Internet & cloud outages — March 2026

X (formerly Twitter) is experiencing a global outage today, with users across multiple countries reporting that feeds are not loading, login failures, and inability to post content. Reports surged on Downdetector throughout the day. No official statement from X has been released yet. The X status page (status.x.com) is also unreachable. Cloudflare is undergoing scheduled maintenance at its Zurich (ZRH) datacenter (11:00–17:00 UTC), which may cause slight latency increases for regional users. Cloudflare also resolved a Network Error Logging issue and is monitoring a minor Spectrum Analytics dashboard display bug — neither affects actual traffic or protection. All major cloud providers are operating normally: AWS, Google Cloud, and Microsoft Azure report no active incidents. GitHub and OpenAI are fully operational. No BGP routing incidents have been reported today. Looking back at the past week, ThousandEyes tracked 277 global network outage events (March 16–22), a slight 2% increase from the prior week. Notably, Tier 1 provider Arelion (formerly Telia Carrier) experienced an outage on March 20 that impacted customers across 20+ countries — that incident has since been resolved. Overall, the internet is up and functioning normally for the vast majority of users. The X outage is the most significant incident today but is isolated to a single platform.

The internet is largely stable today with no widespread outages or major infrastructure disruptions. Cloudflare is experiencing several minor service issues. Users may encounter intermittent difficulties with Challenge Pages and Turnstile verification, though a fix has been deployed and is being monitored. Cloudflare is also investigating intermittent data gaps in Network Error Logging (no customer traffic impact) and a display issue with Schema Validation block reasons. Scheduled maintenance is underway or planned today at datacenters in Atlanta, Palermo, Singapore, Cape Town, and Paris — some users in those regions may notice slightly increased latency. OpenAI reports a degraded service issue with ChatGPT file uploads and downloads, which has been ongoing for approximately 9 hours. A mitigation is being implemented. OpenAI APIs remain operational at 99.99% uptime. All major cloud platforms — AWS, Google Cloud, and Microsoft Azure — report normal operations with no active incidents. No BGP routing anomalies or DNS infrastructure issues have been detected. Looking back at recent days: on March 22, Refinitiv experienced a 37-minute outage affecting users in the US, Spain, Switzerland, and Japan. On March 20, Arelion (a global Tier 1 backbone provider) had a 1-hour-38-minute outage impacting multiple regions. Both incidents have been fully resolved. Overall, the internet is up and functioning normally. The Cloudflare and OpenAI issues are localized and do not affect general internet connectivity.

The internet is operating normally overall on March 24, 2026, with a few minor provider-level incidents that do not affect broad connectivity. Cloudflare is experiencing increased errors on its Workers API, with changes to Cron Triggers also delayed. Already-deployed Workers remain unaffected. Cloudflare has also identified a dashboard bug where schema validation block reasons are not displayed; a fix is being implemented. Additionally, Cloudflare is conducting scheduled maintenance across multiple datacenters including Singapore, Taipei, Osaka, Bogota, and several European and US locations through March 27. OpenAI reported elevated error rates on its Realtime API SIP endpoints starting around midnight Pacific time. The team identified the root cause, applied a mitigation, and is monitoring recovery. Core OpenAI services including ChatGPT and standard APIs remain operational with high uptime. All three major cloud providers — AWS, Google Cloud, and Microsoft Azure — report no active incidents. Fastly and Akamai CDN services show no disruptions. No BGP routing incidents or DNS issues have been detected. Localized ISP outage reports were noted in Houston, TX and Phoenix, AZ, but these appear to be routine regional issues rather than systemic problems. Overall, the internet remains stable and fully operational for the vast majority of users worldwide.

No major internet-wide outages or disruptions detected today. All major cloud providers are reporting normal operations. AWS, Google Cloud, and Microsoft Azure show no active incidents. OpenAI services are fully operational. GitHub resolved a brief Git operations latency issue for west coast users overnight (resolved at 00:58 UTC). Cloudflare is conducting scheduled maintenance at its Montréal (YUL) data center through 12:00 UTC today, with additional maintenance windows planned for Los Angeles (LAX) later today and several European and Asian locations next week. An HTTP 530 error spike affecting traffic routed through Atlanta (ATL) on March 19 was resolved within minutes. Akamai resolved a Certificate Provisioning System issue in their control center on March 19. Fastly reports no active incidents since a regional CDN performance issue in India on March 4. No BGP routing incidents or DNS infrastructure issues were detected. Looking back at recent days: the most notable event in March remains the AWS ME-CENTRAL-1 (UAE) data center incident on March 1-2, which involved structural fires and emergency power shutdowns. That incident has been fully resolved. A Cloudflare WAF incident on March 3 that broke challenge actions globally has also been resolved. Overall, internet infrastructure is stable and operating normally today.

The internet is functioning normally overall today, with only minor, localized incidents to report. Cloudflare is currently investigating increased HTTP 530 errors affecting traffic routed through its Atlanta (ATL) data center, reported as of 11:36 UTC on March 19. Additionally, Cloudflare has scheduled maintenance windows in progress at its San Jose (SJC) and Bogota (BOG) facilities, with several other data centers in Africa, Latin America, and parts of Asia experiencing partial outages or maintenance. These are localized and not affecting global internet connectivity. Yesterday (March 18), X (formerly Twitter) experienced a brief but widespread outage lasting approximately one hour, from around 10:47 AM ET to 11:44 AM ET. Over 26,000 users in the US reported issues including blank timelines, login failures, and inability to load posts. The platform has fully recovered and no cause has been publicly disclosed. All major cloud providers are reporting normal operations: - AWS: No active incidents (the major UAE region outage from March 1-2 is fully resolved) - Google Cloud: All services operational, no broad severe incidents - Microsoft Azure: All services operational, no active events - OpenAI: Fully operational across all systems No BGP routing incidents, DNS failures, or undersea cable issues have been reported today. Overall, the internet remains healthy and stable. The Cloudflare Atlanta investigation is worth monitoring but has limited impact.

Cloudflare is experiencing an active incident involving increased HTTP 520 errors, specifically affecting HEAD requests. The issue was identified at 09:53 UTC on March 18, and a fix is currently being implemented. Separately, Cloudflare is performing scheduled maintenance at its LAX (Los Angeles) datacenter through March 19 at 00:00 UTC, which may cause slight latency increases for traffic routed through that facility. All other major cloud and infrastructure providers are reporting normal operations: - AWS: No active incidents - Google Cloud: All services operational - Microsoft Azure: All services operational - OpenAI: All systems operational - GitHub: No reported issues No new BGP routing incidents have been detected today. The most recent BGP-related events were a Cloudflare BYOIP route withdrawal on February 20 and a route leak on January 22, both of which have been fully resolved. On the local level, the City of Midland, Texas reported an internet outage affecting some city services on March 17, which may still be ongoing. Overall, the internet is functioning normally with a minor, limited-scope incident at Cloudflare that is actively being addressed.

A few localized incidents are affecting services today, though core internet infrastructure remains stable. Cloudflare is monitoring degraded performance on its Workers platform after an increase in errors was detected on March 16. A fix has been implemented and is under observation. Additionally, Cloudflare has scheduled maintenance at its Los Angeles (LAX) datacenter today between 09:00 and 15:00 UTC, which may cause brief latency increases for some users. OpenAI is experiencing elevated errors with ChatGPT sign-in and account creation, an issue that began at 02:42 UTC today. A mitigation has been applied and the team is monitoring recovery. API services remain unaffected. Verizon users reported connectivity issues beginning around midnight Eastern Time, with thousands of reports on outage tracking sites. The scope and resolution status remain unclear. An Xfinity outage that disrupted internet service across the San Francisco Bay Area on March 16 has been resolved, with most users reporting restored connectivity. All three major cloud providers - AWS, Google Cloud, and Microsoft Azure - report normal operations with no active incidents. No BGP routing anomalies or DNS infrastructure issues have been detected. Overall, the internet is up and functioning normally. Todays incidents are localized and do not indicate any systemic or widespread disruption.

A brief but widespread Xfinity outage hit the San Francisco Bay Area this morning, with reports of internet dropouts stretching from the North Bay to the South Bay shortly after 9 AM local time. Service appeared to recover before 10 AM, though Comcast did not provide a root cause. Downdetector also logged minor spikes for AT&T and AWS in the same timeframe, suggesting possible upstream disruption. Cloudflare experienced increased error rates for Workers scripts starting at 21:41 UTC on March 16. A fix was identified and implemented, and the team is monitoring the resolution. Separately, Cloudflare is performing scheduled maintenance at its Hamburg (HAM) datacenter from 01:00 to 05:00 UTC on March 17, which may cause slight latency increases for users routed through that facility. Microsoft 365 saw approximately 160 user-reported issues over the past 24 hours, though no formal incident was declared by Microsoft. All major cloud providers — AWS, Google Cloud, and Azure — report normal operations. GitHub, OpenAI, and Fastly show no active incidents. No BGP hijacks or routing anomalies have been reported. Overall, the internet is functioning normally with only localized and largely resolved disruptions.

No major internet-wide outages or disruptions detected today. Cloudflare is performing scheduled maintenance at its LAX (Los Angeles) datacenter between 07:00 and 15:00 UTC. Traffic is being re-routed, with possible slight latency increases for users in the affected region. All core Cloudflare services (CDN, DNS, Workers, API) remain fully operational. OpenAI is monitoring an ongoing issue with ChatGPT file downloads. A mitigation has been applied and recovery is being tracked. The ChatGPT API and other OpenAI services are unaffected. In recently resolved incidents: Anthropic Claude AI experienced an outage on March 11 lasting approximately 2 hours and 45 minutes, affecting Claude.ai and Claude Code users with login issues and degraded performance. The Claude API was not impacted, and full service was restored the same day. Azure OpenAI Service had a platform issue from March 9-10 that has also been fully resolved. AWS, Google Cloud, and Microsoft Azure report no active incidents. No BGP routing anomalies or DNS infrastructure issues have been detected. Internet backbone connectivity remains stable worldwide. The internet is up.

The global internet infrastructure is operating normally today with no widespread outages or major disruptions. OpenAI is experiencing three ongoing incidents affecting ChatGPT: file upload errors, file download errors, and Codex unresponsiveness. The team has applied mitigations and is monitoring recovery. These issues have persisted for approximately 12-24 hours. Cloudflare is conducting scheduled maintenance at its Los Angeles (LAX) datacenter today (07:00-16:30 UTC) and Seoul (ICN) datacenter (17:00-21:00 UTC). Traffic is being rerouted with possible slight latency increases in affected regions. Several minor Cloudflare incidents from March 10 — including Zero Trust Dashboard UI issues, Access configuration delays, and Challenge Page errors — have all been resolved. All major cloud providers are operating normally: AWS, Google Cloud, Microsoft Azure, and Azure DevOps report no active incidents. The significant AWS outage in the UAE region (ME-CENTRAL-1) from March 1-2 and the Oracle Cloud outage that affected TikTok on March 3-4 are fully resolved. No BGP routing incidents, DNS failures, or undersea cable issues have been detected. Overall internet health remains strong.

The internet is largely stable today with no major outages affecting global connectivity. OpenAI is experiencing an ongoing issue with its Codex service, which has been unresponsive since March 9. The team has identified a likely root cause and is actively implementing mitigations. Other OpenAI services including ChatGPT and the API remain operational. Discord suffered a significant global outage on March 9 that lasted approximately 50 minutes, with millions of users unable to send or load messages. The issue was resolved the same day. Cloudflare is performing scheduled maintenance across several datacenters including Frankfurt (09:00-11:00 UTC), Los Angeles (07:00-15:00 UTC), and Sao Paulo (06:00-12:00 UTC). Some smaller points of presence in Africa, Asia, and South America are showing partial degradation, likely related to traffic rerouting during maintenance. All three major cloud providers — AWS, Google Cloud, and Microsoft Azure — report fully operational status. GitHub is operational following minor Webhooks and Codespaces issues on March 9 that have since been resolved. Fastly and Akamai report no active incidents. No BGP routing incidents or DNS infrastructure issues have been reported today. Overall, the internet is up and functioning normally with only isolated, service-specific incidents.

No major internet-wide outages or disruptions detected today. Google Cloud reports all services operational with no active incidents. Cloudflare is performing scheduled maintenance at its Portland (PDX) datacenter (14:00–23:59 UTC) and London (LHR) datacenter (23:00 UTC March 9 – 07:00 UTC March 10), with some partial outages at smaller edge locations in Africa, Asia, and Latin America — typical for edge points of presence and not impacting core services. AWS and Azure report no official incidents, though Azure OpenAI saw 57 user-submitted outage reports in the past 24 hours without an official acknowledgment. GitHub is operational with 26 user reports from the past 24 hours now resolved. No active BGP routing incidents were detected. The most recent notable BGP events — a Cloudflare route leak on January 22 and a Verizon route leak in February — are fully resolved. Virgin Media in the UK saw some user-reported disruptions earlier today, but no widespread ISP outages are confirmed. Overall, the internet is functioning normally with only routine maintenance activities underway.

The global internet is largely stable today with no widespread outages. The most significant ongoing incident is the AWS me-central-1 (Middle East) outage that began on March 1 after drone strikes physically damaged data center infrastructure in the UAE. Two availability zones remain impaired, with DynamoDB, S3, and EC2 networking APIs experiencing elevated error rates and latency. AWS continues recovery efforts and is advising affected customers to migrate workloads to alternate regions. This remains a regional incident with no impact on other AWS regions. Cloudflare is reporting minor partial outages across several regional data centers, including locations in Africa (Dar es Salaam, Djibouti, Kinshasa, others), Asia (Male, Naha, Kochi), Europe (Geneva, Skopje), and North America (Charlotte, Norfolk, Ottawa). These appear related to maintenance windows and are causing slight latency increases rather than full service disruptions. All other major infrastructure providers are reporting normal operations: - Azure: No active incidents - Google Cloud: All services operational - Fastly: Operational - GitHub: Operational (Copilot degradation on March 5 resolved) - OpenAI: Fully operational Recently resolved incidents from the past 72 hours include a Cloudflare GraphQL API outage on March 4 (DNS metrics queries affected for ~5 hours) and a Claude.ai disruption on March 2. Overall classification: incidents-detected. Notable regional issues persist in the Middle East, but internet infrastructure globally is functioning normally.

The most significant ongoing event remains the AWS Middle East outage, now entering its fifth day. Drone strikes on March 1-2 physically damaged AWS data centers in the UAE (me-central-1) and Bahrain, knocking out EC2, S3, RDS, and dozens of other services across multiple Availability Zones. AWS has stated that recovery will be prolonged due to the nature of the physical damage and is recommending customers migrate workloads to alternate regions. Financial services, delivery platforms, and enterprise software providers in the region continue to experience disruptions. Elsewhere today, Cloudflare is performing scheduled maintenance at its Chicago (ORD) and Toronto (YYZ) data centers, with possible slight latency increases for traffic routed through those locations. Cloudflare RealtimeKit is also experiencing degraded performance on Android. Amazon.com (retail) saw a separate outage earlier today with over 20,000 user reports, affecting checkout, product listings, and order tracking for roughly five hours. GitHub experienced a 3-hour outage on March 5 affecting pull requests, but appears operational today. An Oracle Cloud outage earlier this week briefly knocked TikTok offline for some US users. Google Cloud, Azure, and OpenAI all report normal operations. No new BGP routing incidents or undersea cable issues have been reported today. The internet is up and broadly functional, though the AWS Middle East situation remains a significant regional infrastructure event.

Several notable incidents are affecting internet services today, though no widespread internet-wide disruption is occurring. Cloudflare is experiencing partial outages across multiple regions including parts of Africa, Asia, Latin America, Europe, and select US cities (Charlotte, Norfolk, San Antonio). Scheduled maintenance is also underway in Istanbul and planned for Toronto, Chicago, and Portland in the coming days. OpenAI reported issues with ChatGPT users being unable to send messages starting around 18:12 UTC. After investigation, mitigations were applied and the team is monitoring recovery. Earlier in the day, elevated error rates hit the Realtime API in the EU region but have since resolved. Oracle Cloud experienced an outage on March 3-4 that knocked TikTok offline for US users. This was the second Oracle data center failure affecting TikTok in roughly a month, both involving the US East (Ashburn) region. Oracle confirmed recovery by 09:24 UTC on March 4. GitHub shows some third-party reports of service degradation, though the official status page lists no active incidents. AWS, Google Cloud, and Azure are all reporting normal operations. The AWS Middle East (me-central-1) physical incident from March 1 appears fully resolved. Overall, the internet is up with some service-specific disruptions. No BGP routing incidents or DNS infrastructure issues were detected today.

Several cloud and CDN providers are experiencing localized issues today, though no widespread internet-wide disruption has been detected. GitHub is reporting degraded availability for Git Operations, with recovery underway as of 19:36 UTC. Cloudflare has partial outages across multiple regions including parts of Africa, Asia, Latin America, and two North American locations (Charlotte and Norfolk), alongside scheduled maintenance in several datacenters. A Cloudflare Zero Trust Gateway issue was identified and resolved earlier today. Fastly is experiencing a partial outage affecting its Singapore and Tokyo points of presence, plus several other components. Akamai reports packet loss in the India region attributed to cable restoration work by a third-party provider. Google Cloud and Microsoft Azure are reporting no active incidents. No significant BGP routing issues have been detected. In the past 72 hours, several notable incidents were resolved: AWS me-central-1 suffered a physical infrastructure incident on March 1 that disrupted the Middle East region. Anthropic Claude experienced a major outage on March 2 affecting its web interface and authentication. OpenAI resolved Sora API errors on March 2. GitHub also resolved a 2.9-hour degraded performance issue with Pull Requests on March 2. Overall, the internet is functioning with scattered provider-specific issues but remains broadly operational.

No major internet-wide outages are currently detected. The core infrastructure — AWS, Google Cloud, Azure, Fastly, and GitHub — are all reporting normal operations. OpenAI is experiencing increased authentication failures affecting some ChatGPT users, which began at 06:10 UTC today. A mitigation has been applied and the team is monitoring recovery. Cloudflare has two minor lingering issues: elevated HTTP request latency at its Newark (EWR) datacenter affecting a subset of requests (a fix was deployed on Feb 26 and is being monitored), and an authentication issue with the Cloudflare Peering Portal due to a third-party auth provider outage (identified Feb 27, does not affect CDN or edge security). Several Cloudflare datacenters have scheduled maintenance windows in the coming week, including Bogota (Mar 3), Tallinn (Mar 3-4), Lisbon (Mar 4), Sao Paulo (Mar 5), Toronto (Mar 6), and Portland (Mar 9-23). Looking back at February, the most significant incident was the Cloudflare BYOIP outage on February 20, where 25% of Bring Your Own IP prefixes were unintentionally withdrawn via BGP for over 6 hours. Cloudflare also experienced a BGP route leak on January 22 affecting IPv6 traffic for approximately 25 minutes. Overall, the internet is up and functioning normally with only minor, localized service disruptions.