Unplugging the silica dream

The pitch for the Pax Silica hub in New Clark City is a staggering pile of money. The Bases Conversion and Development Authority is dangling between USD 40 billion and USD 70 billion in prospective investment. They promise 130,000 jobs. It is an intoxicating vision of semiconductor fabs and artificial-intelligence data centers that would drag the country into an advanced, secure global supply chain.

But visions do not power server racks, and here is the math. BCDA officials are talking about individual hyperscale data centers requiring 300 to 600 megawatts of electricity. A single facility at the top end of that range devours a massive chunk of the Luzon grid’s capacity. AI is thirsty, too. The hub needs 65,000,000 to 90,000,000 liters of water daily. BCDA says a planned surface-water system will harvest 120,000,000 liters a day.

They can probably engineer it, but that is not the point.

The point is opportunity cost. Building data centers is fast. Building power plants and transmission lines takes years. The International Energy Agency expects global data-center power consumption to hit 945 terawatt-hours by 2030 — more than double its 2024 footprint.

When supply lags behind this artificial demand, power gets expensive. And we know exactly who pays for a stressed grid. The residential ratepayer. Every household in Metro Manila and Central Luzon already choking on utility bills will subsidize the transmission upgrades required to keep foreign servers cool.

New York State figured this out. In July 2026, they slammed a temporary moratorium on new hyperscale permits. They are far from anti-tech. They just stopped to write rules that force operators to pay for grid expansion so the costs do not shift to ordinary citizens. If one of the richest states in America feels the need to protect its residents from an AI infrastructure land grab, the government here shouldn’t rush to hand over the keys.

Yes, we need foreign direct investment. Turning away billions in capital is a bitter pill when domestic industry struggles to scale. The jobs matter.

But a highly productive enclave is not a national economic strategy. Economist Daron Acemoglu projects relatively modest productivity gains from AI over the next decade. The hype might outpace the reality. If we lock up our scarcest resources — land, water, reliable power — for facilities that generate massive capital returns for American tech giants but only shallow land rent and limited tax revenue for us, we lose.

The fix is simple, though it will anger the developers. Charge these tech consortiums the true price of admission. Mandate that Pax Silica investors fund their own dedicated renewable generation and water infrastructure upfront, before ground is broken. Force them to partner with local suppliers to keep the value domestic.

If the economics of their AI models fall apart without heavily subsidized Philippine power, then the investment was not sound to begin with. Let them prove the math works. Until they do, the turbines stay off.