CloudFormation Express mode can make deployments up to four times faster, but it does that by changing what “finished” means. That is useful for development loops and agent-driven infrastructure work. It is also exactly why I would not switch it on everywhere without thinking about what runs immediately after the deploy.
AWS explains that Express mode completes when resource configuration has been applied instead of waiting for extended stabilisation checks. Resources keep becoming operational in the background. It works with existing templates, the CLI, SDKs, and CDK through cdk deploy --express, with no additional charge in commercial regions.
For a disposable stack or a tight development loop, that sounds genuinely useful. Waiting a minute for a queue and dead-letter queue on every iteration is just friction. The caveat is that Express mode disables rollback by default and a successful stack operation no longer proves every resource is ready to serve traffic. Any pipeline that deploys and immediately runs integration tests, shifts traffic, or announces success needs its own readiness checks.
I would use Express mode where speed improves feedback and the failure is cheap. I would keep the standard behaviour where CloudFormation’s wait is part of the safety case, or add explicit monitoring and rollback before using Express in production. Faster is good, but only when the next command understands what it was promised.