Here’s another inconvenient truth: the more we use artificial intelligence (AI), the more power and water we’re going to need. No, I am not against building more data centers and servers for AI—particularly in the Philippines. Far from it.

I’m calling on our government to plan ahead, to anticipate the increasing demand so that our economy can benefit from it. But we must also ensure that our power and water needs are not depleted in the process.

Consider this: many Filipino data-centers will draw from a grid largely powered by imported fossil fuels. The more power we need, the more fossil fuel we’ll import. And if water demand rises—say for cooling or supporting server-farms—then we might be forced into costly solutions like desalination: turning salt-water into fresh water, which itself requires a lot of power. That’s electricity to generate water to cool electricity-generating infrastructure — a vicious circle.

But the other option is not to build our own data centers, so that our demand stays down. Yet if we do that we risk being left behind in the AI services market. So what’s the realistic path forward? For me, the answer lies in renewables. Let’s produce more power using solar, wind, geothermal, and hydroelectric technologies. In that way we can chase a larger share of the AI services market and guard our energy security and water security. It’d be ironic if the rich benefited from AI while the poor had no water to drink.

The numbers don’t lie

Here are some figures that underscore how serious this is:

In the U.S., data centers consumed about 183 terawatt-hours of electricity in 2024, more than 4% of total U.S. electricity consumption.

Globally, some estimates say data centers today consume about 1-2% of global electricity.

A study suggests electricity demand from AI workloads could approach 49% of all data center electricity by the end of 2025.

On water, U.S. data centers consumed about 17 billion gallons of water in 2023, largely for cooling. 

One detailed estimate suggests that by 2027, AI demand could lead to water withdrawals of 4.2-6.6 billion cubic meters—equivalent to the annual withdrawal of half the U.K.

These are not trivial numbers. That means in the Philippines we need serious infrastructure planning—not just to support AI growth, but to make sure we remain solvent in our water and energy systems.

A troubling case in point

We don’t have to imagine problems—they’ve already happened. In Newton County, Georgia, U.S., a data center built by Meta Platforms, Inc. (then “Facebook”) has been linked to water-supply issues for nearby residents. Houses relying on wells reported their taps running dry within months of the center’s construction in 2018. 

One family says they’ve spent over US$5,000 trying to fix the problem and still can’t afford replacing their well (cost estimated at US$25,000). This isn’t just a local nuisance—it’s a stark reminder that huge AI infrastructure can impose hidden social and environmental costs.

What I’m urging the Philippines to do

Here are my suggestions:

Forecast infrastructure demand – The government (and private sector) must project how many AI-center MWs we might add in the next 5–10 years, and map out how much extra electricity and water that will require.

Prioritise renewables and local water resources – If we build new power mainly via imported fossil fuel, we risk dependence and pollution. Instead, let’s scale up solar, wind, geothermal (we have it!) and hydro capacity, and ensure new data centres are sited where water stress is minimal or mitigatable.

Regulate responsibly – Just because we can build data centers doesn’t mean we should if the local power grid or water supply cannot handle them. Local governments and national agencies must vet proposed sites for AI infrastructure under resource-stress criteria.

Adopt efficiency and alternative cooling – Data centers don’t have to use archaic cooling that burns up water. Technologies like liquid cooling, air cooling in colder climates, or reuse of waste heat should become standard.

Ensure equitable benefits – As we chase the AI market, let’s be sure the benefits aren’t confined to large tech players. Rural communities, poor households, indigenous communities must not end up sacrificing water or power so that others get the gains.

My question to you — and to our policymakers

Are we ready to chase the AI race and sustain our energy-water-eco systems? Or will we simply build fast, import more fuel, sit in water stress and call it progress?

 In the Philippines we have an opportunity: with our sunshine, our geothermal potential, our hydropower sites, we may leapfrog older power systems and actually build AI infrastructure in a sustainable way. But only if we choose that path.

Because if we don’t: the images from Georgia might look familiar—homes with dry wells, a community forced into water rationing, infrastructure built without local oversight. That cannot be our story.

In the end: Building more AI servers is not the problem. The problem is how we build, where we build, and with what resources. If we get that right, we can have our AI cake and eat it too—with power and water to spare, and people across the Philippines included in the benefits.

Ramon Ike V. Seneres, http://www.facebook.com/ike.seneres

iseneres@yahoo.com, senseneres.blogspot.com 09088877282/03-27-2026