It's a sliding scale of return. If your infrastructure is already decent, you will see a terrible return on your investment, as in the case of Japan's debacle of decades of infrastructure investment that produced negative real growth (due to the debt vs growth generated).
If you're China in 1980 or 1990 (ie third world infrastructure), vast infrastructure spend will faciliate your decades-long growth explosion. Without that, you can't become China of 2019, you can't actually go from a $400b economy to a $14t economy. The ports, roads, rail, bridges, utilities, energy generation, et al. is a requirement and produces extraordinary returns in a context like that. It drastically boosts economic productivity and raises your maximum output potential by a lot.
As another recently discussed (on HN) example, Romania has had one of the world's fastest growing economies in recent years. Economically it's attempting to push into a solid middle-tier economic nation, in the footsteps of countries like Poland, Chile or Slovakia. Romania simultaneously has a horribly lagging roads network that is in desperate need of expansion and improvement. It's very likely that properly building out their roads infrastructure would help facilitate their economic boom continuing; and that not doing so, would act as a serious point of friction on growth long-term. You can't get to $25k GDP per capita, with roads built for a $2k GDP per capita economy.