From the article this seems much less bad than the headline might imply. This idea predates trump and it does not eliminate scientific review of proposals. That said, whether the cost savings itself is worth what might be a diminishment in review quality is hard to say. I can only comment from the NSF side of things to say that peer review of proposals is a mixed bag and will unavoidably run into human error and individual predispositions regarding scientific importance so maybe this isn't a bad approach to try.