DICT’s $34.4 Billion AI Infrastructure Plan Frames a Build-or-Buy Choice for the Philippines
The Department of Information and Communications Technology’s Philippine Artificial Intelligence Infrastructure Masterplan sketches about $34.4 billion of investment to grow domestic AI compute from roughly 50 megawatts toward 1.5 gigawatts by 2033, a build-versus-buy choice explored in a BusinessMirror deep dive on October 8 and earlier BusinessWorld coverage of the masterplan. Corridors around Clark-Bataan and Batangas-Aurora are flagged for power and fibre, while two U.S. hyperscalers are described as eyeing sites near 200 megawatts each.
DICT officials, including references to Secretary Aguda’s framing, cast the plan as a way to stop treating the Philippines only as a consumer of foreign models and cloud regions. Local “sachet AI” offerings priced around ₱49 are meant to widen access for SMEs and students even as large campuses chase exportable inference capacity. The strategic tension is familiar: import GPUs and managed clouds quickly, or spend years assembling power, land and racks under Philippine control.
Power availability, transmission upgrades and water for cooling will decide timelines more than slide decks. Communities near proposed corridors will ask who pays for grid reinforcement; investors will ask whether offtake from government and BPO buyers is bankable enough for 15-year project finance.
What it means in practice
A credible PAIIM changes procurement maths for firms already lifting adoption, as in the AWS finding that Philippine AI adoption has reached 32 percent while skills gaps persist. Productivity studies such as the World Bank view that wider AI adoption and higher-value BPO could lift productivity assume compute is reachable and affordable. Industrial parks like DTI’s Bulacan ecozone push to assemble and export AI server racks are the hardware counterpart to PAIIM’s megawatt targets.
If the Philippines only buys foreign capacity, peso outflows and latency for local language models will stay structural. If it builds poorly sited campuses, stranded assets will haunt the budget. The middle path is phased corridors with hyperscaler anchors and open access for domestic startups.
What to watch: signed hyperscaler MoUs with megawatt figures, ERC and grid operator timelines for Clark-Bataan and Batangas-Aurora, and whether sachet AI pricing survives once subsidised pilots end.