The price of AI: profits, power, water – and a warming Asia

Nvidia is riding the artificial intelligence boom to record profits, even as the data centres built on its chips become some of the world's fastest-growing consumers of energy and water.

That surge in demand, where met by fossil fuels, adds to global warming — which in turn is making disasters such as the recent glacial-melt floods in Asia more likely, according to a growing body of climate and industry data.

No single flood or disaster can be attributed solely to AI infrastructure, or to any one company's earnings. What emerges instead is a chain of well-documented trends — rising AI computational demand, rapid data-centre expansion, higher electricity and water consumption, and a warming Asia experiencing faster glacier loss — that together suggest AI's infrastructure boom carries a real, if indirect, climate cost.

Nvidia's AI-driven windfall

Nvidia reported $96.2 billion in quarterly revenue for the second quarter of its 2027 financial year, with data-centre revenue surging 117% year-on-year to $89 billion, driven almost entirely by AI chip sales. In the first quarter, data-centre revenue reached $75.2 billion, up 92% on the previous year, with net income for the quarter at $58.3 billion.

For the full 2026 financial year, Nvidia's data-centre business generated approximately $197 billion in revenue, up from $115 billion the year before, as hyperscale cloud providers and sovereign AI initiatives raced to build what have been termed “AI factories”.

Data-centre revenue now accounts for the majority of Nvidia's overall revenue, driven by sales of chips used for AI training and inference.

Data centres: growing energy and water demands

The International Energy Agency (IEA) estimates that global data centres consumed approximately 415 terawatt-hours (TWh) of electricity in 2024 — around 1.5% of global electricity demand — projecting this could reach roughly 945 TWh by 2030, with AI-focused facilities tripling their power use over that period. In the US alone, data centres used 183 TWh in 2024, about 4% of national electricity consumption, a figure expected to more than double to 426 TWh by 2030.

Where this additional electricity comes from coal, gas or other fossil fuels, it directly increases carbon dioxide emissions. Analysts at the Brookings Institution and elsewhere have flagged this trajectory as a significant challenge for climate goals, warning that AI's rapid compute expansion could outpace efficiency gains and clean-energy deployment unless actively managed.

Water consumption presents a parallel concern. A June 2026 report from the United Nations University Institute for Water, Environment and Health (UNU-INWEH) projects that the electricity powering global data centres by 2030 will carry an associated water footprint equivalent to the basic annual domestic needs of 1.3 billion people in sub-Saharan Africa.

In the US, data centres consumed roughly 17 billion gallons of water directly in 2023, with hyperscale facilities projected to use between 16 and 33 billion gallons annually by 2028. Much of this supports evaporative cooling, which becomes less efficient as temperatures rise — creating a feedback loop in which warmer climates push data centres to consume even more energy and water.

Asia's 2025 climate record: heat, retreating glaciers, and floods

According to the World Meteorological Organization's State of the Climate in Asia 2025 report, last year ranked among the second- to fourth-warmest on record for the region, with mean temperatures 0.96°C above the 1991–2020 average.

All 23 monitored glaciers across High Mountain Asia lost mass in 2025, driven by above-average temperatures and below-average winter snowfall, contributing to an “upsurge in hazards”, including multiple glacial lake outburst floods and glacier collapses.

The report links these shifts directly to escalating climate risks across South and East Asia, including more extreme heat, heavier rainfall, and increased flood hazards. Field studies in 2025 documented exceptional supraglacial glacial lake outburst floods along the China-Nepal border, including one event that released more than seven million cubic metres of water over roughly 35 hours, causing significant loss of life, infrastructure damage and cross-border transport disruption.

Connecting the dots

Taken together, the data suggests a chain worth examining: rising AI demand drives rapid data-centre expansion, sharply increasing electricity and water consumption. Where power grids remain fossil-fuel dependent, that demand raises greenhouse gas emissions, contributing to global and regional warming.

A warmer Asia is associated with faster glacier melt, increasingly unstable glacial lakes, and more intense rainfall — conditions that make disasters such as the 2025 glacial lake outburst floods and monsoon flooding more frequent and severe.

None of this establishes that Nvidia's earnings, or AI infrastructure specifically, caused any individual flood event, including the recent disaster on the Nepal-China border. What the data does indicate is a broader, measurable climate cost associated with the AI infrastructure boom — one reflected not only in emissions statistics, but potentially in the heightened risk of the extreme weather events increasingly observed across the region.

SOURCES: Nvidia, International Energy Agency (IEA), World Meteorological Organization (WMO), United Nations University Institute for Water, Environment and Health (UNU-INWEH), CNBC, DatacenterDynamics, Euronews, Al Jazeera.