解决 AI 数据中心可持续性挑战，是否找错了方向？

Smarter supply chains could hold the key to reducing the environmental impact of the AI boom

Sustainability leaders across the globe are debating the paradox of AI. On one side, AI’s environmental footprint is growing. But on the other side, AI is actively helping solve some of the most challenging real-world sustainability problems.

AI’s benefits aren’t just in increasing productivity within a business or automating processes; they’re much bigger and more consequential. When AI is applied to scientific research, healthcare, and energy systems, it has the power and potential to save lives. When it is applied to sustainability challenges, AI’s predictive and optimization capabilities can reduce waste, energy use, and emissions. The International Energy Agency (IEA) estimates that if existing AI applications for industrial processes were adopted industry-wide, the energy savings alone would exceed Mexico’s total current energy consumption.

While we can’t turn a blind eye to the current environmental footprint of AI and the data center ecosystem, we also can’t ignore the meaningful impact and promise the technology offers. Abandoning AI now will inhibit us from realizing its full benefits. However, to deliver on its promise, we must build out the physical infrastructure that supports it, and we must look for ways to do so sustainably.

So far, the sustainability conversation around AI has largely focused on data centers. But to meaningfully reduce the environmental footprint of AI, we need to also look beyond the walls of the data center and follow the path of the materials, equipment, and hardware upstream.

One powerful lever for reducing AI’s footprint lies in optimizing the global supply chains behind the AI infrastructure.

The current focus on data centers in discussions around AI’s environmental impact is, in many ways, understandable. Today, AI workloads make up about 20% of data center usage. That is predicted to double to 40% by 2030. AI requires enormous computing capacity, and that capacity requires vast amounts of electricity to run and water to cool the systems. The IEA reports that a typical AI-focused data center can consume as much electricity as 100,000 households.

Some of the greatest minds in sustainability, design, and infrastructure are pushing for wide-scale adoption of renewable energy to power data centers, green building practices in every facility and low-water cooling innovations. These issues are critical, but they tackle only part of the problem. The changes should start with AI-powered supply chain optimizations.

AI’s environmental footprint begins well before a data center starts operating. The data center represents the endpoint of thousands of sprawling supply chains producing and moving semiconductors, servers, networking equipment, transformers, switchgear, cooling systems, steel, and concrete. These products’ supply chains cover multiple suppliers across global networks.

This system is largely left out of AI sustainability conversations. But these supply chains are a significant part of AI’s overall footprint and should be optimized to reduce waste and emissions.

A 2026 review of 112 peer-reviewed studies, published in Cogent Engineering, found that AI can have a transformative role in optimizing supply chains by improving supply chain processes, information sharing, and transparency. Here’s how AI-powered supply chain optimization mitigates the environmental impact of AI:

More efficient manufacturing: When applied to manufacturing, AI can help reduce waste by optimizing energy use, identifying equipment problems causing sustainability issues, and surfacing inefficient processes. Within the AI infrastructure supply chain, these efficiencies are especially critical, since the products being produced (including servers, cooling systems, electrical equipment, semiconductors) are particularly resource- and energy-intensive to manufacture, often requiring critical-mineral extraction, according to BSR (Business for Social Responsibility). Maintaining the computing equipment requires frequently replacing servers and AI hardware every two to three years. This makes optimizing upstream product manufacturing one of the biggest opportunities for reducing the overall environmental impact of data centers.

Improved orchestration to move equipment with fewer wasted resources: After the hardware is manufactured, it’s time to transport it to the data center build site. AI can help reduce unnecessary mileage and resource use by supporting load building, transportation planning, and route optimization. The U.S. Environmental Protection Agency confirmed that this is a core part of improving supply chain sustainability.

Better visibility and scenario planning: With AI improvements to visibility, forecasting, and scenario planning, companies don’t have to stockpile excess inventory or duplicate capacity to prepare for potential disruption. A more precise understanding of where risks lie enables them to develop a targeted response rather than building buffers that create waste.

Networked, streamlined collaboration: With AI embedded throughout their supply chain networks, manufacturers and other partners can make communication and information-sharing as efficient as possible. Real-time collaboration means everyone gets the same update at the same time. No more wasted effort, time, or resources.

Lifecycle visibility extends the usability of equipment: As technology advances rapidly, e-waste is becoming a growing concern. According to the Global E-waste Monitor, the world generated a record 62 million metric tons of electronic waste in 2022, doubling from 2010. AI-enhanced lifecycle visibility can help significantly reduce e-waste by informing decisions about repair, refurbishment, and recycling as data-center hardware is upgraded in the future.

This is key to the circular economy movement, which has enormous potential to make supply chains more sustainable. Within the circular economy, AI can improve material sorting to increase recycling rates, optimize reverse logistics to collect returns for refurbishing, and predictively match recovered materials with demand.

By looking closely at each part of the AI infrastructure supply chain and optimizing operations to be more efficient and sustainable, businesses can significantly reduce AI’s footprint. AI itself is changing what has historically been possible, allowing planners to simultaneously analyze the myriad complex variables of the data center supply chain to make the best possible decisions both for the bottom line and for the environment.

Despite concerns about data centers’ resource use, sustainability leaders remain generally optimistic about AI’s impact— the World Economic Forum’s Chief Sustainability Officers Outlook found that 73% of CSOs expect AI to meaningfully accelerate sustainability progress.

Additionally, AI promises to drive material and semiconductor innovations that recursively improve the technology and make it less resource-intensive. Researchers and engineers are using AI to design more efficient chips, develop more energy-efficient AI models, discover new materials, and improve cooling technologies and data center energy systems.

These innovative solutions, alongside widespread supply chain optimizations, will make the next generation of data centers more efficient and sustainable, shrinking AI’s environmental footprint considerably.

The AI transformation and the push for sustainability do not have to work against each other. By applying this rapidly evolving technology to the data centers and supply chains powering its growth, we can continue unlocking AI’s benefits while building a more efficient foundation for future infrastructure.

Saskia van Gendt serves as the Chief Sustainability Officer in Blue Yonder. In this role, she is responsible for developing and executing Blue Yonder’s sustainability strategy and driving sustainability initiatives in product roadmaps and across the company.

Discover More

mvp-article-cont