The structural problem in the AI-environment debate is that the same infrastructure decision — building a new data center — produces two completely incompatible narratives depending on which number you choose to foreground. Labs emphasize AI's potential to optimize energy grids, improve climate modeling, and accelerate carbon capture research. Critics point to the thin evidence behind hyperscaler climate claims and the immediate, measurable cost of coal and gas generation brought back online to meet data center load. Both can cite legitimate studies. Neither is accounting for the full ledger, and…