Green IT, the sustainability movement that faded from corporate priority lists in the 2010s, is experiencing an unexpected revival, and AI is the reason. As organizations build out massive data centers and deploy energy-hungry AI systems, the cost of ignoring efficiency is becoming impossible to hide. Unlike older IT concerns that felt abstract or voluntary, the energy bills for AI infrastructure are concrete, measurable, and growing fast.

Why Did Green IT Suddenly Matter Again?

For years, green IT was treated as a "transient subject" that companies engaged with voluntarily, often as a public relations gesture. That changed dramatically as AI adoption accelerated. The shift isn't driven by environmental virtue signaling alone; it's driven by economics.

"AI is triggering a 'revival' of green IT discussions as concerns around the 'sustainability footprint of IT infrastructure for AI' grows among businesses," said Bjoern Stengel, Global Sustainability Research Lead at IDC.

Bjoern Stengel, Global Sustainability Research Lead at IDC

The digital sector's energy consumption now represents between 1.5% and 4% of global greenhouse gas emissions, a figure that continues climbing as AI workloads intensify. But what's making executives pay attention isn't the abstract carbon footprint; it's the concrete monthly bill. When a company deploys AI agents that run continuously, consuming electricity without pause, the cost becomes impossible to ignore.

Michael Gale, Chief Marketing Officer at sovereign AI and data company EDB, described this shift as a "radical shift" in how companies approach sustainability. "It's not the PCs you buy anymore or the lithium in that," he explained. "It's the fact your monthly bill is going to go up and you're going to see that".

What Makes AI's Energy Problem Different From Past IT Challenges?

Previous generations of IT infrastructure had built-in visibility into resource consumption. A company knew how many servers it was running, how much cooling they needed, and what the power bill would be. AI changes this equation fundamentally.

AI agents and continuous AI workloads operate differently. "The one thing about agents, they don't have a meter that says, 'Oh, I'm consuming electricity,' They just consume it," Gale noted. This invisibility creates a dangerous situation where organizations can rack up enormous energy costs without realizing it until the bill arrives.

The problem extends beyond just electricity consumption. Cooling data centers requires massive amounts of water, and the manufacturing of semiconductors carries its own significant carbon footprint. The full lifecycle of AI infrastructure, from production to operation to eventual disposal, creates environmental and financial pressures that companies can no longer ignore.

How to Evaluate and Reduce Your AI Infrastructure's Energy Footprint

• Assess Infrastructure Sustainability: Ask your data center and infrastructure providers detailed questions about their environmental practices, including carbon emissions data, cooling technologies being used, and specific energy efficiency statistics before committing to long-term contracts.
• Calculate Return on Investment Per Watt: For every watt of energy your AI systems consume, determine what value they're generating for your organization. Without this calculation, you're simply burning electricity without purpose or accountability.
• Train Citizen Developers on Efficiency: As more non-technical employees build with AI tools, establish internal guidelines and training programs to prevent wasteful practices, such as running unnecessary experiments or leaving resource-intensive processes running indefinitely.
• Monitor Consumption Continuously: Implement real-time dashboards and analytics to track AI energy usage, similar to how companies monitor cloud computing costs, so unexpected spikes can be caught and addressed immediately.

Shanea Leven, CEO of AI service Empromptu AI, emphasized the importance of cultural change within organizations. "You'll see it on Twitter and Reddit of just, 'We burned 1 million, 2 million credits doing this one task.' That should not happen," she said. "Being able to evolve with the kinds of people who are using this technology is something that I think more people besides us need to be doing".

Shanea Leven, CEO of AI service Empromptu AI

What Does This Mean for the Future of AI Deployment?

The convergence of digital transformation and environmental sustainability is no longer theoretical. The COP29 Declaration on Green Digital Action, finalized in Baku with support from the International Telecommunication Union (ITU) and over 40 stakeholders, formalizes this reality into eight common objectives aimed at accelerating climate-positive digitalization while demanding accountability from the technology sector.

Organizations that treat energy efficiency as an afterthought will face mounting costs and potential regulatory pressure. Those that build sustainability into their AI infrastructure decisions from the start will gain competitive advantages through lower operational expenses and better alignment with emerging global standards.

The green IT movement's comeback isn't driven by nostalgia or environmental idealism. It's driven by the simple fact that AI's energy hunger has become too expensive to ignore. For companies deploying AI at scale, the question is no longer whether to care about efficiency; it's how quickly they can implement it before their energy bills become unsustainable.