Data centres support AI technologies. They hold servers, storage systems, and networking gear needed to handle large amounts of data. AI is growing fast, especially models like large language models, so data centres need more power and water.
In 2023, data centres in the U.S. used about 4.4% of all electricity. This is expected to triple by 2028 because AI work is increasing. By the 2030s, data centres might use up to 20% of the world’s electricity. This puts pressure on the power grid and causes more greenhouse gas emissions, especially if power comes from fossil fuels.
Training AI models uses a lot of energy. It takes thousands of graphics processing units (GPUs) running nonstop for weeks or months. Studies show a single AI request to tools like ChatGPT uses ten times more electricity than a normal Google search. Training large models over and over increases energy use even more.
Data centres also use a lot of water. Water is needed to cool servers so they don’t overheat. Some advanced cooling systems use large amounts of water, especially in dry regions. Globally, this water use may soon be six times more than Denmark’s total water usage, a country with 6 million people. This can stress local water supplies and affect nearby communities, especially in parts of the U.S. prone to droughts.
High-performance computing gear wears out quickly, creating more electronic waste. Servers, GPUs, and cooling devices don’t last long compared to other equipment. Also, making this hardware needs rare minerals, which often come from mining that harms the environment. The whole cycle of getting and throwing away these materials adds to the environmental effect of AI data centres.
Medical practice administrators who use AI and IT can consider green data centres. These are designed to save energy and resources while reducing harm to the environment.
Power Usage Effectiveness (PUE) measures how energy efficient a data centre is. A PUE near 1 means almost all power goes to IT equipment, not cooling or lighting. Green data centres try to keep PUE low through better design.
New cooling methods help save energy. Free cooling uses outside air to cool servers, cutting down the need for energy-heavy chillers. Liquid cooling takes heat directly from parts, controlling temperature exactly with less power. Also, “hot aisle/cold aisle” setups separate hot and cold air to improve airflow and lower energy use.
Virtualization lets many virtual servers run on one physical server. This lowers the total number of servers and uses less energy. Using all these methods helps cut data centres’ carbon emissions.
A key way to make data centres greener is to use power from renewable sources like solar, wind, and hydropower instead of fossil fuels. In the U.S., using renewable energy in data centres is growing, though some challenges remain.
Renewables greatly cut the carbon impact. For example, a company in Brazil named ODATA uses wind power for its data centres. This lets them run mostly on clean energy and lowers environmental harm. Big cloud providers in the U.S. are trying similar ideas.
But using renewable energy is not always simple. It can be intermittent, so energy storage and grid upgrades are needed. Some data centres spread work across locations and time zones to match when renewable energy is available. This lowers the need to use fossil fuels during busy times.
Certifications like LEED, ISO 14001, ISO 50001, and ENERGY STAR show that a data centre meets certain environmental standards. The Uptime Institute’s Tier Certification checks how well data centres run. These help medical IT providers check if data centres are sustainable and encourage them to be accountable.
Healthcare offices use AI tools more to improve how they work. Tools like front-office phone systems and answering services use AI to help.
AI helps by cutting wait times and making patient communication better. But each AI action uses data centre power and storage. So, every AI interaction adds to the environmental effects we talked about before.
Still, AI can help sustainability if used carefully:
Medical practices in the U.S. have special challenges and chances with AI and data centre sustainability:
AI growth, like the tools made by Simbo AI, creates new challenges for managing data centre environmental impact. Medical practice administrators and IT managers should know about the heavy energy, water, and waste used by AI infrastructure. Data centres need a lot of electricity and water, especially when using fossil fuels. They also create electronic waste from fast hardware turnover.
Green data centres use energy-saving cooling, virtualization, renewable power, and certifications to cut environmental harm. In the U.S., using renewable energy is a good but complex way to reduce emissions and costs.
Healthcare providers using AI for workflow should think about the environment too. Picking green AI services and improving energy use inside the practice can help keep healthcare IT both useful and responsible to the environment.
AI’s environmental problem includes high energy consumption, electronic waste generation, heavy water usage, and reliance on rare minerals often mined unsustainably, leading to significant greenhouse gas emissions and resource depletion.
AI can detect data patterns, anomalies, and predict outcomes, enhancing environmental monitoring accuracy. It helps track harmful emissions such as methane and assists governments and businesses in making sustainable decisions and improving resource efficiencies.
Data centres consume massive electricity, mostly from fossil fuels, generate electronic waste with hazardous substances, require large water quantities for cooling, and depend on rare earth minerals mined destructively, cumulatively impacting the environment negatively.
Higher-order effects include unintended consequences such as increased greenhouse gas emissions from AI-enabled technologies like self-driving cars and the potential spread of misinformation about environmental issues, undermining public awareness and climate action efforts.
UNEP recommends measuring AI’s impact with standardized methods, enforcing transparency in environmental disclosures, improving algorithm and hardware efficiency, greening data centres with renewables, and integrating AI policies into broader environmental regulations.
AI-related infrastructure may consume up to six times more water than Denmark, a country of 6 million, posing serious concerns given global water scarcity and the lack of access to clean water by many.
A single request to an AI virtual assistant like ChatGPT uses ten times more electricity than a Google Search, illustrating AI’s high energy demands relative to conventional digital activities.
The number of data centres surged from 500,000 in 2012 to 8 million currently, driven largely by AI’s explosion, thereby greatly intensifying energy and resource consumption associated with computing infrastructure.
While data centre impacts are better understood, AI’s broader environmental effects are uncertain due to unpredictable application usage, second-order consequences, and potential behavioral changes driven by AI technologies.
Over 190 countries have non-binding ethical AI recommendations including environment, with some legislation in the EU and USA; however, environmental considerations are often absent in national AI strategies, highlighting a policy gap in sustainable AI governance.
Simbo is using AI agents to automate all phone related workflows. Connect now to know more.
Schedule AI Agents Demo Now© Since 2019, Simbo AI.
