Great article, I have been wondering for a while why prices of energy commodities have not reacted as strongly as it was initially expected, these tend to be some of the most efficient markets in the world. Are we missing something here, specially gas and coal price movements are really tiny. My theory is that oil smuggling high initial inventories and floating inventory have calmed the market, but still is a big doubt that I have. What could we be missing?
The AI buildout being bottlenecked by transformers and switchgear is the story most AI optimists don't want to hear. Compute scales fast, the physical layer doesn't. Anyone underwriting AI capex right now without a hard read on Chinese electrical components is modeling fantasy numbers. Are you seeing any operators pricing this constraint into their build timelines yet, or is everyone still planning around 2024 lead times?
The Naval quote nails it. By the time something is investable to the public, most of the alpha is priced in and the institutions are stepping back for a reason. The smarter play with AI right now isn't buying the IPOs anyway, it's using the tools to take cost out of your own operation. I run a small services business and what AI has done for our admin and dispatch already pays for itself many times over before any IPO ever prints. Are you seeing operators in commodities and EM lean into AI on the ops side, or is it still mostly a public-market trade for them?
Great article, I have been wondering for a while why prices of energy commodities have not reacted as strongly as it was initially expected, these tend to be some of the most efficient markets in the world. Are we missing something here, specially gas and coal price movements are really tiny. My theory is that oil smuggling high initial inventories and floating inventory have calmed the market, but still is a big doubt that I have. What could we be missing?
The AI buildout being bottlenecked by transformers and switchgear is the story most AI optimists don't want to hear. Compute scales fast, the physical layer doesn't. Anyone underwriting AI capex right now without a hard read on Chinese electrical components is modeling fantasy numbers. Are you seeing any operators pricing this constraint into their build timelines yet, or is everyone still planning around 2024 lead times?
🙏
Electrical component manufacturing, critical mineral production, buybacks -- you got me turning into a China bull!
The Naval quote nails it. By the time something is investable to the public, most of the alpha is priced in and the institutions are stepping back for a reason. The smarter play with AI right now isn't buying the IPOs anyway, it's using the tools to take cost out of your own operation. I run a small services business and what AI has done for our admin and dispatch already pays for itself many times over before any IPO ever prints. Are you seeing operators in commodities and EM lean into AI on the ops side, or is it still mostly a public-market trade for them?