Research suggests AI infrastructure projects are worth the investment less than 30% of the time

Who could have foreseen this?