How Can Big Tech Companies Get To Grips With AI’s Energy Footprint?
Companies worldwide are facing a double-edged sword when it comes to investing in new forms of AI, with research showing the vast energy consumption of the technology could counteract corporate policies on reducing carbon output.
According to research by Goldman Sachs, AI is posed to drive an alarming 160% increase in data center power demand by 2030 – leading to their carbon dioxide emissions more than doubling.
Also Read: AI helps Data Engineers be Distinguished Data Engineers
US utilities will need to invest around $50 billion in new generation capacity just to support data centers alone. But where does this leave business leaders who are trying to balance the need for innovation with environmentally friendly practices?
Indeed, the higher the share of fossil fuels in the present energy landscape, the more companies have to deal with rising CO2 emissions as a result of the expansion and construction of new data centers. For example, Google’s CO2 emissions have increased by 48 percent in five years due to the rapid growth of AI. According to the International Energy Agency (IEA), training a single AI model consumes as much electricity as 100 US households in a year. The IEA therefore predicts that data centers will be responsible for six percent of the global CO2 balance by 2030 – compared to 1 to 1.6 percent in 2022.
This rising power demand for AI data centers doesn’t pose an immediate crisis for the US power grid (according to the Columbia’s Center on Global Energy Policy), however better energy efficiency and renewable sources are ‘pivotal’ for the future of AI and the environment.
To address the challenge of meeting their climate commitments while investing in AI, some tech giants have decided to enter into agreements to offset CO2 emissions through carbon credits. While it can help them achieve their goals now, it’s more of a short-term stopgap as the industry looks for permanent solutions to reduce AI’s reliance on fossil fuels.
Can AI contribute to reducing the CO2 footprint?
While some fear the environmental impact of AI, the majority believe that the benefits of this technology will outweigh the risks in combating the climate crisis. With the rise of energy-efficient algorithms and groundbreaking cooling systems, AI itself can be adapted as a powerful ally in the fight against climate change.
The technology is already starting to redesign and optimize sustainability programs, such as the operation of wind turbines, solar panels, electric vehicles or batteries. Thus, AI could contribute to more energy efficiency and eventually reduce overall consumption by minimizing unnecessary processes.
For example, most organizations (57%) believe digital twin technology is critical to improving sustainability efforts. Digital twins are virtual representations of a tangible process, informed by real-time business data to predict the outcomes of proposed changes. The technology can reduce the carbon footprint of an office building alone by up to 50%.
There are many variations of digital twins, and depending on the type, it may cost as little as $65,000 to over $1 million to deploy a twin. The energy, materials and mobility sectors are the three highest carbon emitters, yet they also have the greatest potential to reduce emissions by leveraging digital process improvement technologies.
Another path to significantly reducing energy consumption is by using specialized or purpose-built AI such as small language models (SLMs). Advanced artificial intelligence platforms such as generative AI and large language models (LLMs) have fallen under scrutiny for their voracious energy usage, stemming from the massive stores of data that must be navigated to yield results.
To mitigate this, enterprises have begun to pivot to purpose-built AI specialized for narrower tasks and goals. These contextual AI and machine learning solutions are tailored to enable straight-through processing with high accuracy in real-world scenarios.
Hugging Face and Nvidia, in partnership with Mistral AI and OpenAI, have each launched small language models (SLMs) to democratize access to advanced natural language processing, which could have a significant impact on the environment.
The Potential of AI to Optimize Data Centers
Regardless of environmental concerns, investment in AI is booming and could rise to $200 billion by 2025. In general, data center operators try to keep energy consumption at a constant level in order to control costs while ensuring reliable performance. However, they remain dependent on the electricity grids that supply them as well as renewable energy sources that are subject to fluctuations. If there is a lack of storage capacities such as pumped storage power plants, “dark doldrums“ can currently only be compensated for with fossil energy.
AI can overcome these challenges by optimizing energy use. For example, it could predict the availability of solar energy by using weather data and predictive analysis. In this way, data centers could shift workloads according to the peaks of renewable energy generation, thus reducing their dependence on fossil fuels. AI could also increase efficiency by monitoring data in real-time to reduce energy consumption without compromising performance.
Also Read: How we built GT Protocol’s AI Assistant
The future of AI will be determined by regulation
AI is still developing rapidly, and in many cases is subject to only a very limited or even unclear legal framework. The industry still needs clear legislation that encourages companies to drive innovation while maintaining best environmental practices. In other words, it is possible to make further progress, but it does not come at the expense of other important considerations.
To address these challenges, the Organization for Economic Co-operation and Development (OECD) has defined a set of principles to guide the implementation of trustworthy AI for the benefit of humanity. However, better defined national, regional and international frameworks are needed for energy consumption, especially in regard to role of the energy sector in the global economy and its importance for achieving crucial climate goals.
Unfortunately, there has been a lack of ecological focus from recent legislation like the EU AI act and President Biden’s executive order, which concentrate mostly on other facets of AI responsibility such as ethics and privacy. While this is laudable, these regulations need to be expanded to include obligations for companies to reduce their AI-related emissions and energy consumption.
However, it is up to business leaders themselves to take a step back from the technology’s potential for value and look inside their organization for more ways to sustainably leverage AI.
Some major AI players such as Microsoft have begun to proactively self-regulate and work towards sustainable AI use, which is to be commended for consciousness and accountability.
Companies bold enough to tackle this challenge not only stand to reap significant economic benefits, but also set a new bar for sustainable innovation.
Comments are closed.