Why the Philippines should build its own AI infrastructure

Originally published in Manila Bulletin

If the Philippines aspires to be a producer rather than a passive consumer in the AI economy, it must treat compute, data, and connectivity as strategic infrastructure. Across Southeast Asia, peers are already moving at infrastructure scale: Singapore’s National AI Strategy 2.0 centers shared national compute; Malaysia has partnered with Nvidia and YTL Power to set up AI supercomputing in Johor; and Indonesia is backing an Indosat–Nvidia AI center in Central Java. These are not pilots but platform bets that anchor research, entrepreneurship, and exportable digital services.

The Philippines has begun to lay pieces of this foundation. The Department of Trade and Industry adopted the National AI Strategy Roadmap 2.0 on July 3, 2024 and launched the Center for AI Research (CAIR), supported by additional budget allocations. In parallel, private players have commissioned new capacity: PLDT’s VITRO Sta. Rosa—an “AI ready” hyperscale facility designed for up to 50 MW—Amazon’s Manila Local Zone to cut latency for cloud workloads, and Equinix’s entry via the acquisition (and in 2025, completion) of three data centers to deepen interconnection. What is missing is not momentum but a coherent plan to stitch these assets into a national AI infrastructure.

The economic case is straightforward. AI will not displace the Philippines’ services engine if the country upgrades the factors of production that make its IT BPM sector globally competitive. That sector closed 2024 with 1.82 million jobs and $38 billion in revenue; its next productivity surge depends on secure access to model training, fine tuning, and inference at scale, close to data and users. Domestic capacity lowers latency and data egress costs and, crucially, enables compliance with the National Privacy Commission’s model contractual clauses for cross border transfers under the Data Privacy Act. Firms that can certify their pipelines as privacy-preserving and locality-aware will capture higher value tasks rather than see them arbitraged to jurisdictions with better provisioned compute.

Sovereignty and resilience add a second, non negotiable rationale. The Philippines repeatedly tops the WorldRiskIndex as the country most exposed to natural hazards, a status that turns AI for early warning, logistics, and relief from a luxury into a necessity. Keeping critical models, datasets, and decision support systems onshore—and caching them at the edge—reduces dependence on international links that are vulnerable to typhoons and seismic events. The state’s National Fiber Backbone, launched in April 2024, supplies the terrestrial spine to move data among government, universities, and carriers; an AI infrastructure strategy should prioritize GPU clusters and secure data lakes on this backbone, with mirrored sites across islands to avoid single points of failure.

Energy economics further strengthen the case. Data centers are power hungry, but they can also be demand anchors for firm, clean generation and grid upgrades. The Department of Energy targets renewables at 35 percent of generation by 2030 and 50 percent by 2040; setting up AI campuses alongside new solar, wind, storage, and transmission can lower long run costs while hardening the grid. With the Philippines bearing Southeast Asia’s second highest electricity tariffs, the only path to cost competitive AI is to fuse compute expansion with the country’s accelerating pipeline of clean energy projects, including the recently announced UAE–Masdar investment program.

Critically, the Philippines should exploit locational advantages that are already materializing. The country sits on an expanding web of subsea cables: PLDT has landed APRICOT branches at Baler and Digos, while the Bifrost trans Pacific system has reached Davao. Proximity to these landing stations, together with Manila’s growing cloud edge and interconnection fabric, can minimize latency for regional users and attract AI tenants whose workloads straddle Asia and the United States. Policy should designate “compute connectivity corridors” around these nodes and align land use, permitting, and spectrum policy to reduce friction for both operators and researchers.

Governance must move in lockstep with investment. The Data Privacy Act and the NPC’s transfer clauses provide the legal scaffolding for sovereign AI “trust zones” in finance, health, public records, and justice. Executive Order 18 on Green Lanes should be used to fast track data center builds and to standardize municipal approvals and rights of way. Public procurement can create anchor demand: a shared “GovCompute” program for agencies and state universities would consolidate spending, enforce cybersecurity baselines, and keep research grade compute accessible beyond a handful of corporations.

The choice, then, is not between buying cloud services abroad or building at home; it is about sequencing both so that domestic capacity becomes the gravitational center for high value work. A credible AI infrastructure strategy—explicit targets for national compute, a siting plan tied to renewables and cable landings, and a regulatory regime that lowers frictions while raising trust—would convert today’s scattered wins into a durable advantage. If the Philippines acts with that clarity, it will graduate from renting other people’s platforms to exporting its own AI enabled services and tools.

This opinion column is published under the Creative Commons Attribution 4.0 International License (CC BY 4.0). You are free to share, adapt, and redistribute this content, provided appropriate credit is given to the author and original source.