On Jan. 17, 1893, the Hawaiian Kingdom was overthrown by a group made up largely of sugar planters.

The legacy of plantation agriculture is still visible across Hawaiʻi in diverted water, depleted soils, food import dependency and an economy structured around extraction. A place once rich in water, life and culture, that sustained a million people, had its waiwai redirected into producing sugar and pineapple that enriched the “Big Five.”

Plantations reorganized the entire ecosystem of Hawaiʻi to serve distant markets, concentrating land and water under a few owners while exporting wealth off-island. The result was ecological simplification, cultural disruption and long-term dependency. Today, the commodity is no longer sugar or pineapple or sandalwood. It is machine intelligence — computer, data and the infrastructure that powers them.

We are a collective of university professors in fields of AI and genome sciences, nonprofit leaders, and community organizers, asking a simple but urgent question: As AI data centers are proposed to come to Hawaiʻi, who will own the data, control the resources that power them and hold these systems accountable?

A Special Commentary Project

Civil Beat is focusing on transparency, accountability and ethics in government and other institutions. Help us by sending ideas and anecdotes to sunshine@civilbeat.org.

Read More

From places like Singapore, Malaysia, and across the continental United States, we have already seen the consequences for local communities — and we refuse to see Hawaiʻi follow the same path. We want AI to support our livelihoods, prosperity and culture — but only if island communities have a real say in how it is built and governed.

In rural places around the U.S. and Malaysia, as well as abutted to urban centers in Singapore, data centers now operate at a scale that rivals — and in some cases exceeds — the extractive intensity of 19th century plantations.

Plantation-era sugar production in Hawaiʻi required vast amounts of water — ranging from tens to hundreds of gallons per pound of sugar (depending on how diversion and losses are counted) supported by ditch systems that rerouted tens to hundreds of millions of gallons per day, ultimately amounting to hundreds of billions to over a trillion gallons annually, or the equivalent of hundreds of reservoir-scale water systems like Wahiawā each year. A single modern hyperscale data center can consume 1 million to 5 million gallons of water per day, over a billion gallons a year.

For context, the Wahiawā reservoir system on Oʻahu — built to support plantation agriculture and now associated with Dole operations — stores on the order of billions of gallons of water (approximately 2 billion to 3 billion gallons), meaning one large facility could approach that entire volume within a year. With Wahiawā serving a population of roughly 18,000 to 19,000 residents, a single data center’s daily demand can easily rival or exceed the municipal-scale water needs of an entire community.

Beyond water, the environmental and infrastructure footprint is equally stark: data centers already account for 1% to 2% of global electricity use, with large facilities drawing power comparable to over 100,000 homes, while lifecycle impacts — from concrete-intensive construction to server manufacturing — add substantial embodied carbon.

Data centers of this size are used for training AI models. A single large AI model can emit hundreds of metric tons of CO2 and consume millions of liters of water, and grid-dependent operations — especially those relying on fossil fuels and diesel backup generators — contribute to particulate pollution and public health burdens in surrounding communities.

In Hawaiʻi, these impacts would likely be amplified: higher ambient temperatures increase cooling demand, imported fossil fuels raise emissions intensity, freshwater is already constrained, and geographic isolation magnifies the environmental and economic cost of every gallon of water, ton of material, and kilowatt-hour consumed.

Hawaiʻi has higher land and energy costs than the rural regions of the United States where hyperscale data centers are often sited. But cost alone does not prevent this intensive, extractive infrastructure from arriving. Singapore — a tropical island nation with deep agricultural history, dense subsea cable connectivity, and proximity to major financial markets — illustrates this clearly. By the early 2020s, Singapore had built out roughly 876 MW of data center capacity, placing it among the largest global hubs.

Data centers already consume around 7% of the country’s electricity, despite its hot, humid climate where cooling demands are especially high. When land constraints tightened, development didn’t stop — it spilled into nearby regions like Malaysia, where large new hyperscale campuses are now being built to serve the same regional demand. With data centers, it is always location, location, location — cost be damned.

We need to make it clear that extractive AI is not acceptable here. Hawaiʻi’s location in the middle of the Pacific makes us a routing point for data traveling through undersea cables. Smaller data center buildouts that are happening in Hawaiʻi, like the acquisition and expansion of AlohaNAP and DRFortress, aren’t used for the resource-intensive training of AI models, but they do support the flow of data for AI queries between Asia, Pacific Islands and the U.S., which could grow over time.

These data centers generate value for their out-of-state owners while relying on Hawaiʻi’s natural resources. Military data centers located in Hawaiʻi, about which less is publicly known, do the same.
The costs of large-scale data centers are staggering, but there is another path.

An Ahupuaʻa-Inspired Model

“I ka wā ma mua, i ka wā ma hope” — we must look to the past as a guide to the future.

Long before plantations, Hawaiʻi sustained a large population through a sophisticated system of land and resource management known as the ahupuaʻa: a division of land from mauka to makai that integrated forests, agriculture, freshwater, and fisheries into a self-sustaining whole. Resources were managed locally, with deep knowledge of place and an ethic of stewardship that maintained abundance over generations.

What would it mean to design our islands’ computing infrastructure with those principles in mind? What would it mean if we, Kānaka Maoli and Kamaʻāina, were the leaders in ensuring the AI data centers that inevitably do arrive on our lands were built differently?

An ahupuaʻa-inspired model of computing would look very different from today’s data centers. It would prioritize local ownership and data sovereignty, ensuring that communities — not distant corporations — control how resources and information are used. It would be energy and water aware, operating within the limits of available renewable power rather than demanding constant, industrial-scale supply. It would be smaller, distributed, and integrated into local ecosystems, responsive to weather, land, and community needs.

In other words, instead of extracting intelligence from Hawaiʻi, it would cultivate intelligence within it. If we are not careful, Hawaiʻi will become a server rack for someone else’s empire. Hawaiʻi has a choice.

It can accept another wave of extractive infrastructure — or it can lead the world in defining what responsible, place-based computing looks like.

There is already a word for that future: ea — life, sovereignty, breath. And it is worth building systems that sustain it.