Blog

The Hidden Costs and Unexpected Gifts of AI in Energy and Nature

Jan Gunash · Co-Founder & CGO · March 9, 2026

The best questions are not always the comfortable ones. Like: what does AI actually cost the planet? Not in carbon, we have started to ask that one. In water. In biodiversity pressure. In grid strain. And then, the question that flips everything: what if AI's enormous appetite is accidentally one of the best things to happen to clean energy?

The answers are emerging, and they are complicated in the best way.

The Water Nobody Is Talking About

We have spent years debating AI's carbon footprint. That conversation matters. But there is another resource quietly disappearing into data centers that almost nobody is talking about: water.

A recent study found that AI server expansion across the United States could generate an annual water footprint ranging from 731 to 1,125 million cubic meters between 2024 and 2030, depending on the pace of expansion (source: Nature Sustainability). That is not a rounding error. That is a river.

Data centers use water in two ways: directly, to cool servers through evaporative systems, and indirectly, through the water embedded in electricity generation. Both add up fast. And unlike carbon, water does not move. When a data center draws from a stressed aquifer in Arizona or a drought-prone region in Southern Europe, that impact is local, immediate, and largely invisible to the people buying the compute.

The same study found that the AI server industry is unlikely to meet its net-zero aspirations by 2030 without substantial reliance on carbon offset and water restoration mechanisms that remain deeply uncertain (source: Nature Sustainability). In other words, the current trajectory has a gap between the commitments being made and the infrastructure being built.

This is not an argument against AI. It is an argument for honesty about what we are building and where we are building it. The data exists. The question is whether the industry is willing to treat water the way it is beginning to treat carbon: as a real cost that shows up on a real balance sheet.

The Biodiversity Data Explosion Nobody Can Read

At the same time AI is straining water systems, something remarkable is happening in nature data. Early 2026 has seen a wave of new datasets, platforms, and satellite products enter the nature data space, from ecosystem integrity mapping to mangrove monitoring, mining risk, and AI-powered land observation (source: Nature Tech Collective). NASA released harmonized vegetation index products combining Landsat and ESA Sentinel satellites. The Global Biodiversity Information Facility expanded marine and freshwater spatial filters. Open AI-powered land monitoring systems now let any user track ecosystem changes in plain language, getting back answers with satellite imagery, maps, and statistics.

The problem is not the data. For the first time, we have enough of it. The problem is the same one facing energy data: fragmentation, access barriers, and a lack of shared standards that would let all of this information actually talk to each other.

Environmental DNA is a good example of where things are heading. By collecting samples from water, soil, or even air, scientists can now produce comprehensive species lists and biodiversity metrics for a fraction of the cost of traditional surveys (source: NatureMetrics). Companies are using this for regulatory disclosure under frameworks like TNFD and CSRD. Conservationists are using it to monitor ecosystems at scales that were impossible five years ago. But the data still lives in silos. Different labs. Different formats. Different standards for what counts as a valid sample.

The 2026 Global Horizon Scan, published in Trends in Ecology and Evolution, flagged low-power TinyML devices and optical AI chips as emerging tools that could enable real-time biodiversity detection in remote landscapes without internet connectivity (source: UNEP-WCMC). That is genuinely exciting. But the same scan raised the question of transparency, data access, and whether these tools will serve communities with limited digital infrastructure or simply add another layer of complexity that benefits the already-resourced.

More data without better data infrastructure is just louder noise. The biodiversity crisis does not need more satellite images nobody can act on. It needs what energy data needs: standards, open access, and platforms built for collaboration rather than proprietary advantage.

The Accidental Catalyst

Here is the twist most people miss.

AI's enormous appetite for electricity, the thing that is straining grids and drawing down water tables, may also be one of the most powerful forces accelerating the clean energy transition.

Renewables accounted for over 90 percent of new utility-scale generating capacity added in 2024, partly because they are faster to deploy and easier to scale than fossil fuel alternatives (source: Optera). That is not a coincidence. When hyperscalers and data center operators need guaranteed, large-scale, long-term power and they need it fast, the math keeps pointing to solar and wind. They are cheaper. They are quicker to permit and build. And increasingly, they are the only way to credibly hit the net-zero commitments that investors and regulators are watching.

The result is that AI's power hunger is becoming an inadvertent catalyst for clean energy investment at a scale that climate policy alone has struggled to achieve (source: Optera). Utilities and governments are accelerating renewable projects not because of Paris Agreement targets but because a data center is coming online in eighteen months and it needs a guaranteed power purchase agreement today.

This does not make AI's environmental costs disappear. But it changes the shape of the problem. The question is no longer whether AI will drive energy demand. It will. The question is whether the infrastructure built to feed that demand will lock in fossil fuels or accelerate the transition off them. Right now, the economics are increasingly favoring clean energy. That is worth paying attention to.

What Needs to Happen

These three forces, water consumption, biodiversity data fragmentation, and clean energy acceleration, are not separate stories. They are the same story about what happens when a powerful technology scales faster than the governance and infrastructure around it.

Water accounting needs to become standard practice. The energy sector has spent years building carbon reporting frameworks. Water deserves the same rigor, the same transparency, and the same expectation that companies disclose where their compute is drawing from and what that means for local ecosystems.

Nature data needs the same infrastructure push that energy data is starting to get. Common formats, open repositories, and collaborative platforms where satellite data, eDNA results, and field observations can be layered together and acted on. The data is arriving. The pipes to move it usefully are still missing.

And the clean energy opportunity created by AI demand needs to be seized deliberately, not left to market forces alone. Governments and utilities that move fast to build renewable capacity now are positioning themselves to capture an enormous and durable source of demand. Those that hesitate are likely to end up locking in fossil fuel contracts that will look very expensive very quickly.

The Path Forward

The intersection of AI, energy, and nature is not a simple story of salvation or destruction. It is a systems challenge that rewards honest accounting and penalizes wishful thinking.

AI is thirsty, and we should know exactly what it is drinking and from where. AI is generating more nature data than we have ever had, and we should build the infrastructure to actually use it. And AI's power demand, for all its costs, is pushing clean energy deployment at a pace that climate commitments alone could not achieve.

The technology is moving. The question is whether the data, the standards, and the governance can move with it.

That is the work worth doing.

Sources:

Want to contribute your thoughts or findings in data and AI with climate, energy, or nature? Message Jan Gunash on LinkedIn or send an email to jgunash@projectcoeus.com.