Posts

Featured Post

The Math Says Jack Clark Might Be Right

Image
  Copyright: Sanjay Basu A Technocrat’s Discernment Jack Clark  said something in September 2025 that most people either celebrated or dismissed, but very few actually sat down and did the math on. He said he continues to believe that the sort of powerful AI system described in  Dario Amodei’s  “Machines of Loving Grace” essay will be buildable by the end of 2026, with many copies running in 2027. He doubled down on this at a Congressional hearing, too. End of 2026. Truly transformative technology. I have spent years building GPU and generative AI infrastructure at Oracle Cloud Infrastructure. I have watched training clusters scale from curiosity projects to multi-billion dollar deployments. I have seen what happens when you give smart people enough FLOPS and enough will. And honestly, when I first read Clark’s prediction, my instinct was skepticism. Not because the goal seemed impossible in some abstract sense, but because the gap between “buildable” and “deployed a...
Post a Comment
Read more

The Illusion of Mining Bitcoin on GPUs

Image
Copyright: Sanjay Basu There is a certain kind of optimism that never quite dies in our industry. It shows up every few years. A new wave of hardware arrives. Someone looks at a powerful GPU cluster and asks a simple question. Can this print Bitcoin? It is a fair question. It is also the wrong one. Because what you are really asking is not whether it works. It does. In the same way you can cross an ocean on a rowboat. The question is whether it makes any sense at all. The answer, if we stay honest, is no. The Scale Problem Nobody Escapes Let’s start with first principles. Bitcoin mining is not about compute in the abstract. It is about compute shaped in a very specific way. SHA 256 hashing. Deterministic. Brutal. Repetitive. No room for cleverness. This is why ASICs exist. A modern Bitcoin ASIC pushes well above 100 terahashes per second. That is not a typo. That is the baseline. Entire racks scale into exahashes. Now place a GPU next to it. Even a powerful accelerator like an A100 is ...
Post a Comment
Read more