But what we're slowly learning is that a huge percentage, perhaps most of that "millions upon millions of dollars and man hours" ends up producing terrible papers that don't replicate (in science) or don't even say anything coherent at all (e.g. basically all of gender/critical race theory).
COVID-19 has been an excellent demonstrator of this: at a time when it's become clear that epidemiologists and virologists don't actually seem to know much about viruses, and know even less about stopping them, we have widespread riots and protests against 'systemic racism' that can't be found in any data sources. All this badness is coming from universities which look bloated, over-funded and nowhere near scrutinised enough.
Then finally you look at the tech sector, you look at commercial organisations at the cutting edge. The sad reality is they don't use much university output. They don't. Look at CS, most of the breakthrough papers are coming from corporate labs. The last 10 years saw multiple new programming languages take off and gain critical mass, none of them bear any resemblance to Haskell or other research languages popular in academia. A lot of the improvement in tech, whether it's electric cars, self driving cars etc, it's been coming from corporate labs.