Artificial intelligence (AI) and machine learning are rapidly changing the data center industry. AI tools and technology already are immensely popular in the United States, and it’s anticipated that this adoption will be globally widespread. Think of how global internet access fundamentally changed society (and our critical infrastructure) just a few decades ago; AI is on track to disrupt virtually every industry at a similar scale – are you ready? Hyperscalers, developers and investors need to plan strategically for the digital evolution ahead by considering the following major ways that AI and machine learning are transforming data centers:
With AI becoming more prevalent in today’s digital world, data centers must be upgraded to keep up. To support increased accuracy of AI and machine learning, exponential growth in data computing is needed. High computing capabilities always have been required of data centers, but not as consistently as demanded today. Servers are consistently operating at much higher loads, closer to their peak design limits than ever before.
Operating this close to capacity (and potentially failure) creates risks that need to be addressed, calling for significant upgrades to older systems and facilities. Consistently operating at max capacity leads to higher temperatures and higher voltages; cooling and electrical system performance and design basis should be continually evaluated and most likely augmented. High-density racks must be installed, requiring additional power and water integration to the facility.
Data centers are one of the most power-intensive building types, consuming up to 50 times the power of a typical commercial office building and accounting for about 2% of the United States' total power use.1 Previously, data center server racks were designed with 5 to 10KW capacities; the industry is now seeing a trend towards 35 to 100KW per rack, and full data center campuses are being developed. These infrastructure modernizations are leading to a substantial increase in power consumption, further integrated water re-use for cooling systems, and demand for specialized equipment.
Data center owners and operators have an extremely limited time to react to power disruptions; backup power is critical to resiliency, and outages are unacceptable. If any type of power failure were to occur, thermal battery systems and cooling systems must be run on facilities’ uninterruptible power supply (UPS) systems. Consider implementing the following resiliency strategies:
Renewables and Microgrid Integration. While utilities have adequate generation capacity, their distribution infrastructure may not be capable of carrying the amount of energy when and where your facility needs it. Microgrids serve as control centers to seamlessly transition to different types of power generation — such as renewables, combined heat and power, fuel cells or onsite generators — that allow data center owners and operators to diversify power supplies.
Integrated Water Resources. In many cases, data center owners can make operations and maintenance changes to optimize the use of onsite water sources with strategic up-front capital investments that offer cost savings down the road. For example, stormwater capture and wastewater reuse are sustainable alternatives for data center processes that do not require potable water to cool servers and other equipment.
Download Six Steps to Resiliency eBook
With the increase in cloud-based AI technology, access to “always on” power is critical. In some regions, energy demands of data centers may exceed current and future grid capacities. When data centers no longer can rely solely on the traditional power grid, they must seek creative solutions. To meet the energy demands of data centers in the near and long-term, follow these six steps towards resiliency.
Community members are influencing the future of data centers; data centers are physically growing to support AI, but this growth may negatively affect public perception. No matter which type of cooling system is being used, data centers are notoriously high water and power consumers. In areas where water supply and grid resiliency are already issues, residents and other businesses are reluctant to share with new data centers being developed. If additional transmission lines or roadways need to be constructed to support a new data center, communities may be opposed to their tax dollars funding this infrastructure. Construction and operations of data centers can be disruptive in various ways: generators and commercial HVAC systems are loud, major facilities lead to additional traffic, and multi-story data centers in urban settings raise visual impact concerns.
Data centers must comply with any federal, state, and local environmental regulations – just like any other commercial or industrial facility would. Air quality regulations already are driving big changes at these facilities; data centers often are restricted in how many emergency generators they can install. In addition, noise restrictions influence where data centers are located and noise suppression strategies that must be implemented. Although there aren’t necessarily any regulations specific to data centers (yet), there are several ways that owners and operators can prepare for stricter reporting mandates in the future, and lend sensitivity to community concerns:
Establish water and energy usage data collection processes
Optimize facility performance to increase output per megawatt hour of energy consumed
Maintain current knowledge of evolving industry benchmarks and other metrics for energy efficiency
Develop a robust community engagement plan
Consider water recharge and replenishment strategies
Harness the Benefits of AI
Data centers make AI possible for the rest of the world, but data centers themselves also can use AI as a tool to optimize performance. To truly stay ahead of the curve, data center owners and operators should view AI as a major opportunity rather than an obstacle.
Although society’s increased AI utilization has the potential to strain data center infrastructure, AI and machine learning also can be used to optimize facility operations. For example, AI-enabled predictive analytics allow facility managers to make improvements to cooling systems, loads and room configurations in real-time.
The data center industry is at an inflection point, with multiple significant changes happening simultaneously. Data center infrastructure is growing to support AI, demanding more power and water; communities and governments are calling for less disruption and more sustainable business practices. While these challenges are certainly not impossible to overcome, it takes many technical experts to make it all work. Data center owners and operators should consider collaborating with an expert advisory partner that can perform site due diligence, optimize the design and integrate supporting infrastructure. The data center industry can leverage this major inflection point to be even better stewards of natural resources and drive further technological advancements to benefit utilities, municipalities and communities.
How Did AI Answer This Question?
Just for fun, we asked a popular AI platform, “What are three ways AI is transforming data centers?” and this is what it came up with:
AI is transforming data centers in three main ways.
Firstly, it enables predictive maintenance by analyzing data from sensors and other sources to predict when a component is likely to fail. Secondly, it optimizes energy consumption by analyzing data on energy usage and adjusting settings accordingly. Lastly, AI helps identify and respond to security threats in real-time by analyzing network traffic and user behavior, allowing security teams to detect and address potential threats before they become a major issue.
Black & Veatch has more than 100 years of delivering critical infrastructure around the world. We provide integrated data center, water, power, environmental, site due diligence and facility modernization solutions to optimize the design and performance of your assets throughout the entire infrastructure lifecycle.
Contact us to discuss customized solutions for your business and stay tuned for our next article on utilizing liquid cooling for data centers to address the growing demand for AI.
