The Hidden Costs of the Global South’s AI Data Center Boom

The arithmetic of the modern AI economy is skewed. In Malaysia, data center projects approved between 2021 and 2025 required roughly RM101 million in capital for every projected job, compared to just RM1.37 million per job in the manufacturing sector. This capital-intensive model places immense strain on local resources—land, water, and power—without providing the broad-based industrialization or skill transfers that defined previous development eras.

Investors often bypass this friction by relying on political intermediaries to fast-track permits. However, this approach is fragile. In Chile, Google faced significant setbacks when an environmental court forced a redesign of a Santiago data center due to aquifer concerns. Such incidents demonstrate that political access is not a substitute for a social license. When infrastructure projects ignore local environmental or economic needs, they trigger community resistance, leading to delays and increased long-term operational risk.

To avoid becoming mere resource hosts, governments must move beyond simple investment announcements. Chile’s National Data Centers Plan serves as a potential blueprint, integrating investment facilitation with strict environmental and capability-building criteria. By linking public resource allocation—such as power and water access—to measurable outcomes like local procurement, compute access for universities, and workforce development, states can ensure that AI infrastructure delivers tangible, long-term value rather than just serving as a digital theater for short-term political gain.