Search
IDCNOVA

CoreWeave Reports Service Disruption Across Cloud Platform, Schedules Network Maintenance Through Late October

By: IDCNOVARegion: North America
CoreWeave, a major AI-focused cloud infrastructure provider, has reported a service disruption affecting multiple components of its cloud platform, including its core infrastructure, object storage, services, and inference offerings, according to the company's status page. The incident, which impacted the US-EAST-15 availability zone, comes as the company simultaneously announced a series of planned network maintenance windows stretching through the end of October 2026.

The disruption affected CoreWeave Infrastructure, CoreWeave Object Storage, CoreWeave Services, and CoreWeave Inference, though the company indicated that several other components—including its premium shared filesystem volumes, logging, block storage volumes, network, bare metal, networking, and cloud console—remained operational. The status page listed six incidents over the past 30 days and 12 maintenances over the same period, reflecting an active operational period for the AI cloud provider.

Separately, CoreWeave detailed a planned network maintenance window running from Monday, October 19, 2026, at 12:00 UTC to Friday, October 23, 2026, at 00:00 UTC, with work conducted during daytime hours each day. The company said it would apply a vendor-recommended firmware update to the boot drives of network switches across several availability zones, including US-CENTRAL-09A, US-CENTRAL-08A, US-EAST-04B, US-CENTRAL-10A, US-EAST-12A, and US-CENTRAL-08B. According to CoreWeave, the update resolves a known issue that can cause a switch to become unresponsive and improves overall network reliability, with no impact expected as the update is applied live without switch reboots or node drains.

CoreWeave noted that in rare cases a switch may require a brief restart after the update, and where this affects backend GPU-to-GPU network switches, a small number of nodes may temporarily lose part of their backend network bandwidth until the restart completes, potentially affecting multi-node jobs running on those nodes. Engineers and on-site technicians will monitor throughout the window, and affected customers will be contacted directly, with no customer action required.

In addition, CoreWeave will perform scheduled maintenance to update CNI networking in CoreWeave-managed clusters in US-EAST-04 beginning Monday, October 19, 2026, at 9:00 AM ET (1:00 PM UTC) and continuing through Friday, October 30, 2026, at 5:00 PM ET (9:00 PM UTC). The company said it would actively monitor the situation during this period and directed customers with questions to its support team. CoreWeave classified the activity as Planned Maintenance under its Terms of Service Maintenance Policy. A further scheduled maintenance was listed for October 26–29, 2026, affecting CoreWeave Infrastructure in the CA-EAST-01 location, along with a global CoreWeave Cloud maintenance window on October 26.

The operational notices underscore the scale and complexity of maintaining the high-performance networking and storage infrastructure that underpins CoreWeave's GPU cloud services, which are increasingly relied upon for AI training and inference workloads. As AI demand continues to surge, the reliability of backend GPU-to-GPU networking and storage systems has become a critical concern for cloud providers serving large-scale model developers, and even brief disruptions can ripple across multi-node training jobs. CoreWeave's proactive firmware updates and transparent status communications reflect the industry's growing emphasis on operational resilience as AI infrastructure scales.