The technology giants behind the artificial intelligence revolution are coming under intensifying scrutiny as their sprawling data centre operations pump increasing quantities of carbon into the atmosphere. With public opposition growing on both sides of the Atlantic, calls are mounting for governments to impose dedicated levies on the industry—effectively an AI tax—to account for the environmental damage inflicted by facilities that consume electricity equivalent to entire nations.
The Scale of the Problem
America’s largest technology corporations—Amazon, Microsoft, Google, and Meta—have amassed a combined market value exceeding $12 trillion, roughly twice the economic output of Germany. Yet their environmental credentials stand in stark contrast to their financial muscle. According to analysis by the Financial Times, the 60 largest planned data centres across these four companies alone would generate more than 100 million tonnes of carbon dioxide annually once operational. To contextualise that figure: it matches the yearly emissions of 27 coal-fired power stations or approximately 24 million petrol-driven vehicles.
The situation shows no signs of abating. Amazon disclosed that its carbon footprint expanded by 16 percent in 2024 compared with the previous year, while Microsoft reported a 25 percent surge in emissions over the same timeframe. As these companies accelerate their infrastructure rollouts, analysts anticipate such double-digit increases could become the standard trajectory—placing their publicly-stated net-zero commitments, including Amazon’s 2040 target, under serious question.
The Energy Dilemma
Data centres require vast quantities of electricity to power servers and, critically, to manage the extraordinary heat they generate. Cooling systems are essential, and the International Energy Agency reveals that natural gas currently supplies over 40 percent of the electricity consumed by existing American data centres. Renewables account for 24 percent, nuclear for 20 percent, and coal for the remaining 15 percent.
Here lies the fundamental contradiction at the heart of the AI industry’s environmental strategy. While technology executives publicly champion emissions-free futures powered by nuclear reactors, solar farms, and wind turbines, the reality on the ground tells a different story. New gas-fired power stations can be commissioned within three years; shuttered plants can sometimes be restarted within months. By contrast, nuclear facilities frequently require decades to progress from planning to full generation capacity. Faced with this timeline disparity, companies are inevitably gravitating towards the quickest and cheapest energy solutions—which means fossil fuels.
A June report from the Berkeley Lab in California suggests American data centres could account for between 11.8 and 15 percent of total national electricity consumption by 2030. That projection raises uncomfortable questions about resource allocation as the planet warms: will sufficient power remain available to cool both residential homes and these computational hubs, and at what cost to consumers?
Calls for an AI Levy
Public sentiment is hardening against the industry. Research from Gallup indicates that 70 percent of Americans would oppose data centre construction in their local areas, citing environmental concerns alongside anxieties about job displacement and distrust of AI-driven medical recommendations. Canadian polling reveals similar reservations, with most respondents opposing government subsidies for AI infrastructure.
The argument for taxation rests on a straightforward principle: corporations routinely privatise profits while externalising environmental costs onto society at large. AI companies benefit from taxpayer-funded incentives including property and sales tax exemptions, generous tax credits, and subsidised electricity rates. An AI tax would recalibrate this imbalance.
Virginia, home to one of America’s densest concentrations of data centres, has emerged as a pioneer. State legislators approved an electricity consumption levy specifically targeting AI facilities, projected to generate $600 million in its first full year. Economists have proposed various alternatives: taxes on electricity usage, charges based on carbon intensity, or revenue-linked excise duties. Proceeds could fund retraining programmes for workers displaced by automation, subsidise renewable energy development, or cushion household electricity bills from any cost increases.
Why it Matters
The AI sector presents itself as an engine of future prosperity, yet its environmental footprint is already substantial and expanding rapidly. As extreme heat bakes Europe, rivers recede, and agricultural yields decline, the incongruity of data centres consuming ever-greater resources while receiving public subsidies grows harder to stomach. The technology industry’s combined valuation dwarfs the GDPs of most nations—these are not struggling enterprises requiring protection. If the AI revolution is to be sustainable, both environmentally and politically, the sector must bear a fairer share of its own costs. An AI tax is not a punishment; it is recognition that innovation without accountability is simply another form of externalisation. Governments that fail to act risk becoming complicit in the very climate crisis they purport to address.
