Oracle Cloud Outage Disrupts TikTok Service in US, Highlighting Infrastructure Reliability Concerns

Summary

Oracle Cloud Infrastructure (OCI) experienced a significant outage on March 3-4, 2026, that disrupted TikTok service for US users. The incident affected Oracle's US East (Ashburn) region, causing connection timeouts, errors, and increased latency for approximately 10 hours before being resolved. As TikTok's cloud infrastructure provider under the TikTok USDS Joint Venture—in which Oracle holds a 15 percent stake—the outage prevented some creators from posting content and degraded the user experience.

The outage began at 1324 UTC on March 3, with Oracle engineers working through the night to identify the root cause and implement mitigations. Service health metrics showed recovery by 0703 UTC on March 4, and the issue was marked as fully resolved by 0924 UTC. This marks the second TikTok disruption related to Oracle infrastructure in just over a month, following a winter storm that took down one of the company's datacenters in February.

The incident has drawn particular attention due to Oracle CEO Larry Ellison's 2022 claim that the company's cloud "doesn't go down." User frustration was evident on social media, with complaints about the extended downtime and questions about Oracle's infrastructure reliability. The outage also raises broader questions about single-vendor cloud dependency, especially for critical services with millions of users. Oracle has not yet provided a detailed explanation of what caused the failure or offered assurances about preventing future interruptions.

• The incident highlights risks of single-vendor cloud dependency for critical consumer services with millions of users