Twelve mapped first.

The compute floor itself: siting, power, water, and what a facility looks like when the cooling plant is no longer part of it.

The industry whose entire margin is electricity per unit of work, and therefore the first place a change in that ratio shows up on a balance sheet.

Inference is a latency business. What changes when the compute can sit near the user instead of in a remote campus built around a substation.

Where the toll on a transaction actually goes, and what a settlement layer costs when conversion stops being the expensive step.

An industry that already pays enormous sums for microseconds, meeting a logic element measured in femtoseconds.

Hardware-anchored attestation, and what becomes possible when trust is rooted in a physical device rather than asserted in software.

Who carries the light, and where it lives. The transport layer between the switch and the site.

Machines whose capability is bounded by how much compute they can carry and how much heat they can shed while carrying it.

Platforms where power, thermal budget, and mass are the binding constraints on what can be computed at the edge. Market map only.

Optical sensing and the diagnostic instruments that follow when the sensing element and the processing element share a substrate.

The generation side of the same physics, and what a compute campus looks like when it makes a meaningful share of what it draws.

The consumer end of the market, where latency is felt directly and rendering budgets are set by thermal limits.