However, companies scraping high volumes of freely available Wikipedia knowledge for AI training has driven up server demand and, subsequently, costs at the non-profit, whose primary source of income is small donations from the public.

It’s framed around cost.