12 tracked incidents · 2025–2026
Incident tracker
Major outages and failures across data centers and clouds — factual, dated, sourced. Want an email when something big breaks? Set an alert.
2026-07-15
Google Cloud europe-west4 outage after power and cooling failure
Google Cloud · Eemshaven, Netherlands (europe-west4-a)
A utility power disturbance triggered protective shutdowns of power and cooling systems in one zone, causing a cascading failure across multiple Google Cloud services. The incident lasted roughly 15 hours before full recovery.
2026-06-05
Fire at STT GDC/Tata Next-Gen Tower data center in Delhi
ST Telemedia Global Data Centres India / Tata Communications · Greater Kailash, New Delhi, India
A fire that started in a lithium-ion UPS battery room caused extensive damage to the STT Delhi 2 facility, injured two firefighters, and disrupted Google Cloud, Netflix and local ISPs across Delhi-NCR. Google Cloud reported degraded network capacity in the Delhi metro for about three weeks, and some colocation clients feared permanent data loss.
2026-05-07
Fire at NorthC data center takes IBM Cloud Amsterdam offline
NorthC Datacenters (tenant: IBM Cloud) · Almere, Netherlands
A fire broke out at the NorthC Almere facility at around 08:45 and took emergency services until the evening to bring under control, cutting power to the site. IBM Cloud's AMS03 data center went offline, and Utrecht University systems, a regional water board and Transdev bus/tram services were among those affected; restoring redundant power took about a week.
2026-03-01
AWS me-central-1 outage after drones strike UAE data centers
Amazon Web Services · United Arab Emirates (me-central-1)
AWS reported that 'objects' — later confirmed as drones amid regional attacks — struck two availability-zone data centers, causing sparks and fire; fire crews cut power to one facility and its generators. With two AZs down, more than 100 services in the me-central-1 region were disrupted as the remaining zone became overloaded.
2026-02-20
Cloudflare BYOIP outage withdraws customer routes via BGP
Cloudflare · Global
A bug in Cloudflare's Addressing API during an automated cleanup task caused about 25% of Bring-Your-Own-IP prefixes worldwide to be withdrawn from BGP, making affected customers unreachable from the internet. The incident lasted 6 hours and 7 minutes; Cloudflare confirmed it was not an attack.
2025-11-28
CyrusOne CHI1 cooling failure halts CME futures trading
CyrusOne (tenant: CME Group) · Aurora, Illinois, US
A chiller plant failure at CyrusOne's CHI1 data center on the evening of November 27 caused server rooms to overheat, forcing protective shutdowns. CME Group halted trading across its global futures and FX markets for roughly 10 hours on November 28, freezing about 90% of global listed derivatives trading.
2025-11-18
Cloudflare global outage from oversized bot management feature file
Cloudflare · Global
A database permissions change caused Cloudflare's Bot Management feature file to double in size, exceeding a hard limit and crashing the core proxy software across the network. Large parts of the web returned 5xx errors for several hours, affecting sites and apps including X, ChatGPT and Spotify.
2025-10-29
Azure Front Door configuration error disrupts Microsoft services globally
Microsoft Azure · Global
An inadvertent configuration change deployed to the Azure Front Door platform produced DNS and routing failures from about 16:00 UTC, taking down or degrading the Azure Portal, Microsoft 365 web apps, Xbox/Minecraft authentication and thousands of customer sites. Microsoft rolled back to a last-known-good configuration, with recovery declared after roughly 8 hours.
2025-10-20
AWS us-east-1 outage after DynamoDB DNS automation failure
Amazon Web Services · Northern Virginia, US (us-east-1)
A latent race condition in DynamoDB's automated DNS management wiped the service's DNS records in us-east-1, cascading into roughly 15 hours of disruption across dozens of AWS services and major dependent platforms including Snapchat, Fortnite, Slack, Ring and Coinbase. It was one of the largest cloud outages on record, with over 17 million user reports on Downdetector.
2025-09-26
Battery fire at South Korea's national government data center
National Information Resources Service (NIRS) · Daejeon, South Korea
A lithium-ion battery fire that broke out during UPS battery relocation work at the NIRS facility knocked more than 600 government digital services offline, including the Government24 portal and postal services. Full restoration of affected systems took weeks, prompting a national review of battery safety and government cloud resilience.
2025-06-12
Google Cloud global outage from Service Control crash
Google Cloud · Global
An invalid automated quota policy update triggered a null-pointer crash loop in Google's Service Control API management binaries worldwide, as the failing code path lacked error handling and feature flags. Dozens of Google Cloud and Workspace services failed for up to about 3 hours (with us-central1 recovery taking longer), also disrupting Cloudflare, Spotify and other dependent platforms.
2025-05-22
Battery room fire at Oregon data center leased by X
Digital Realty (tenant: X Corp) · Hillsboro, Oregon, US
A fire ignited in a lithium-ion backup battery room at the PDX11 facility in Hillsboro Technology Park, prompting a building evacuation and a lengthy firefighting response. X (Twitter) suffered a widespread outage, with degraded login, messaging and timeline functionality persisting into the following days.