Microsoft says Azure now spots failing hardware early and cut hardware-related reboots by more than 36%. Businesses running key software there should check their alerts.
Microsoft says its Azure cloud now tries to catch failing computer hardware before it knocks your software offline. The system is called Azure Failure Prediction and Detection, or AFPD. Microsoft reports it has cut hardware-related reboots across Azure by more than 36%. That figure comes from Microsoft itself.
The old pattern was familiar on any cloud. A part wears out, your software stops, an engineer investigates, and then a reboot or a hardware swap follows. AFPD aims to break that chain. It uses machine learning and telemetry, which is data the machines report about their own health, to predict failures. Then it acts on them. It became fully operational in 2024.
Microsoft says the goal is that the interruption never reaches you. Eligible workloads can be moved off a struggling machine, and customers are notified when they need to act. The company also says its approach won the Best Paper Award at the 2025 IEEE Cloud Summit.
Microsoft points to SAP, the finance and supply chain software many companies run on. These systems often hold the official records for money, orders and operations. Even a short outage can hurt. That makes early warning more valuable.
Microsoft also admits a gap. Many SAP customers do not yet know about the alert and automation options they can use. Three of the channels it names are Resource Health, Resource Graph and Event Grid. Microsoft says the content of the alerts is the same in all three. The choice comes down to scale and how you want to connect them to your own tools.
SAP is not the only beneficiary. Microsoft says AFPD runs across several kinds of Azure computing, plus some storage:
Microsoft says this is not a premium tier and does not need to be set up at the infrastructure level. It runs automatically across Azure. Anyone with software that cannot afford downtime has a reason to care.
If your business runs key software on Azure, ask your IT team or provider which alert options are turned on. The system works on its own, but the notices only help if someone sees them. Microsoft says automated responses are practical for companies running at scale.
The 36% figure is Microsoft's own, and the post gives no outside check. It does not say how many failures are predicted correctly, or how often a workload still goes down. The fourth alert channel is also not named in the material available here.
If key software runs on Azure, ask which failure alerts are switched on, because the early warnings only help if someone is set up to receive them.
Major environmental groups now use AI to study climate data and track fires, while warning that the data centers behind AI carry their own environmental cost.
Read more: Environmental groups use AI to fight climate change, despite its energy costReports say the Trump administration is weighing underwater datacenters off Maine and Alaska, and Lake Erie groups want a plan before anyone proposes one nearby.
Read more: Underwater datacenters may be next, and Lake Erie groups are worriedVoters in Upper Sandusky, Ohio, banned large data centers after a $900 million plan. Here is how a town of 6,600 did it, and what other places can learn.
Read more: Small Ohio town votes to ban a $900 million AI data center
