SIGNAL

Perspectives on the infrastructure behind AI

The public argument about AI infrastructure is being conducted almost entirely in the wrong units.

It is framed as a question about artificial intelligence, and it is really a question about electricity, land, water, capital, and consent. It is fought at city councils, which hold the wrong instruments. It is defended with engineering specifications, which answer questions nobody asked. And it is reported as a story about technology when it is a story about who pays, who decides, and who benefits.

Signal exists to argue about that properly.

These are commentaries. Each one takes a specific piece of reporting and works out what it actually shows, which is often not what the headline concluded. We write them because the quality of this debate will determine what gets built in Canada over the next decade, and the debate is currently poorer than the decision deserves.

Five things we think, stated plainly so readers know where we begin.

Placement is a choice, not a constraint. The strain on urban grids is not created by compute in the abstract. It is created by putting enormous continuous loads where the grid is already tightest. Most of that is a choice.

Disclosure beats prohibition. A project that can state its load, its water, its emissions, its grid behaviour, and its local benefit before anyone demands them should be permitted. A project that will not say should not. Moratoriums exist because that standard does not, and the industry's reluctance is the reason.

Engineering quality is not a consent strategy. A well-designed facility still owes its neighbours an answer about what they receive. The sector has been substituting the first for the second and has been surprised by the result.

A community is owed something specific. Not construction jobs, not a sustainability claim, not a national interest argument. Something a resident can verify without taking anyone's word for it. Very few projects can answer this, and the ones that cannot are the ones being stopped.

Sovereignty is a reason for a country to build, not a reason for a community to accept. Those are different arguments for different audiences, and conflating them has cost this industry more than it realises.

We will get things wrong, and we will say so when we do. We will also credit good work by people we compete with, because the alternative is not commentary, it is marketing.

Published irregularly, whenever something is worth arguing with.

VOLTEDGE