They only care because open-weight helps to drive the GPU business notably thanks to inference providers (Baseten, Together, Mistral, ...).

At least their support helps the open-weight ecosystem.