How do you do expensive bleeding edge research with no money? Sure you might get some grants in the millions but what if it takes billions. Now lets assume the research is no small feat, its not just a handful of individuals in a lab, we need to hire larger teams to make it happen. We have to pay for those individuals and their benefits.
My take is its not cheap to do what they are doing and adding a capped for-profit side is an interesting take. Afterall, OpenAI's mission clearly states that AGI is happening and if thats true, those profit caps are probably trivial to meet.