In May, we published a deep technical breakdown of a fatal AWS Lambda bug — Node.js functions inside a VPC making outbound HTTPS calls would silently crash after returning a 201. No logs, no stack trace, no errors. AWS eventually reproduced the issue internally.

This post isn't about the bug.

It's about what AWS did next.

They blamed our code — after confirming the crash themselves.

They revoked a previously offered $4,000 credit — without explanation.

They charged our founder’s personal card mid-dispute — with no invoice or warning.

They routed technical escalations through sales — not engineering.

They re-offered the same credit — now as a settlement contingent on silence.

And they privately asked us not to publish.

We published anyway.

This follow-up post is about what happens after you prove a platform failure in AWS — and why the support model, not the runtime, is the real risk.

Read the follow-up post here