 355 million — a scale that makes a 7-hour outage significant infrastructure news, not a routine blip. Previous outages had never crossed the 2-hour mark for consumer users, making this one a statistical outlier and a credibility test for a company that has built its reputation on lean, efficient deployment.
The incident also adds texture to an ongoing question: as Chinese AI models gain global traction, can they maintain the reliability that enterprise and developer users demand? DeepSeek's compute efficiency has been a selling point, but efficiency doesn't buy you uptime when something fails at scale.