Overview:
Jim Roche, Senior Vice President of Engineering at CyrusOne, and Kyle Myers, Vice President of Environmental Health, Safety and Sustainability at CyrusOne, recently led a Tech Talk webinar, “The Intersection of Data Center AI and Sustainability,” hosted by Virtualization & Cloud Review.
In this discussion, the panelists break down data center sustainability, the impacts data center AI has on energy consumption, strategies for energy efficiency and considerations for integrating sustainability best practices.
The panelists also address the biggest data center obstacles currently hindering the widespread adoption of AI and sustainability, how sustainability practices and designing for AI affect operational costs and long-term profitability, and advice for organizations looking to navigate the path for leveraging AI and sustainability.
While the mass adoption of cloud computing and growth of AI continues to steal the spotlight, navigating environmental sustainability as it relates to green energy, energy-efficient hardware and water conservation should be top-of-mind. The data centers of tomorrow start at the intersection of AI and sustainability today.
A special thanks to John K. Waters, editor-in-chief of the Converge360 group of 1105 Media, for serving as the moderator of the panel discussion.
Watch the Webinar:
Read the Transcript Highlights:
Environmental Impact of Data Centers
Question: Talk about the environmental impact of data centers. In what ways do data centers offset things like carbon emissions, and why has sustainability become a top-of-mind issue?
Answer: Kyle Myers explained that at CyrusOne, sustainability is considered in four areas: climate impact, water usage, biodiversity and circularity. He highlighted that the goal is not just to do less harm but also to do more good. For instance, in high-water-stress regions, we aim to return more water to the watershed than we consume.
[3:31-4:45]
Sustainability Practices, Operational Costs & the Evolution of Data Center Design
Question: How do sustainability practices and designing for AI within data centers affect operational costs and long-term profitability?
Answer: Jim Roche highlighted that sustainability practices can lead to significant cost savings by using less land, concrete and steel. By packing more into a smaller footprint, we can achieve greater efficiency and reduce carbon footprints. This approach helps to lower operational costs and improve long-term profitability while meeting sustainability goals.
[4:48-7:49]
AI Requirements Shaping Data Center Design
Question: In what ways are AI requirements shaping the architectural design of future sustainable data centers?
Answer: Kyle Myers highlighted that AI allows more flexibility in the siting of data centers, enabling them to be located where renewable energy is more cost-effective and available. This reduces the impact on biodiversity and water consumption. AI tools also help in better predicting resource availability and interruptions, making data center design more efficient.
[15:07-19:00]
Challenges of Integrating AI Technologies
Question: What would you say are the primary challenges encountered in integrating AI technologies into data centers?
Answer: Jim Roche identified power as the biggest challenge. He explained that the demand for power has drastically increased, with current data centers requiring significantly more power than before. This demand affects the siting of data centers, as they need to be located where power generation and availability can meet these high demands.
[19:01-21:16]
Regulatory Hurdles & Future Trends
Question: What regulatory hurdles are currently affecting the data center industry, and what trends do you foresee in the future?
Answer: Kyle Myers identified regulatory hurdles as a significant challenge, particularly in terms of power availability and transmission. He mentioned that there are ongoing efforts to educate policymakers about the benefits and needs of data centers. In terms of future trends, both speakers agreed that the industry will continue to see a focus on efficiency and sustainability, driven by advances in AI and a growing emphasis on environmental impact.
[21:19-27:40]
Q&A Session Insights:
Importance of PUE, CUE and WUE Metrics
Question: How important are Power Usage Effectiveness (PUE), Carbon Usage Effectiveness (CUE) and Water Usage Effectiveness (WUE) as considerations for companies outsourcing data centers?
Answer: Kyle Myers explained that PUE is a well-known metric used to measure how efficiently a data center uses energy, particularly how much energy is used by the computing equipment in contrast to cooling and other overhead. However, he noted that the metric might not be ideal for measuring overall efficiency. CUE and WUE are becoming increasingly important as they provide a more holistic view of a data center’s sustainability by considering carbon and water usage, respectively. This broader approach allows for better alignment with green financing and competitive benchmarking.
[31:10-34:54]
Collaboration with Utilities for Site Selection
Question: How are data center developers beginning to think like utilities in terms of site selection and collaboration?
Answer: Jim Roche explained that there is now more collaboration with utilities than ever before. Instead of simply building and demanding power, data centers are now selecting sites based on power availability. This proactive approach helps ensure efficient power distribution and supports the rapid growth and power needs of data centers.
[45:52-47:09]
Changes in Backup Power Architecture
Question: With the MW growing to 100 to 300 MW per data center, are you changing your architecture for backup power?
Answer: Jim Roche discussed how we are exploring ways to reduce reliance on large numbers of generators, which sit idle most of the time. We are working on innovative solutions to maintain backup power without the extensive use of traditional generators. This might involve using UPS systems more strategically and ensuring only critical infrastructure is supported by generators.
[47:10-49:12]
Higher Voltage Distribution in Data Centers
Question: Are you doing higher voltage distribution in any of your data centers?
Answer: Jim Roche confirmed that we are implementing higher voltage distribution, starting from the transmission side with transmission-grade voltage coming into the buildings. Many data centers are moving towards 415V solutions, with some considering 480V systems. This shift reduces the need for multiple voltage transformations, enhancing overall system efficiency.
[49:18-50:38]