As someone who writes high throughput PHP code I can confirm that it’s much more about technique than language capabilities (though in an embedded setting things with dynamic GCs are simply unusable unless static memory management can be enabled with a compiler switch).
For most projects you’d be much more rewarded for focusing on tools/framework/libraries available for the different languages (since that’s where most initial effort will go) and then build up any missing functionality as needed ontop of that base.
Most languages can do pretty much anything these days. The technical advantages are much smaller than the impact the right approach will have… it’s one reason that I hold “maintainability” as the most important attribute of a project.
NIH (Not Invented Here) is absolutely the downfall of many tech companies. Code costs constant money to maintain - it may sound illogical since it’s unchanged from the day it was written but it absolutely is the case.