The deeper point is that efficiency rarely reduces total demand when the input becomes cheap enough. For AI, the constraint is not chips or models but reliable, affordable electricity at data-center scale. A system that uses fewer watts per task can still trigger a surge in overall power use if lower costs make more training runs, cloud services, and consumer applications economically viable.
For Philippine firms, this is not abstract. Companies are already weighing AI tools for customer service, logistics, payroll, and demand forecasting. If those tools become standard, office buildings, warehouses, and digital platforms may face higher electricity bills even as each task becomes more efficient. The local challenge is that power supply remains sensitive to fuel prices, typhoons, grid bottlenecks, and transmission constraints. A jump in data-center or commercial load could pressure tariffs unless generation, interconnection, and distribution keep pace.
Policymakers should watch demand-side signals, not just carbon goals. The issue is whether the grid can absorb a faster rise in commercial and digital load without pushing up costs for households and manufacturers. This may make energy storage, distributed generation, and efficient data-center siting more relevant than before.
Watch for rising electricity tariffs tied to large commercial customers, new rules on data-center interconnection, and whether AI adoption begins showing up in utility demand forecasts. If global players are planning very large power projects while local capacity expands slowly, the gap could widen into a cost disadvantage for Philippine firms trying to compete digitally. The risk is that digitalization gains get partly offset by energy costs, especially for firms that cannot pass them through to customers.