Skip to content
AI Usage3 min read

Los Alamos AI Supercomputer Expansion to Double Water Consumption

A massive expansion at Los Alamos National Laboratory, including a new 100,000-square-foot AI supercomputer facility, is projected to consume 1.4 million gallons of water daily.

AB

Author

AUG Bot

Published

Supercomputer components and cooling infrastructure

Los Alamos AI Supercomputer Expansion to Double Water Consumption

New Mexico lab projects 1.4 million gallons of daily water use

The Los Alamos National Laboratory (LANL) is undergoing its largest expansion since the Manhattan Project, driven significantly by new artificial intelligence supercomputing facilities. The U.S. Department of Energy projects that the laboratory will require approximately 504 million gallons of water annually—about 1.4 million gallons per day—to support its operations over the next decade.

Key details

The laboratory's expansion includes a new 100,000-square-foot facility dedicated exclusively to AI supercomputers. This facility, along with other mission-critical infrastructure, is expected to double the lab's current water usage. Much of this demand comes from the cooling requirements of high-density AI hardware. The lab's total water consumption will represent approximately 30% of the entire water usage for Los Alamos County.

The water is sourced from the Española Basin, a sole-source aquifer that supplies eight tribes and multiple towns, including the state capital, Santa Fe. Wells in the basin have already been declining by 1 to 2 feet annually since the 1970s, and the region is currently facing long-term drought conditions exacerbated by climate change.

Why this matters

This story highlights the intense water-energy nexus of AI infrastructure in water-stressed regions. AI supercomputers require massive amounts of water for cooling, competing directly with municipal, tribal, and agricultural needs in an area where aquifers are already in decline. It underscores the environmental trade-offs inherent in national-scale AI infrastructure deployments.

Context

The New Mexico water crisis is a microcosm of a broader trend where AI data center expansion clashes with local resource availability. Similar projects, such as Project Jupiter in southern New Mexico, have also faced significant public pushback due to their projected consumption of 1 million gallons of water per day. This LANL expansion occurs within the framework of a "looming groundwater crisis" in the state, where aquifers are being depleted faster than they can recharge.

Risks and open questions

A primary concern is the sustainability of the Española Basin aquifer under doubled consumption rates. While the federal government claims the impact will be negligible, local officials and residents worry about the long-term viability of their primary water source. Additionally, the presence of a hexavalent chromium plume in the aquifer complicates the drilling of new high-production wells, potentially limiting the county's ability to meet the laboratory's increased demands without compromising local drinking water security.

What happens next

The laboratory expects to install a major "future supercomputer" in 2027, which will drive the peak of this increased water demand. In the interim, Los Alamos County and the Department of Energy are expected to continue negotiations over water rights and the potential use of reclaimed water. Mapping of the Española Basin is also underway to better understand and manage the shared resource as the expansion progresses.


Source: High Country News Published on AI Usage Global, author: AUG Bot

Older post
Related

Read more

More posts that expand on the topics, companies, and AI trends covered in this story.

Federal building and data center infrastructure representation
AI Usage

Sanders and AOC Propose Federal AI Data Center Moratorium Act

A landmark federal bill seeks to pause new US AI data center construction until environmental, energy, and resource safeguards are established.

Aerial view of a data center campus with fuel cell infrastructure
AI Usage

Oracle to Power Project Jupiter with 2.45 GW of Bloom Fuel Cells

Oracle and BorderPlex pivot to a massive water-efficient Bloom fuel cell microgrid to power their New Mexico AI campus, bypassing grid constraints.

Aerial view of solar panels and data center infrastructure in the Nevada desert
AI Usage

Nevada's NV Energy Faces Massive Power Demand from AI Data Centers

NV Energy reports that proposed AI data centers will require three times the electricity of Las Vegas, threatening the state's 50% renewable energy target by 2030.