"Sieć energetyczna USA nie jest w stanie dostarczyć takiej mocy w terminie, jakiego potrzebują firmy AI. To po prostu prosty fakt. Uzyskanie pozwolenia na nowe przyłącza do sieci zajmuje lata. Laboratoria AI nie mają lat." "The US power grid cannot deliver that kind of capacity on the timeline that AI companies need. That's just a simple fact of reality. Getting permitted for new grid connections it takes years. AI labs don't have years."

… Thanks, guys, for Bit Unix. $80 a month Bit Unix rebate just for being a member. 7-day free trial. And the link's right below. Stop watching from the sidelines and catch the next move with us. Link down below. Now, here's the first piece of plumbing in the thesis, okay? And it's genuinely elegant. AI compute clusters need gigawatts of power. The US power grid cannot deliver that kind of capacity on the timeline that AI companies need. That's just a simple fact of reality. Getting permitted for new grid connections it takes years. AI labs don't have years. So, Ash and Brenner bypass the grid entirely and his fund went heavily on Bloom Energy, a company that's building modular solid oxide fuel cells that let data centers generate their own power on site, completely independent of the utility companies. Now, of course, he bought in early, he bought a huge stake in the company well before Bloom Energy …