Concentrated infrastructure keeps failing in public
A Google Cloud outage took down HSBC, Revolut, Bank of Scotland, Lloyds and Equifax. A Revolut user had no account access, no withdrawals and no card payments, with no phone number to call. One comment: 'Countries talking about AI sovereignty, when they forgot compute sovereignty.' An earlier thread showed AWS billing glitches where hobby accounts saw bills from $286M to $284 billion, and people got real scares. Another thread showed Docker pull failing in Spain because Cloudflare IPs get blocked during La Liga football matches.
The same theme shows up lower in the stack. French supermarket fridges, made by Danfoss, were exposed online for months. One commenter in grocery software says the industry is decades behind. Nobody noticed hundreds of fridges suddenly entering defrost mode.
The pattern: a few providers sit under everything, regulators and courts hit them in odd ways, and the people at the edge, whether a bank customer or a grocer, have no fallback. Nothing in these threads says multi-cloud is easy, only that single-provider dependence is now a visible risk.
So what?
Write down your single points of failure, especially for anything customers touch during a crisis like login, payments and support contact. Have a non-cloud channel, like a phone line or status page hosted elsewhere, because 'no one to call' is the complaint that sticks.
Read these
Google cloud outage takes down banks
AWS: Inaccurate Estimated Billing Data – $1.7 billion
Tell HN: Docker pull fails in Spain due to football Cloudflare block
France's supermarket fridges were vulnerable online for months