What actually stops an AI data center from earning what it was built to earn.
A deterministic instrument that computes what physically and economically constrains an AI compute investment, and shows why that constraint binds — every number carrying its source and grade, every refusal shown as a finding.
Illustrative reference cases — precomputed from the physics model; open one to adjust its assumptions.
Nordic 40 MW GB200 — timeline binds before hardware
Everything this site needs can be had — just not yet. Grid connection, permits, construction and deliveries stack into one long schedule, and no revenue flows until the last step clears. Waiting itself is the cost: every month of schedule is a month of financing paid with nothing earned.
40 MW · SE-SE1 · a margin of 1.7 % over hardware
Adjust the assumptions in the constraints builder →Helsinki Waste-Heat Campus 80 MW — capital binds before timeline
The physics of this site clears; the financing does not. At the return its backers require, the income the site can earn does not cover what it costs to build and borrow for — so money, not power, heat or chips, is the boundary here.
80 MW · FI-01 · a margin of 35 % over timeline
Adjust the assumptions in the constraints builder →Norway Narvik 230 MW — hardware binds before timeline
The scarce thing here is the chips themselves: they are sold out far ahead, and new orders join a queue measured in months. The building and the power are ready before the silicon arrives, so the wait for hardware sets the pace of everything else.
230 MW · NO-1 · a margin of 18.3 % over timeline
Adjust the assumptions in the constraints builder →Labor × Chips Collision — network binds before capital
A token only earns money once it reaches its buyer, and this site's connection to the outside world is narrower than what its machines can produce. Part of the output could never reach a customer, so the data links, not the computers, cap the revenue.
60 MW · US-AZ · a margin of 15 % over capital
Adjust the assumptions in the constraints builder →