The Pentagon is preparing to make Palantir’s Maven Smart System an official “program of record,” a move that typically cements long term funding and pushes a tool across the entire force.
In a March 9 letter reviewed by Reuters, Deputy Secretary of Defense Steve Feinberg told senior leaders to embed Maven broadly by the end of the current fiscal year, which closes in September.
It sounds like a pure military and tech story, and to a large extent it is. But AI at scale runs on electricity, and modern combat still runs on fuel, smoke, and supply chains. So what does battlefield AI have to do with climate risk?
Maven becomes the default AI layer
Maven is command and control software that ingests data from satellites, drones, radars, sensors, and intelligence reports, then uses AI to help surface potential threats and targets. The Pentagon’s Chief Digital and Artificial Intelligence Office describes Maven Smart System as a tactical platform for real time object detection, tracking, and decision support in combat operations.
Feinberg’s letter orders oversight to move from the National Geospatial Intelligence Agency to the Pentagon AI office within 30 days, with the Army set to manage future contracting.
Reuters reported Maven has been used in operations tied to thousands of recent targeted strikes against Iran, which shows how quickly software can shift from “pilot” to “infrastructure.” Palantir says the software does not make lethal decisions and humans remain responsible for selecting and approving targets.
Palantir’s win and a supply chain snag
For Palantir, the designation is a business milestone as much as a Pentagon one. Reuters reported the company has landed a stream of government awards, including a U.S. Army deal worth up to $10 billion, helping lift its market value to nearly $360 billion.
A “program of record” can also make budgeting and procurement simpler for the services, which is how vendors turn a contract into a long runway.
The messy part is what sits inside the stack. Reuters has reported that Maven relies in part on Anthropic’s Claude, and the Pentagon recently labeled Anthropic a supply chain risk, ordering a phase out that users say will be difficult and disruptive to execute.
If a core model has to be swapped out, it is rarely a clean plug and play moment, especially in systems that must be recertified for classified use.
AI’s power appetite meets a military climate footprint
The energy side is not theoretical. The International Energy Agency estimates data centers consumed about 415 terawatt hours of electricity in 2024, around 1.5 percent of global power use, and projects demand could roughly double to about 945 terawatt hours by 2030.
In the U.S., Pew Research Center summarized IEA estimates showing data centers used about 183 terawatt hours in 2024, more than 4 percent of national electricity, with a projection of 426 terawatt hours by 2030.
When demand rises that fast, someone has to build power plants, upgrade transmission, or both, and that can get pricey.
Reuters recently reported AI driven data center growth is already reshaping clean energy buying, pushing up long term power purchase agreement prices as demand climbs and project supply tightens.
In everyday terms, it is another large load competing for the same electrons that keep homes cool in summer and factories running year round.
Then there is the military’s own emissions baseline. A 2025 study in PLOS Climate reported that, based on Department of Defense figures, DoD Scope 1 and Scope 2 emissions totaled more than 636 million metric tons of CO2 equivalent from 2010 to 2019, and the authors note this is conservative because it excludes Scope 3 supply chain emissions.
Separate research from the Conflict and Environment Observatory and Scientists for Global Responsibility estimated the global military carbon footprint at about 5.5 percent of worldwide emissions.
Human control, environmental harm, and what comes next
The debate over AI in targeting usually centers on law and ethics, not ecology, but the lines are starting to blur.
The United Nations notes Secretary General António Guterres has called lethal autonomous weapons systems “politically unacceptable and morally repugnant,” and has urged states to agree on a binding instrument by 2026 to prohibit systems that function without human control or oversight.
The UN Security Council has also warned that AI is outpacing governance and could threaten human control over weapons systems.
Environmental damage is part of that same picture, because conflicts often hit energy infrastructure and industrial sites.
In a March 13 statement, the United Nations Environment Programme warned that large oil fires and spills can cause extensive contamination and pose serious health risks from smoke, particulates, and toxic emissions, with effects that can linger long after the fighting slows. AI does not create those pollutants, but a faster tempo can raise the stakes for how often targets are generated, reviewed, and struck.
In the months ahead, the big tell will be what the Pentagon measures, not just what it buys. If Maven becomes a permanent layer of U.S. warfighting, transparency on energy use, emissions, and environmental harm will matter almost as much as speed and accuracy.
The official statement was published on UNEP.
