AI is having a huge impact on pretty much every part of the economy—and the environment. Researchers expect the AI industry to consume at least 10 times the amount of electricity it demanded in 2023 by 2026, which is leading to big questions about the technology, energy-related emissions, and climate change.
In looking for ways to decrease that potentially massive carbon footprint, experts have championed solutions like optimizing workflow, using sustainable hygiene (like not running things the enterprise isn’t actively using), and even using AI itself to monitor emissions.
Organizations like SustainableIT.org, a nonprofit spearheaded by tech execs, are aiming to set standards in the IT industry for best practices and provide training in order to reduce AI’s environmental impact.
Individual companies, like the multicloud and AI solutions company Rackspace Technology, have set sustainability goals. Rackspace aims to become 90% net-zero in carbon emission by 2045, with 50% net-zero emissions by 2032.
Ben Blanquera, VP of tech and sustainability at Rackspace, told IT Brew sustainability actually centers on optimizing teams’ workload. He suggested fundamental hygiene for ensuring sustainability and reducing environmental impact includes asking if the programs running are actually needed, because that translates into less energy usage
“When you talk about the impact of AI, you can get caught up by that, but you actually have to start with the basics of architecting what you need from a workload, minimizing the starting point of the resources you’re using, and that’ll get you 80–90% of the way there,” Blanquera said.
Brigit Hirsch, the Environmental Protection Agency’s press secretary, provided IT Brew with a statement saying the agency recommends data center owners and operators,consult with state and local air permitting authorities for information about the Clean Air Act requirements.
The EPA “regulates air emissions from different types of sources, which could include data centers” under the Clean Air Act.
Sustainability, sustainability, sustainability. Right now, too much of the IT market is not looking at sustainability in the long term, according to Forrester Senior Analyst Abhijit Sunil. He said that many organizations are considering the short-term gains of emerging technologies like AI while not always considering “negative impacts that might catch up” with them.
“There will be a scramble for green energy, the power purchase agreements, all those different types of things,” Sunil said. “As a framework for IT leaders to look at these technologies, it will be really specialized in terms of what they’re looking at.”
Top insights for IT pros
From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.
Forrester released a report in July detailing how technology leaders can use AI to guide sustainability—along with AI’s need for computing power, materials, and scale that could offset the benefits of innovation.
The study showed that advanced data centers can use sustainability technologies like advanced cooling and heat reuse to help save energy and boost efficiency. While the efficiency benefits can offset the savings through encouraging higher demand, Forrester recommended that data center operators choose modular and durable infrastructure equipment to encourage a sustainable IT life cycle.
Sunil said, “Even as an IT leader, if they’re taking a decision on their data center expansion or things like that, they can leverage some of these analytics solutions that will help them make better decisions, that will enable them to have better business resiliency.”
Using AI to manage your AI. One choice for an IT leader to make is how to leverage an AI solution internally to help make better decisions.
According to Sunil, IT professionals can use models like GenAI to assist with measuring and reporting emissions through their implementation in a climate risk analytics software solution.
“AI can help with not just the assessment of various risks and making it easily accessible for various leaders within [a] company, but also to be able to make the data more accurate, the analysis easier for multiple personas within the organization to read and use,” Sunil said.
Rackspace has implemented an AI process that takes invoices from facilities around the globe and puts the data into an inventory of usage metrics. The technology pulls out the information from the invoice and categorizes it.
Blanquera explained this solution arose from a need to have complete and accurate data.
“We know that’s not sexy, or pretty, but it works,” he said. “We have put in our basics to make sure you can run with AI down the line. Without the basics, all those fancy algorithms, AI really don’t work.”