MIT spends ~1B/y over ~5k researchers (salary+research equipments, etc). That's ~200k/y per researcher, or 0.1% of what top researcher at Meta Superintelligence Lab makes. What causes this massive discrepancy? Many at MSL are recent (2-3y) graduates of MIT/Stanford/etc

one hypothesis is that what MSL pays is closer to their true value to the economy (like making a 1M GPU cluster more efficient). Then top schools should find better ways to enable their values (graduate earlier/give more resources) and capture a part of it (like VCs)