The Decomposition Discipline

We can now assemble the pieces into a working method. The claim of this course is that any significant cross-border expansion decision is a blend of three motives — infrastructure, tax, and data residency — and that most public explanations foreground one while staying quiet about the others. The practitioner's skill is to take the decision apart and estimate how much weight each motive actually carries. This is not cynicism; it is bookkeeping. Companies are entitled to their stated reasons. You are entitled to check whether the stated reason can bear the full weight of the decision, and usually it cannot.

Begin with infrastructure, because after the global floor it is the motive with the strongest claim on the physical map. A data centre is, before it is anything else, a very large electrical load with a cooling problem attached. Its economics are dominated by the price and reliability of power, the availability of cooling — cold air, cold water, or the energy to chill warm air — the presence of fibre to move data in and out, and the ability to secure land, permits, and a grid connection in a reasonable time. These are stubbornly physical facts, tied to geography and to a particular grid's condition. They do not move because a rate changed. When you see a large build, the infrastructure hypotheses should be your first, not your last: is there firm cheap power here, is there a grid connection available when others wait years in the queue, is the climate forgiving to cooling? Frequently the whole decision is explained here, and the tax story is a rounding error.

Then tax — but tax after the floor, which is a diminished thing compared to a decade ago. The minimum rate has compressed the extremes, so the question is no longer "which jurisdiction lets us book profit at near zero" but the subtler "where can we operate at or modestly above the floor, and where do the carve-outs and credits actually land." Remember the substance-based carve-out: the minimum tax reduces the top-up in proportion to real payroll and tangible assets in a country. That detail quietly rewards putting genuine physical operations — including data centres and the people who run them — in a jurisdiction, because tangible assets on the ground shrink the top-up. So tax has not vanished; it has partly fused with the infrastructure question. Building real things in a place can now improve the tax position directly, which is a different logic from the old paper-shifting and one you should watch for.

Third, data residency, the motive most likely to be underweighted by an outside observer because it is legal rather than economic and often unglamorous to state. A law requiring that citizens' data, or a regulated category of it, be stored or processed inside the national border is a hard constraint on the physical map. It cannot be satisfied by a contract or a holding company; it demands iron in the country. Financial data, health records, government data, and personal data under strict privacy regimes are the usual triggers. When a company builds in a country whose market it could plausibly serve from next door, a residency rule is often the unstated reason it did not. Always ask what the local law says about where the relevant data must live, because that single legal fact can override every economic argument and force a build that infrastructure and tax alone would not justify.

The method, then, is a disciplined three-way split. For any expansion, ask: how much of this is the grid and the land? How much is the effective rate and the carve-outs, understood post-floor? How much is a data-residency mandate compelling a physical presence? Assign rough weights and notice which motive is load-bearing — the one that, if removed, would change the decision. The stated reason and the load-bearing reason are often different, and the gap between them is where the real analysis lives. A company citing "proximity to customers" for a build that a residency law required has told you something true and something incomplete in the same breath.

One discipline to guard against: do not let the most visible motive crowd out the others. Power deals make good press and are freely discussed. Tax is discussed carefully, through advisers, in the language of effective rates. Data residency is often barely mentioned because it is a compliance obligation, not a boast. The quietest motive is frequently the most binding. Weighting the three honestly means resisting the pull of whichever one the company most wants to talk about, and asking, coldly, which fact would have to change for the decision to flip. That fact is your answer.