When the cloud goes down, does your plant keep running?

Two years of outages

Here’s a short list. Every item on it stopped real work, not only websites.

• July 19, 2024. A faulty CrowdStrike update crashed about 8.5 million Windows computers. Delta alone cancelled about 7,000 flights.
• September 2025. Ransomware hit the check-in software that Collins Aerospace runs for airports. Heathrow, Brussels and Berlin went back to manual check-in.
• October 20, 2025. A DNS fault in one AWS region caused about 15 hours of failures for the services that depend on it.
• October 29, 2025. A bad configuration change in Microsoft’s Azure Front Door caused about eight hours of failures. Alaska Airlines and Hawaiian Airlines were among those hit.
• November 18, 2025. An internal bug at Cloudflare took a large part of the web offline for hours. No attacker was involved.

These aren’t all the same kind of failure. One was a bad update, one was an attack on a vendor, and three were plain cloud outages. What they share is this: the people who stopped working had done nothing wrong and could do nothing to fix it. They waited.

Why almost everything is cloud-only

Cloud is great for the company that sells the software. There’s one copy to run, one place to update, it scales easily, and the bill repeats every month. I know, because vMaint runs in the cloud today, for exactly those reasons.

It’s good for the buyer too, on day one. Nothing to install, nothing to back up, and it works from any phone. For a lot of shops that’s the right trade.

But a plant isn’t an office. If email is down for three hours, people grumble. If the system that holds your work orders, PM schedule, spare parts and downtime log is down for three hours on second shift, the floor goes back to paper and memory.

What the 90s got right

Software used to come on a CD-ROM. You installed it and it ran. It didn’t ask a server in another state for permission to open, and if the internet was down you didn’t notice. Anyone who has worked in a plant has seen a machine still running on a PC from that era.

The 90s got plenty wrong too. No updates, no backups unless someone remembered, and the knowledge of how it all worked left with whoever installed it. Nobody should want that part back.

Local isn’t safe by default

This is the part most cloud-versus-local posts skip. The plant shutdowns that made the news in 2025 weren’t cloud outages. Jaguar Land Rover stopped production for about five weeks after a cyberattack. Asahi halted most of its 30 plants in Japan after ransomware. Both were attacks on the companies’ own systems.

Sophos asked manufacturers hit by ransomware in 2025 how the attackers got in. The top answer, at 32 percent, was an exploited vulnerability. In plain words: something nobody had patched.

So if software runs on your own PC, these jobs are yours:

• Backups, and a restore test. A backup nobody has restored is a hope, not a backup.
• Updates and security patches, on a schedule.
• The hardware. Disks die, and the spare should already be on the shelf.
• Access. Who can reach it, from where, and who gets the call at 2 a.m.

A cloud vendor does all of that for you. That’s the real thing you pay them for.

What changed: the upkeep got cheaper

Here’s my bet. I’m calling it a bet because I don’t have a study that proves it, and the research on AI and productivity is mixed.

Small plants didn’t give up on local software because of the software. They gave up because of the upkeep. Somebody had to script the backups, read the logs, plan the update, and write down how it all works.

AI tools are good at that kind of work now. An engineer who knows the plant can ask for a backup script, a restore drill or an update plan and get a solid draft in minutes. A person still has to review it and own it. But it no longer takes an IT department.

The AI itself can run on site too. Small models now fit on one PC with a graphics card, so plant data doesn’t have to leave the building to be useful.

Local first, cloud as an extra

I don’t think the choice is cloud or local. It’s about which one comes first.

• The floor comes first. Work orders, PMs, parts and downtime run on a PC on the plant network, and keep running with the internet unplugged.
• The cloud is an extra. Use it for off-site backup, remote access, and reports for people outside the building.
• The data is yours. It sits in a database you can open, copy and export without asking anyone.

A good test for any plant system: unplug the internet for a day and see what stops.

Where vMaint is going

vMaint is cloud today, so this post is about my own product too. I’m building the local install now. The goal is the same app on a PC on your plant network, with backups and updates treated as part of the product, not left as homework.

Local install is coming soon.

Get notified →