Quick Read <\/p>\n
Anthropic\u2019s AI demand dramatically exceeded internal projections, exposing a critical infrastructure bottleneck where megawatts have become the limiting factor for serving frontier AI users.<\/p>\n
Hut 8 (HUT) secured a $7.0 billion 15-year lease with Fluidstack for 245 MW of AI data center capacity at its River Bend Louisiana campus. The stock is up 114% year-to-date and 595% over the trailing year with an $11.09 billion market cap.<\/p>\n
The analyst who called NVIDIA in 2010 just named his top 10 stocks and Hut 8 wasn’t one of them. Get them here FREE<\/a>.<\/p>\n On a recent episode of The Pomp Podcast, investor Jordi Visser relayed an admission from Anthropic CEO Dario Amodei that reframes how to think about AI infrastructure demand. Anthropic, the maker of the Claude assistant, watched AI demand blow past every internal scenario they had modeled: “We planned our business for 2 or 3x [growth], and then we went all the way out and said the outlier would be if we could 10x our business. And he goes, the problem is that we didn’t 10x, we 80x’d the business.”<\/p>\n The Physical Infrastructure Constraint <\/p>\n Visser\u2019s takeaway cuts directly into the core challenge facing investors in AI right now. If the companies building frontier AI systems underestimated demand by that magnitude, then the constraint becomes physical infrastructure. More users require more compute, more data centers, and ultimately more power capacity.<\/p>\n The analyst who called NVIDIA in 2010 just named his top 10 stocks and Hut 8 wasn’t one of them. Get them here FREE<\/a>.<\/p>\n In AI, megawatts are becoming the equivalent of restaurant seats. Host Anthony Pompliano offered a restaurant analogy: “If you create a restaurant and all of a sudden you’ve got a line out the door, well, you need more seats.” The seats, in this case, are megawatts.<\/p>\n Where the Capital Is Flowing: Hut 8 <\/p>\n