Most lines of code aren't necessarily async/await calls, so the benchmark results might be negligible, because they don't represent realworld usecase. I don't predict it would do anything for most Node.js applications, but I would personally feel more comfortable using it once it's properly implemented, there's no reason not to, otherwise it feels like a compromise, looking at the benchmarks. Here's a PR in progress that addresses the performance issues, hopefully it will be ready soon: https://github.com/nodejs/node/pull/48528