Deep Dives
Moats, tokens-per-watt, and capex—read through Nvidia and the AI build-out. · Started Apr 2026 · 4 of 7 sections published
Moats, tokens-per-watt, and capex—read through Nvidia and the AI build-out.
Tokens per watt, bottleneck inversion, and the Hopper–Blackwell arithmetic.
Tacit knowledge, CUDA as institutional memory, and who captures AI value when models commoditise.
What the inference market and Nvidia's no-surge-pricing tell us about strategic decisions in multi-dimensional spaces.
Grids, GPUs, and the physics of running intelligence at planetary scale.
Watts, wires, and who controls the stack—not the headline model. · Started Mar 2026 · 5 of 5 sections published
Watts, wires, and who controls the stack—not the headline model.
Smil on why transitions take generations, welfare at ~110 GJ, and why energy enables but never decides.
Land and power density as the real limit on decarbonisation—and why hyperscalers are buying nuclear.
Europe's climate funding gap vs AI capex—same capital pool, different clocks.
Yergin's cartographic method—shale, Malacca, path dependence—and why the AI race is an energy and supply-chain story.
From trust loops to deployment metabolism: how products actually reach the market.
How frontier tech earns trust in the field, and a simple map of the constraints that set deployment speed. · Started Oct 2025 · 3 of 3 sections published
How frontier tech earns trust in the field, and a simple map of the constraints that set deployment speed.
You need trust to deploy, but you need deployment to earn trust. How to resolve the circular dependency.
A framework for diagnosing your five deployment forces and understanding how constraints determine which strategic approaches can work.
