The article doesn't mention one very important expense for nearly every research field: getting access to the research other people have published. Some fields (e.g. computational linguistics) have long had a policy of letting authors keep copyright, so they can post their own papers online, and the collections are also free; but the norm is that nearly all publication goes through publishers that paywall. So just to see what everybody else has been doing, you need to have institutional support.

Only slightly less important is access to conferences, which are not always super cheap: travel + lodging + registration is often in the $1000 range even for the relatively accessible conferences.