DeepSeek Suffers 13-Hour Outage — Longest Since January 2025 Launch

DeepSeek's consumer chatbot platform (355 million registered users) went offline at 21:35 CST on March 29 in what became the service's longest outage since its January 2025 global launch. Engineers partially resolved the first incident at 23:23 CST, but a second failure began at 00:20 CST and lasted until 10:33 AM CST on March 30 — a combined downtime of over 13 hours across two back-to-back incidents. No public explanation was issued. The outage revealed a significant infrastructure reliability gap: DeepSeek has scaled to hundreds of millions of users but its backend infrastructure was not architected for the level of redundancy that US hyperscalers maintain. Analysts noted the incident underscores the challenge Chinese AI labs face in building production-grade infrastructure while simultaneously developing frontier models under chip restrictions.