On May 11, the world celebrated World Technology Day, applauding how innovation has transformed human life. Yet, as tech giants race to lead the AI revolution, an inconvenient truth shadows this progress: the enormous and growing environmental toll of artificial intelligence.
The Illusion of Seamless Progress
At 7 a.m., Divya’s smart home gently wakes her. Her AI assistant has already mapped out her commute, summarized her work emails, and reviewed her health data. By 8 a.m., she’s video conferencing with colleagues worldwide, relying on real-time translation. It’s a futuristic life, made possible by AI.
This hyper-connected world, once science fiction, is now reality. AI shapes how we work, live, communicate, and even heal. It writes, creates, predicts—and it’s only just getting started.
But behind this sleek digital frontier lies a grim paradox: these very technologies are accelerating environmental degradation. The AI systems powering our lives are devouring energy, emitting carbon, and pushing Earth’s ecosystems closer to the brink.
A Century-Old Warning
In 1908, scientists predicted that doubling atmospheric CO₂ would increase Earth’s surface temperature by 4°C. Over a century later, we’re edging toward that grim milestone.
Consider this: in North America alone, data centre power demand surged from 2,688 megawatts in 2022 to 5,341 megawatts by the end of 2023—nearly doubling in just one year. According to AI researcher Sasha Luccioni, generative AI uses 30 times more energy than a traditional search engine. One average data centre can consume enough electricity to power 50,000 homes annually.
Meanwhile, electronic waste has soared to 57 million tons a year—comparable in weight to the Great Wall of China.
Tech’s Broken Climate Promises
Despite bold climate pledges, tech giants are fuelling the very crisis they claim to fight. Here’s a snapshot of the growing disconnect between sustainability and AI ambition:
Microsoft
- Climate Pledge: Carbon negative by 2030
- AI Goal: Ubiquitous integration of AI through Copilot, $13B invested in OpenAI
- Reality: Over 10 terawatt-hours of annual data centre consumption; 30% rise in greenhouse gas emissions in 2023
- Climate Pledge: 24/7 carbon-free energy by 2030
- AI Goal: Generative AI in search, productivity, DeepMind, Bard
- Reality: 18.3 terawatt-hours used in 2022; emissions up 48% since 2019
Amazon
- Climate Pledge: Net-zero carbon by 2040
- AI Goal: AI tools in AWS, logistics, Alexa
- Reality: 29 terawatt-hours of electricity consumed in 2022
Meta
- Climate Pledge: 100% renewable electricity
- AI Goal: LLaMA models, AI agents across platforms
- Reality: ~9 terawatt-hours annually; rising demands from large models
In 2020, Microsoft launched its bold “carbon moonshot.” By 2023, its emissions had climbed dramatically. Google’s own environmental report shows a 13% year-over-year emissions increase, totaling a 48% rise since 2019.
At current trends, data centres could consume 1,050 terawatt-hours of electricity by 2026—ranking them fifth among the world’s highest-consuming “nations.” They already account for 1% to 1.5% of global electricity use and 0.6% of total carbon emissions.
By 2030, AI could consume 1% to 3% of global electricity.
Water, the Other Hidden Cost
It’s not just power. Water is increasingly a concern. Data centres require enormous volumes of water to prevent overheating. In the UK, government proposals to scale AI capacity have triggered concerns over future water shortages—a looming environmental cost that isn’t reflected in carbon reports.
The Painful Irony
The real tragedy? AI could help solve environmental crises—if used responsibly. Instead, its rapid expansion is worsening them.
Here’s what training just one AI model can cost:
| Model | Energy Used | CO₂ Emissions |
|---|---|---|
| GPT-3 (OpenAI) | ~1,287 MWh | ~550 tons CO₂e |
| GPT-4 (Estimate) | 3–5x GPT-3 | ~2,000–3,000 tons |
| Google’s PaLM | ~2,500 MWh | ~1,000 tons CO₂e |
| Meta’s LLaMA 2 | 500–1,000 MWh | 200–400 tons CO₂e |
For context, training GPT-3 consumes as much energy as 128 households do in an entire year.
Two Futures
From here, we face two possible futures:
1. Business as Usual
Tech giants continue scaling AI without serious climate accountability. By 2030, AI energy demands soar, emissions skyrocket, water scarcity intensifies, and the global climate crisis deepens.
2. Responsible Transformation
Companies adopt transparency in reporting AI’s carbon footprint. They invest in renewables, cap energy use for training, and embed sustainability into every layer of AI development.
It’s not unprecedented. The Industrial Revolution wreaked environmental havoc until society introduced regulation, cleaner technologies, and new standards. AI now demands a similar reckoning.
Glimmers of Hope
Some firms are beginning to chart a more responsible path. Apple is prioritising on-device AI to cut energy-intensive cloud computing. Salesforce is integrating AI into climate tracking tools. Others are investing in renewable energy and efficient cooling technologies.
According to the latest IPCC report, we must halve global emissions by 2040. Without urgent action, global temperatures could rise 4°C by 2100—a tipping point that could render regions uninhabitable, collapse ecosystems, and deepen inequality worldwide.
The Choice Before Us
As we honoured World Technology Day, we must ask: are we celebrating innovation—or applauding the seeds of our own extinction?
The question is no longer whether we can make machines smarter. It’s whether we can use that intelligence wisely—to protect our planet rather than drain it.
The future of AI must align with the future of Earth.
