In data centers, AI-bolstered computing is continually pushing up the power draw of equipment, putting excessive load on energy grids.
The New York Times reports that artificial intelligence servers shipped worldwide are estimated to guzzle between 85 to 134 terawatt hours (TWh) annually by 2027. According to the peer-reviewed analysis that the article draws on, this is equivalent of what a small country like Argentina, Sweden or the Netherlands, consumes round the year.
This amounts to roughly 0.5% of the current global electricity consumption. But it is not a small number considering that a majority of data centers get their power from nonrenewable energy sources.
Burning of fossil fuels releases planet-warming emissions into the atmosphere which contributes to higher surface temperatures and events like frequent storms, heatwaves and wildfires.
The rising levels of carbon emission can profoundly impact climate change not so long in the future, making the 2050 net-zero target a fantasy.
Managing Environmental Impacts With AI
The soaring energy demands of AI have raised a host of complex questions about the sustainability impacts of the technology.
Tech Field Day, a division of The Futurum Group, has addressed these concerns through candid discussions with corporations and practitioners about emission reduction initiatives and adoption of green technologies.
The conversations suggest that while strong currents of change are flowing across the tech landscape, consumers and providers both share concerns about the environmental impacts of energy-intensive AI technologies.
But former CEO of Microsoft, Bill Gates, holds an optimistic outlook. Gates is excited about AI’s transformative potential. He estimates that the value accrued from AI would far outweigh the energy demand spike, and in some cases even propel decarbonization efforts.
A Microsoft-commissioned report published by PwC UK confirms this. The paper that studies the environmental and economic impacts of AI adoption in the sectors of agriculture, energy, water and transportation, suggests that artificial intelligence can break the mutual exclusivity of economic growth and ecological preservation, and contribute to building a future where technologies like it can sustainably drive advancements, while helping mange environmental issues.
AI applications like precision farming, AI-enabled disaster prediction and green supply chains, are the emerging megatrends that are helping organizations fight the sustainability battle.
The research predicts that environmental applications of AI can boost the global GDP by around 3.1% to 4.4% by 2030. Depending on the maturity of the adoption model, in just the four mentioned sectors, AI can yield approximate gains of US$3.6 to 5.2 trillion for economies.
At par is AI’s promise of sustainability. Exploring the potential of AI when working in tandem with supporting technologies like robotics, automated vehicles and IoT, the paper found that parallelly, AI has the potential to cut down greenhouse gas (GHG) emissions by up to 4.4% or 2.4 gigatons. For perspective, that is the estimated combined annual emission of Canada, Japan and Australia in 2030.
But only select countries will be beneficiaries of this, the study says. Latin American and Sub-Saharan African countries, owing to their low digital readiness, will not be among the most benefited nations.
Rebound effects of deforestation, environmental pollution, and land degradation are already being monitored using AI applications. Data picked up by sensors are analyzed at scale through AI tools to track degradations and changes. This is helpful for creating early warning systems that can result in development of targeted programs to address issues locally.
In food and fashion industries, AI-based waste reduction can help companies meet sustainability goals while obtaining sizeable cost-savings. The Roland Berger report states that AI-aided solutions offer great promise for these industries, and will be widely adopted for reducing waste through the value chain.
