Meta CEO Mark Zuckerberg’s recent manifesto painted a rosy picture of a future utopia that can be created with Meta’s “superintelligence.”

The essay pushes back against AI doomers, arguing that the democratization of AI, as broadly as possible, is the path to both economic growth and safety.

AI, via data centers, has begun consuming water at a significant rate due to the cooling systems required to operate them.

Within the essay, Zuckerberg says that Meta plans to become water-positive by 2030 with a goal of restoring 200% of consumption in high-water-stress areas. But what does this mean in practice? Is this a positive development or PR management?

To gain a better insight, Channel Insider asked an expert from a water technology company. Read our Q&A with Kevin Gast, CEO and Chairman of VVater.

What “water positive” should mean for AI data centers

From a water technology perspective, what does “water positive” actually mean, and how should companies measure whether they are genuinely offsetting the water their data centers consume?

“Water-positive” should mean more than replenishing a greater volume of water than you consume. Companies should measure where the water was withdrawn, when it was used, what condition the watershed was in at the time, and whether restoration actually improves the same local water system affected by the facility. 

A credible approach should track direct consumption, upstream water use, reuse rates, local watershed conditions, and the quality and timing of restored water, not just an annual gallon-for-gallon balance.

Meta says that it plans to restore 200% of the water it consumes in high-water-stress areas. From your perspective, does that actually address the local impact of a data center’s water consumption, or can “water-positive” claims obscure the realities of where and when water is being used?

A 200% restoration goal is meaningful, but the number alone does not tell us whether the local impact has been addressed. If a facility consumes water during the hottest, driest months and restoration occurs later, elsewhere in the watershed, or in a form that does not immediately increase usable supply, the local system can still experience stress. 

The right question is not only how much water is restored, but whether that restoration improves availability and resilience where and when the demand occurs.

Is restoring water elsewhere an adequate substitute for reducing the amount of freshwater a data center consumes at the source, particularly when the facility is operating in a water-stressed community?

I do not think restoration should be treated as a substitute for reducing freshwater consumption at the source. 

The first priority should be to reduce the amount of potable or high-quality freshwater a facility needs through reuse, recycling, alternative water sources, and more efficient cooling. 

Restoration is valuable, but the strongest strategy is to reduce local demand first and then replenish the watershed on top of that.

What questions should the industry be asking about Meta’s water-positive commitment that aren’t necessarily answered by the headline 200% restoration figure?

The industry should ask where the restored water is going, how quickly it becomes available, whether it is usable by the same communities and ecosystems affected by the facility, and whether the calculation includes the broader water footprint associated with electricity generation and equipment manufacturing. 

We should also ask how much of a facility’s cooling water is reused, what percentage comes from potable municipal supplies, and what happens during drought or extreme heat. The 200% figure is useful, but it is only one metric in a much larger water strategy.

How data centers can reduce freshwater consumption

What are the most effective technologies available today for reducing freshwater consumption in data center cooling, particularly as AI increases power and cool requirements?

The biggest opportunities are closed-loop cooling, water reuse, advanced treatment, and the use of reclaimed or lower-quality water sources instead of potable freshwater where technically appropriate. 

Decentralized treatment can also allow facilities to treat and reuse water on-site rather than continually drawing from municipal systems. The goal should be to keep the same water circulating for as long as possible and reduce the amount of new freshwater required to replace losses.

Where do you see the biggest opportunities and gaps for technology providers to help data center operators build more water-efficient infrastructure as AI deployment accelerates?

The biggest opportunity is moving from a linear model of withdrawal, use, and discharge to a circular model built around treatment and reuse. 

Technology providers can help operators use lower-quality source water, recover more water from their own operations, monitor water quality in real time, and build closed-loop systems that reduce dependence on municipal freshwater. 

The biggest gap is that water strategy is still too often treated as a utility issue instead of a core infrastructure decision. AI companies plan power years in advance. Water needs to receive the same level of attention.

Why water needs to shape AI data center site selection

When companies select sites for new AI data centers, what water-related factors should they evaluate beyond simply whether a site has access to an adequate water supply?

Availability is only the starting point. Operators should evaluate long-term watershed stress, competing community and agricultural demand, drought exposure, water quality, wastewater capacity, permitting requirements, treatment infrastructure, reuse opportunities, and how demand changes during periods of extreme heat. 

They also need to understand whether the local utility can support the facility without requiring major upgrades or shifting costs onto residents. 

Water needs to be part of site selection from day one, not something addressed after the land and power are already secured.