Q4 2026: 1,651 incidents on public status pages
Preview. This covers October 1 to 2026-10-21 22:03 UTC, and every figure changes until the quarter ends on December 31. The published report will be a frozen snapshot.
- Vendors with an incident
- 252 of 368
- Vendor said why
- 25%
- Named a major cloud or AI provider
- 6%
- At AI vendors (23% of those tracked)
- 19%
How long incidents lasted
| Tier | Incidents | Per tracked vendor | Median to mitigation | Median to resolution | p95 to resolution |
|---|---|---|---|---|---|
| Infrastructure (78 tracked) | 496 | 6.4 | 53m | 1h 32m | 29h 46m |
| Apps and SaaS (206 tracked) | 838 | 4.1 | 1h 4m | 2h 27m | 2d 12h |
| AI (84 tracked) | 317 | 3.8 | 22m | 18m | 9h 9m |
Mitigation is the first "monitoring" update, and only counts for vendors whose pages have that stage.
What failed
1565 incidents tagged by claude-haiku-5-5 from their updates. "Not enough information" means the vendor described symptoms only.
Why, when the vendor said
Knock-on incidents
- Google Cloud: Multiple services are experiencing issues in the europe-west2 region. 2026-10-20 06:08 UTC, 12 vendors followed
- Google Cloud: Google Cloud Storage (GCS) customers experienced elevated latencies and error rates when performing upload or write operations. 2026-10-08 11:00 UTC, 6 vendors followed
- GitHub: Delays for Pull Requests and Webhook deliveries 2026-10-21 14:52 UTC, 2 vendors followed
- GitHub: Incident with Actions, Pull Requests and Issues 2026-10-12 12:21 UTC, 2 vendors followed
Longest incidents
| Vendor | Incident | Duration |
|---|---|---|
| Grafana Cloud | IRM Unable to Send SMS to Thailand Numbers | 13d 19h |
| Cloudways | Degradation Impacting Vultr Server Operations | 13d 5h |
| GeForce NOW | Customers are unable to start sessions | 8d 4h |
| Notion | Degraded performance in Notion and the Notion API | 8d 1h |
| Crypto.com | Crypto.com Main App Mission Reward Daily Check In | 7d 22h |
| Astronomer | Capacity constraints in GCP us-central1 may delay or fail some task executions | 7d 21h |
| Circle | XDC Testnet network outage | 7d 8h |
| UpCloud | [SG-SIN1] Managed Object Storage Issue | 7d 7h |
| Retool | Elevated Latency on Query Execution (Retool Cloud) | 7d 6h |
| Mercury | Failed Investments Wire Transfers Out for Some Customers | 6d 17h |
Read this with care
Every figure comes from what vendors chose to post. A vendor that posts every blip will show more incidents than one that posts only big outages, so this report does not rank vendors. The vendor list was built to cover AI providers widely, so the AI share of incidents follows that list and is not an industry-wide figure; compare tiers by incidents per tracked vendor instead. AI failures inside other products rarely reach a status page at all.
For 4 vendors, the recorded history starts after October 1, because their status pages no longer show earlier incidents. Azure posts reviews days after an incident, so its latest incidents arrive late. A vendor posting the same title again within an hour counts once. See the methodology for the rest.
Download the 1,651 incidents in this report (CC BY-NC 4.0).