Yeah. It's a chicken and egg thing. Imagine if resources applied to PyTorch was spent implementing in Julia.

But then there's not enough users... so the cycle continues until one day Julia hits critical mass and a tipping point is reached.