I'd rather see a benchmark of a more real-world case. These hello world microbenchmarks don't show anything.
True, these hello world benchmarks are not indicative of real world performance, but suspect React would be significantly slower than node.js at a hello world showdown.
It's useless for absolute perf but to compare 2 solutions doing the same job, it's a good start.