The Coming Surge in Electricity Demand

The most staggering projection in the UN report concerns electricity consumption. By 2030, global data centers powering AI are projected to consume 945 TWh of electricity annually . This figure represents roughly 3% of the world’s total electricity production and is nearly triple the combined annual electricity use of Pakistan, Bangladesh, and Nigeria—countries that are collectively home to more than 650 million people .

The growth is driven almost entirely by AI. The power demand from AI-specific data centers is expected to be 11 times higher in 2030 than it was in 2023. At that point, AI workloads alone will require as much electricity as all conventional data centers consume today, making AI a primary force behind energy demand in the entire data infrastructure sector .

Carbon Emissions: Adding Millions of Cars to the Road

This spike in energy use carries a heavy carbon cost. The report projects that by 2030, AI data centers will emit an additional 24 to 44 million metric tons of CO₂ annually . To put that in context, the upper end of that estimate is the emissions equivalent of adding 10 million cars to U.S. roadways . Other characterizations in the report’s coverage compare total data center emissions to the entire annual emissions of the United Kingdom, potentially reaching 400 million tonnes of CO₂ equivalent .

These emissions are projected to rise even under the assumption of a carbon-neutral electricity supply by 2040, highlighting the sheer scale of AI's expanding energy demand .

A Hidden Thirst: The Water Footprint of AI

While much public discussion centers on electricity and carbon, the report emphasizes that water is a critically overlooked resource. The water required to cool the servers running complex AI models is immense.

In the United States alone, the deployment of AI servers is projected to consume between 731 and 1,125 million cubic meters of water per year by 2030—roughly equal to the annual household water usage of 6 to 10 million Americans . On a global scale, the water footprint becomes staggering. The report finds that AI’s total water consumption in 2030 will be equivalent to the basic annual domestic water needs of 1.3 billion people in sub-Saharan Africa .

Land, E-Waste, and a Lifecycle of Impact

The environmental cost of the AI boom doesn’t end with operational resources. The report highlights two more critical dimensions of its footprint:

• Land Use: The physical expansion of data centers places heavy pressure on scarce land resources, competing with agriculture, housing, and natural ecosystems. The land required to operate AI—including its infrastructure and supply chains—is projected to exceed 14,500 square kilometers, an area roughly twice the size of the Jakarta metropolitan area, which houses over 32 million people .
• Electronic Waste: The hardware behind AI, from specialized chips to entire server racks, has a limited lifespan, generating electronic waste at a mounting scale. The report adds to growing concerns about an e-waste crisis, with one projection likening the annual waste to discarding 250 Eiffel Towers .

Calls for Radical Transparency and Recognizing Trade-Offs

The UNU-INWEH report is not just a catalog of problems; it is a policy call to action. It warns that the true cost of AI extends across the entire lifecycle, and that this cost is being distributed unjustly, with developing nations bearing a disproportionate share of the environmental burdens while often seeing fewer of the economic benefits .

The report’s central demand is for mandatory, standardized environmental reporting.

• It urges AI firms to disclose their full environmental footprint, moving beyond simple carbon metrics to account for water, land, and material impacts .
• It calls on governments to require standardized reporting from AI providers, creating a level playing field and a clear picture of the sector’s total impact .
• It frames the future of AI as a series of trade-offs, where the benefits of the technology must be weighed against accelerating resource depletion and pollution. The industry, the report concludes, cannot continue to grow without accounting for its full lifecycle impacts .

The report even places a share of responsibility on end-users, suggesting they choose less energy-intensive AI tools for tasks when possible, a small step that could collectively reduce pressure on strained power grids and water supplies .

The Coming Surge in Electricity Demand

The most staggering projection in the UN report concerns electricity consumption. By 2030, global data centers powering AI are projected to consume 945 TWh of electricity annually . This figure represents roughly 3% of the world’s total electricity production and is nearly triple the combined annual electricity use of Pakistan, Bangladesh, and Nigeria—countries that are collectively home to more than 650 million people .

The growth is driven almost entirely by AI. The power demand from AI-specific data centers is expected to be 11 times higher in 2030 than it was in 2023. At that point, AI workloads alone will require as much electricity as all conventional data centers consume today, making AI a primary force behind energy demand in the entire data infrastructure sector .

Carbon Emissions: Adding Millions of Cars to the Road

This spike in energy use carries a heavy carbon cost. The report projects that by 2030, AI data centers will emit an additional 24 to 44 million metric tons of CO₂ annually . To put that in context, the upper end of that estimate is the emissions equivalent of adding 10 million cars to U.S. roadways . Other characterizations in the report’s coverage compare total data center emissions to the entire annual emissions of the United Kingdom, potentially reaching 400 million tonnes of CO₂ equivalent .

These emissions are projected to rise even under the assumption of a carbon-neutral electricity supply by 2040, highlighting the sheer scale of AI's expanding energy demand .

A Hidden Thirst: The Water Footprint of AI

While much public discussion centers on electricity and carbon, the report emphasizes that water is a critically overlooked resource. The water required to cool the servers running complex AI models is immense.

In the United States alone, the deployment of AI servers is projected to consume between 731 and 1,125 million cubic meters of water per year by 2030—roughly equal to the annual household water usage of 6 to 10 million Americans . On a global scale, the water footprint becomes staggering. The report finds that AI’s total water consumption in 2030 will be equivalent to the basic annual domestic water needs of 1.3 billion people in sub-Saharan Africa .

Land, E-Waste, and a Lifecycle of Impact

The environmental cost of the AI boom doesn’t end with operational resources. The report highlights two more critical dimensions of its footprint:

• Land Use: The physical expansion of data centers places heavy pressure on scarce land resources, competing with agriculture, housing, and natural ecosystems. The land required to operate AI—including its infrastructure and supply chains—is projected to exceed 14,500 square kilometers, an area roughly twice the size of the Jakarta metropolitan area, which houses over 32 million people .
• Electronic Waste: The hardware behind AI, from specialized chips to entire server racks, has a limited lifespan, generating electronic waste at a mounting scale. The report adds to growing concerns about an e-waste crisis, with one projection likening the annual waste to discarding 250 Eiffel Towers .

Calls for Radical Transparency and Recognizing Trade-Offs

The UNU-INWEH report is not just a catalog of problems; it is a policy call to action. It warns that the true cost of AI extends across the entire lifecycle, and that this cost is being distributed unjustly, with developing nations bearing a disproportionate share of the environmental burdens while often seeing fewer of the economic benefits .

The report’s central demand is for mandatory, standardized environmental reporting.

• It urges AI firms to disclose their full environmental footprint, moving beyond simple carbon metrics to account for water, land, and material impacts .
• It calls on governments to require standardized reporting from AI providers, creating a level playing field and a clear picture of the sector’s total impact .
• It frames the future of AI as a series of trade-offs, where the benefits of the technology must be weighed against accelerating resource depletion and pollution. The industry, the report concludes, cannot continue to grow without accounting for its full lifecycle impacts .

The report even places a share of responsibility on end-users, suggesting they choose less energy-intensive AI tools for tasks when possible, a small step that could collectively reduce pressure on strained power grids and water supplies .