AI manufacturing boom triggers global surge in energy use and emissions

AI manufacturing is accelerating—but so is its environmental toll

A new Greenpeace report highlights how global energy inequities and fossil fuel dependence are deepening as AI expands

As AI supercomputers and chip production surge across the globe, so too does a quieter crisis: the staggering environmental footprint of the artificial intelligence industry. While nations like the United States race to dominate AI innovation, organizations such as Greenpeace are sounding alarms over the escalating energy consumption and fossil fuel dependency driven by AI hardware production and data processing.

Last week, Nvidia announced a $500 billion U.S. infrastructure investment, bringing AI chip fabrication and supercomputer production stateside. But a Greenpeace East Asia report, released in tandem with this news, paints a grimmer picture of the industry's unchecked growth and its global power imbalance—particularly in East Asia, where most AI hardware is built and electricity is increasingly sourced from climate-damaging fossil fuels.

AI chipmaking is outpacing clean energy capacity

Greenpeace research reveals that electricity consumption tied to AI hardware manufacturing jumped 350% between 2023 and 2024. Over the next five years, it's projected to grow 170-fold, eclipsing the annual electricity use of entire nations—Ireland among them.

These figures don’t even include the high-profile AI data centers in the U.S., but rather focus on the early lifecycle stages of AI technology: chip fabrication, testing, and packaging. And these operations are heavily concentrated in Taiwan, South Korea, and Japan—regions still dependent on coal and gas.

“Fabless giants like Nvidia and AMD profit from the AI boom while ignoring its environmental costs,” says Katrin Wu, Greenpeace East Asia supply chain project lead. “Their supply chains are justifying new fossil fuel infrastructure.”

Wu adds that renewables like wind and solar remain underutilized in East Asia, despite their availability, as chipmakers continue relying on traditional grids.

Domestic production doesn’t mean a cleaner footprint

In the U.S., enthusiasm over Nvidia’s investment—which includes supercomputer facilities in Texas and chip manufacturing in Arizona via TSMC—has been framed as a win for jobs and tech leadership. Nvidia CEO Jensen Huang said the facilities will be “the engines of the world’s AI infrastructure.”

But the International Energy Agency (IEA) offers a stark counterpoint: by 2030, U.S. electricity consumption for AI processing alone is set to exceed energy used for all heavy industry combined, including steel and cement.

Currently, 40% of American data centers run on natural gas. Renewable energy will not scale fast enough to meet projected demand, potentially forcing states to fall back on coal and gas, imperiling emissions goals and clean energy targets.

More than just power: AI's full environmental burden

The Greenpeace report also highlights parts of the AI lifecycle that receive less scrutiny than data centers. From chip fabrication to testing, energy usage is immense. And it’s not just electricity: cooling systems that support AI infrastructure carry a growing water footprint.

According to researchers at UC Riverside, a user interacting with an AI chatbot like ChatGPT 10–50 times a day consumes about 2 liters of water, used indirectly in cooling the model’s supporting servers.

“AI's full lifecycle, not just inference and training, must be considered in climate planning,” warns Alex de Vries, the report’s author.

Governments and corporations face mounting choices

The Trump administration's recent move to revitalize the coal industry, including fast-tracking leases for domestic coal mining, could further embolden AI-linked fossil fuel use. The administration frames coal as cleaner than in the past—but most scientists caution it’s still incompatible with emissions reduction goals.

And while some AI manufacturers have turned to alternatives like nuclear power—with Microsoft, Amazon, and Google pursuing nuclear energy deals—such strategies remain limited in scale. In one notable example, tech companies plan to reopen the Three Mile Island nuclear plant to power future AI operations.

AI may also drive energy innovation, the IEA notes, but the current momentum favors fossil expansion over sustainability.

The AI sector must take climate leadership seriously

As AI grows into a foundational pillar of the global economy, its environmental footprint becomes harder to ignore. Companies profiting from AI innovation—including chipmakers, data center operators, and software developers—will need to invest aggressively in clean energy, or risk compounding the climate crisis they may also claim to solve.

“AI has the potential to accelerate climate solutions,” says de Vries. “But without responsibility in manufacturing and energy sourcing, it risks becoming a driver of the problem.”

With massive public and private investments on the horizon, including from policymakers seeking to cement U.S. tech dominance, the question becomes whether sustainability will scale alongside performance—or be sacrificed in its shadow.

Stay with The Horizons Times for ongoing coverage of AI innovation, global manufacturing, and the climate cost of emerging technologies.