AI Infrastructure Bullwhip: Bottlenecks Cascade Costs

The bullwhip framing matters because it reframes AI infrastructure strain as a structural supply-chain problem rather than a series of isolated shortages. Tunguz's point is that a constraint at any single layer doesn't stay contained — it propagates and amplifies costs across the stack, which means capacity planning and procurement assumptions built on smooth scaling are likely wrong. The open question is whether the industry treats these cascading bottlenecks as a coordination problem to engineer around, or simply absorbs them as a persistent cost of doing business.