The energy and environmental impact of AI and how it undermines democracy – Image for illustrative purposes only (Image credits: Unsplash)
The rapid rollout of AI infrastructure is no longer an abstract concern. Data centers are expanding at a pace that directly challenges energy transitions in key markets, while local communities absorb the immediate costs in higher bills, strained resources and lost control over their surroundings. This pattern has become especially visible in Australia, now the world’s second-largest destination for data center investment after the United States. The developments raise urgent questions about who benefits from the technology and who bears its growing footprint.
Why the Current Buildout Matters More Than Future Promises
Industry claims often point to potential long-term climate gains from AI, yet recent analyses show those benefits remain largely unproven. A February 2026 report backed by Beyond Fossil Fuels found that 74 percent of such assertions lacked solid evidence, and no clear cases emerged where consumer tools like ChatGPT delivered measurable emissions reductions. At the same time, a Greenpeace Germany assessment from 2025 projected that AI-related electricity demand could rise elevenfold by 2030 without stronger oversight. These figures underscore how the infrastructure race itself is locking in higher fossil-fuel dependence rather than easing it.
Australia illustrates the tension clearly. Operators there have proposed facilities that would consume as much power as a small city, with some exploring new gas plants to meet their needs. Without firm limits, the country risks repeating patterns seen elsewhere, where communities face elevated electricity costs and environmental pressures while large technology firms secure priority access to resources.
Local Pushback Reveals a Pattern of Uneven Costs
Resistance has surfaced in multiple countries as residents confront the practical effects of data center projects. In Perth, plans for a three-storey, 120-megawatt facility were withdrawn after concerns over culturally significant sites gained traction. New Brunswick, New Jersey, removed data centers from a redevelopment plan following public input that highlighted noise, water use and grid strain. San Marcos, Texas, saw its city council reject a proposal by a 5-2 vote after more than 100 residents spoke out during an extended meeting.
Similar actions have occurred farther afield. South Dublin County Council in Ireland approved a motion in September 2025 calling for a national moratorium or strict conditions, including full reliance on renewables. In the United Kingdom, campaigners secured the right to challenge a 90-megawatt hyperscale project in Buckinghamshire after officials acknowledged an error in the original approval. These cases show communities rejecting the assumption that they must simply accommodate distant corporate priorities.
Supply-Chain Pressures and Corporate Scale Add to the Strain
The hardware foundation of the AI surge brings its own environmental ledger. Nvidia reported annual revenue of US$215.9 billion, with data centers and AI chips now accounting for the vast majority of that figure. Greenpeace East Asia documented a 4.5-fold jump in emissions from AI chip manufacturing in a single year, driven largely by production in regions still heavily dependent on fossil power. The company ranked last in a supply-chain decarbonization review, highlighting how rapid revenue growth has outpaced efforts to reduce upstream impacts.
Broader corporate trends reinforce the imbalance. Amazon recorded more than US$77 billion in profits in 2025 while reducing its workforce by around 30,000 positions amid heavy AI-related spending. Such outcomes reflect a model in which record earnings coincide with workforce reductions and rising capital outlays, leaving questions about how gains are distributed.
Democratic Safeguards and a Different Direction
The concentration of power in a handful of firms has prompted wider scrutiny of AI’s role in surveillance, content generation and public information. Tools have produced inaccurate responses to election-related questions more than half the time in some evaluations, raising risks for voter understanding. At the same time, movements such as the QuitGPT boycott have drawn more than 2.5 million participants, reflecting unease over military contracts and the normalization of generative systems in daily life.
A more accountable path would require data centers to operate on additional renewable energy, full disclosure of energy and water use, and genuine community consent before projects advance. Governance structures would need stronger public oversight, limits on monopoly influence and protections for privacy and creative labor. These steps would shift technology from an engine of concentrated control toward tools that serve shared needs without eroding local autonomy or ecological stability.
What matters now: The decisions made in the next few years on data center approvals, energy sourcing and regulatory guardrails will determine whether AI infrastructure reinforces existing imbalances or supports more equitable outcomes.
