i attended a talk on energy use at industrial chemical plants a few months ago - one of the main points emphasised by the academic giving the talk was that industry refuses to invest in deploying any new technique/approach coming out of research until it is demonstrated that the new approach definitely works at scale. maybe part of "knowing" that it works is having something operational at large-scale and running for long enough to help identify and iron out problems that weren't anticipated during design and smaller scale trials. if it costs hundreds of millions to billions to build the thing you want to guarantee that the approach is solid.
I think that is exactly correct, it is also why Facebook and Google have built more data centers in the last 10 years than any of the "established" internet service providers. They operate on a different evaluation strategy, and establishing a source of energy that they controlled and was 'green and unencumbered' has been high on Google's list for a while. They helped build a very large solar farm in the Mohave desert for just that reason.
And they are sitting on billions of dollars of cash that is returning maybe 1.2% in returns.
