As artificial intelligence (AI) reshapes industries, tech companies are racing to build greener data centers that slash energy costs without sacrificing performance.
Firms like Sustainable Metal Cloud (SMC) are at the forefront of this shift, developing cooling solutions for data centers that are trying to cut energy while boosting computing power. These advancements come as the tech industry grapples with surging demand for AI capabilities and mounting pressure to reduce its environmental footprint.
“Between exponential cloud growth and changes associated with AI, we are facing a critical challenge: a power problem and an energy trilemma,” John DeBoer, head of data center vertical market at Siemens Smart Infrastructure US, told PYMNTS. “The energy trilemma is the balancing between energy security, energy sustainability and energy affordability.”
More efficient data centers could significantly reduce energy consumption and operational costs for businesses relying on digital infrastructure. This increased efficiency could lead to lower prices for cloud services and online transactions, fostering growth in eCommerce and digital-based industries.
SMC’s new HyperCubes — containerized Nvidia GPU servers utilizing immersion cooling — are attracting attention from tech industry leaders. The company said that these systems can reduce energy consumption by 50% compared to traditional air-cooled systems while also being 28% more cost-effective than rival liquid-cooling solutions.
The growing importance of energy innovations like Hypercubes is highlighted by Aditi Godbole, a senior data scientist at software company SAP. She told PYMNTS that recent years have seen a substantial rise in the deployment of GPU and high-performance computing systems to power AI and machine learning (ML) applications.
“These workloads consume a lot of power, energy, cooling, network capacity and resources, and also require high-density underlying infrastructure,” Godbole said. “Adopting sustainable AI data centers could offer significant cost savings for businesses and provide efficient and intelligent data center infrastructure management solutions, specifically in terms of energy consumption, operational expenditure, and network resources.”
The technology has far-reaching commercial implications. As AI applications proliferate across industries — from healthcare and finance to autonomous vehicles and smart cities — the demand for powerful, efficient computing resources continues to grow. With their substantial energy requirements and carbon footprints, traditional data centers are increasingly viewed as economically and environmentally unsustainable in the face of this expanding demand.
DeBoer stressed the need to balance innovation with sustainability goals in energy projects. He pointed out several key areas of focus, including alternative power generation and improved energy storage. “Navigating this while not losing sight of sustainability KPIs matters for total cost and project success,” he told PYMNTS.
The potential for cost savings is significant. DeBoer cited a recent example: “Novva Data Centers, a leading provider of enterprise data centers, had recently employed Siemens technology to achieve zero downtime and successfully migrate from water to air cooling at its Colorado Springs facility. As a result of integrating Siemens’ technology, Novva is now saving more than two million kilowatt hours of energy per year — the equivalent of 1,397 metric tons of CO2 or enough energy to power 197 homes annually. Additionally, Novva saves more than $176,000 a year by eliminating inefficiencies.”
SMC is not alone in pursuing sustainable data center solutions.
“Companies like Google and Microsoft have utilized deep learning and neural networks to operate their data centers efficiently,” Godbole said. “Google has consistently achieved a 40% reduction in its energy consumption for cooling operations. Microsoft has implemented waterless liquid cooling designs to bring efficiency and cost savings of at least 15% for their data centers.”
Godbole outlined the challenges facing traditional data centers: “Power is a crucial factor for any data center design, and with AI & ML workloads, the chips need to train AI models that use more electricity than traditional computing. Sustainable AI data centers require redesigning of electrical and power systems to accommodate increased demand for power across data centers.”
While sustainable data centers are similar to traditional ones at their core, “best practices in sustainable design allow data centers to both serve their central compute mission while also becoming an important asset in the energy landscape with both reliable performance and an ability to respond to an ever-changing energy environment,” DeBoer added.
The financial benefits of sustainable data centers extend beyond energy savings.
“Depending on the scale of the network, these savings could range from tens of thousands to millions of dollars annually,” Godbole said. “Companies may also benefit from tax incentives or credits for utilizing green and sustainable technology, which can enhance cost savings.”
We’re always on the lookout for opportunities to partner with innovators and disruptors.
Learn More