DeepSeek Chatbot Suffers Seven-Hour Outage, Longest Since Its 2025 Viral Debut

DeepSeek's AI chatbot went offline for more than seven hours on Monday, according to the company's status page, marking the platform's longest consumer-facing outage since its viral rise in early 2025. The Hangzhou-based startup's initial fix failed. Engineers spent from Sunday evening through Monday morning deploying multiple updates before restoring service at 10:33 a.m. local time, Reuters and Bloomberg reported. With 355 million registered users as of February, the disruption flooded Chinese social media with complaints and sent users scrambling to competitors.

Two incidents, one night, no explanation

The trouble started at 9:35 p.m. China Standard Time on Sunday, March 29. DeepSeek's status page flagged a problem with its web and app chat service. Engineers marked it resolved by 11:23 p.m., according to Techloy. Fifty-seven minutes later, the system broke again.

Engineers pushed fixes at 1:24 a.m. and 9:13 a.m. Reuters called it 7 hours and 13 minutes of downtime. Techloy counted from first report to final fix and landed closer to 13 hours. DeepSeek classified the event as a "major outage" but offered no explanation for either incident, following what Reuters described as company protocol. No root cause. No acknowledgment of what broke. Just a status page entry reading, "a fix has been implemented and we are monitoring the results."

The company did not respond to emailed requests for comment from Bloomberg.

355 million users, one point of failure

"Only after DeepSeek went down did I realise I no longer knew how to work without it," user yezi888 wrote on Xiaohongshu, the lifestyle platform known in the US as RedNote, according to the South China Morning Post.

That frustration rippled through Chinese social media overnight. StatusGator, an outage tracking platform, logged 428 user-submitted reports within 24 hours. Users were locked out of saved chat histories during peak hours, not just blocked from new queries. Coding tasks abandoned. Writing projects stalled. Research workflows frozen. The daily routines that hundreds of millions of people have wired into a single AI chatbot stopped cold.

This is what dependency looks like when the service disappears. If you rely on DeepSeek for your morning work, Sunday night was a reminder that the platform sitting between you and your output has no fallback.

A near-perfect record, broken

DeepSeek has maintained close to 99% uptime since it released its R1 model in January 2025, according to Livemint. Its consumer chatbot had never experienced a major outage exceeding two hours. Monday shattered that.

The company's developer-facing API had a rougher history. Consecutive day-long outages hit during late January 2025 at the peak of DeepSeek's viral moment, when the R1 model rattled Silicon Valley and triggered a global sell-off in tech stocks. DeepSeek blamed those disruptions on large-scale malicious attacks. But the API serves a different audience, developers who integrate the chatbot into custom applications and generally expect some instability during traffic spikes. Monday's outage landed on the consumer product itself. The chatbot. The interface ordinary users open at their desks every morning.

And it arrived when AI platforms can least afford to stumble. When Alibaba's Qwen crashed under 10 million orders during China's Lunar New Year giveaway in February, the incident exposed how fast demand can overwhelm infrastructure that hasn't scaled to match it. DeepSeek now faces the same question, though with a user base many times larger than the audience Alibaba's Qwen served in February.

Rivals gain ground while V4 stays silent

The silence from DeepSeek extends beyond Monday's outage. The company has given no timeline for its next-generation model, widely called DeepSeek V4, despite months of speculation that an update was imminent. Job postings earlier this month hint at where the company is headed. Roles for Agent Deep Learning Algorithm Researcher and Agent Infrastructure Engineer suggest a push into agentic AI, Livemint reported. But no product has materialized.

Competitors are not waiting. Alibaba dropped new models over the Lunar New Year holiday. ByteDance and Tencent matched them. Zhipu AI, MiniMax, Moonshot AI, none of them sat still. Each day without a V4 announcement gives those rivals more room to fill. And for 355 million users weighing whether to explore alternatives, a seven-hour outage makes the decision easier.

DeepSeek went dark the same day Mistral announced $830 million in debt financing to build a data center near Paris equipped with 13,800 Nvidia AI chips. Infrastructure investment as the antidote to reliability risk. DeepSeek's users will be watching to see whether Hangzhou makes the same bet before the next outage hits.

Claude Crashed Three Times in March. Anthropic Has a Fail Whale Problem.

Anthropic's Claude chatbot suffered its third outage in March on Monday, with more than 6,800 users reporting failures on Downdetector by 1:03 p.m. Pacific time. Both the Opus and Sonnet models return

The Implicator

OPINION: Anthropic Knew the Math. It Sold the Tickets Anyway.

On March 23, Anthropic's paying Claude subscribers discovered their sessions had been throttled without notice. Max users handing over $200 a month watched their daily allowance vanish on a single pro

The Implicator

Arm Abandoned Neutrality. Its Biggest Customers Cheered.

The Implicator

Maria Garcia

Los Angeles

Bilingual tech journalist slicing through AI noise at implicator.ai. Decodes digital culture with a ruthless Gen Z lens—fast, sharp, relentlessly curious. Bridges Silicon Valley's marble boardrooms, hunting who tech really serves.