An AI data centre, read through Chainge

How a rule written for the power sector reaches the cost of running a data centre, traced step by step.

What this is

A method demonstration on a synthetic archetype, a model investor in Dutch AI data centres. Not a real client; built to show how the method reads an exposure, and where its own data would take over.

frame

The direction of travel

Any global event reaches a specific exposure in a few steps. A rule drafted for the power sector, a shift in a commodity market, a regulation written for another industry, each travels across boundaries until it lands on something concrete. Most companies never see the steps, because a sector view cannot look across its own edge. Here is one chain, end to end.

chain

From a power-market rule to the cost of the site

An EU electricity-market reform rewrites how large power users are granted access to the grid.

grounded · source: EUR-Lex

The Netherlands transposes it into a flexible-connection regime. Firm, round-the-clock capacity, the thing a data centre needs most, is no longer a given.

grounded · source: EUR-Lex, NL grid register

To hold its always-on commitment, the site leans harder on on-site backup and the balancing market, both more exposed to short-term grid stress.

grounded · source: ENTSO-E

That runs into the EU energy-efficiency and sustainability-reporting duties the same site already sits under, where more backup and higher draw cut against its reporting position.

grounded · source: EUR-Lex

The combined effect on the cost of keeping this site powered and compliant.

held at lower confidence · the step your own data grounds

What the chain shows Everything above the last step is grounded on public primary sources: the legal texts, the grid data, the reporting rules. That grounded chain is what a profile runs on from the first week, with no bespoke wiring. The final step, the one that lands on the site's own economics, the engine holds back at lower confidence on purpose, because it cannot evidence it from public data. That single held step is exactly what an operator's own data grounds. The engine drew the line, not the marketing.

reach

One shock, two sides

The run surfaced a sharper version of the same idea. A single AI-policy shock, a tightening of controls on advanced compute, reaches this archetype on two sides at once: the cost side, through hardware-supply and exclusion risk, and the demand side, through tenants stretching their decision cycles and revenue timing slipping. One distant event, two exposures, found without being looked for. That two-sided reach is the kind of tie a sector view cannot see.

proof

Why it holds

This is more than a clever prompt. It reasons from grounded primary data, not from headlines. Every claim in the chain carries its provenance, the source it rests on or an honest mark where it is a model estimate. And the predictions are dated and marked against a public track record over time, so the method can be audited rather than trusted.

Throughout, the chain lands on an exposure reaching the entity, the cost of running the site, never a verdict on its returns or its solvency. That boundary is deliberate, and it is what keeps the read about the world acting on a business, not the business itself.

Read the method on the how it works page, and follow the calls on the track record.

The real version is your own company

This is a demonstration on an archetype. Run it on your own exposure, and the held step becomes the one your data grounds.

Put your company on Chainge, and see where the world reaches it.