The U.S. still has its sights on winning the global AI race. First stop: Commandeering AI manufacturing.
Announced just last week, a $500 billion infrastructure investment from artificial intelligence giant Nvidia will bring domestic AI manufacturing to the U.S. — that’s half a trillion dollars going toward mass production of the the country’s own AI supercomputers as well as NVIDIA’s Blackwell chips.
The AI supercomputers will take over a million square feet of manufacturing space in Texas, while factories and manufacturing partners across Arizona — operated by the Taiwan Semiconductor Manufacturing Co., which landed a similar deal in March — will be tasked with building and testing chips. Proponents say it’s a welcome investment in the country’s growing AI economy, potentially boosting jobs and aiding in the development of an AI workforce. In the words of Nvidia CEO Jensen Huang: “The engines of the world’s AI infrastructure are being built in the United States for the first time.”
But while the investment may bode well for the country’s position in the AI race, a recent report from Greenpeace suggests an additional worry for such hardware manufacturing chains and AI data centers, at large: Their voracious consumption of electricity.
AI manufacturing eats away at power supply
According to research from Greenpeace East Asia, electricity consumption linked to AI hardware manufacturing increased by more than 350 percent between 2023 and 2024 — It’s expected to increase another 170-fold in the next five years, according to Greenpeace, exceeding the total amount of power consumed by the population of Ireland.
Global hubs for AI manufacturing in East Asia, including Taiwan, South Korea, and Japan, are the largest consumers of electricity and are increasingly reliant on climate destructive fossil fuels, the report finds. Unlike other similarly foreboding reports, these figures apply to the early lifecycle of an AI-powered product, including the creation and testing of chips, and not just the processing power used by AI supercomputers like those built by Nvidia.
“While fabless hardware companies like Nvidia and AMD are reaping billions from the AI boom, they are neglecting the climate impact of their supply chains in East Asia,” said Katrin Wu, Greenpeace East Asia supply chain project lead. “AI chipmaking is being leveraged to justify new fossil fuel capacity in Taiwan and South Korea – demand that could, and should, be met by renewable energy sources. Across East Asia, there are many opportunities for companies to invest directly in wind and solar energy, yet chipmakers have failed to do so on a meaningful scale.”
The need for more and more energy sources
As enthusiasm for AI has exploded over the last several years, so too has its demands on the globe’s already strained energy sources. In the words of experts, AI is an energy hog.
Mashable Light Speed
According to a report by the International Energy Agency (IEA), “the U.S. economy is set to consume more electricity in 2030 for processing data than for manufacturing all energy-intensive goods combined, including aluminium, steel, cement, and chemicals.”
Half of the growth in U.S. electricity demand between now and 2030, which is expected to at least double, will be due to AI — currently, around 40 percent of data centers in the U.S. are supplied by gas power plants. Renewable energy sources, like wind and solar, won’t be able to match the need, says the IEA, necessitating a further reliance on large scale fossil fuels like gas and coal and potentially bottlenecking states’ emissions goals.
The issue doesn’t just pertain to AI’s immense processing power. “The rapidly rising energy costs of AI data centers have captured global headlines, yet the environmental implications of other parts of the hardware lifecycle are often overlooked,” said Greenpeace report author Alex de Vries.
On April 14, President Donald Trump announced a plan to revitalize the U.S. coal industry, including protecting coal-fired power plants and expediting leases for domestic coal mining that would also supply hungry AI data centers. But while coal power is remarkably cleaner than it was in generations past, it’s not a viable path toward reducing carbon emissions.
Similar to other manufacturing sectors, such power demands are inequitably shared among global regions, as well. “The manufacturing process of AI hardware is energy intensive and carries a significant environmental footprint, especially considering the concentration of this manufacturing in East Asia, where power grids still rely heavily on fossil fuels, and chipmakers have taken few steps to procure renewable energy,” de Vries warns.
Consumer demand also exacerbates energy costs. Some researchers have said just a single AI chatbot query consumes the same amount of electricity as what’s required to light a bulb for 20 minutes, while others point to the growing water footprint created by cooling systems for AI servers. According to researchers at the University of California, Riverside, a user asking between 10-50 ChatGPT queries per day uses up about two liters of water.
How AI proponents are tackling sustainability
Even so, AI has the potential to revitalize the need for renewable energy. According to the IEA, the continued AI boom could spur investment in diverse energy sources and cement the importance of renewables and natural gas sources. AI could also accelerate “innovation in energy technologies,” the IEA contends.
For example, some AI manufacturers have sought nuclear power options in response to growing grid demands. Microsoft, Google, and Amazon have announced plans to secure nuclear energy deals to power their in-house AI products, including reopening the Three Mile Island plant in Pennsylvania. Bipartisan lawmakers are seeking exemptions on nuclear power to support a cleaner energy option, too. The relationship is reciprocal: Nuclear power facilities help supply energy to AI processing demands, while AI-powered technology may offer a solution to the complicated maintenance of nuclear reactors.
Other companies, including controversial claims by Chinese-owned OpenAI competitor DeepSeek, are finding ways to reduce the amount of processing power needed to feed their models.
But such energy alternatives need continued investment from AI’s major players, from the companies selling AI products to the manufacturers and politicians aiding in their creation. And as the Trump administration and other U.S. political leadership have struck down its commitments to climate and environmental stewardship, and slashed at the country’s climate science infrastructure, concern over the technology’s environmental strain remains.
