Meeting the power demands of AI-driven data centers requires innovative engineering solutions that prioritize speed to market, reliability and the right site selection. In this Q&A, Adam Benson discusses forward-looking solutions to the challenges data center developers face through custom-built, highly redundant power systems and strategic site selection.
Deploying data centers at the pace demanded by AI growth requires more than just capital. From site selection to power system design and execution, every decision must accelerate timelines without sacrificing reliability. As the world’s largest tech companies race to scale, the challenge lies in building power infrastructure that is not only fast, but also reliable and future ready.
Adam discusses the common goals and challenges in data center development and highlights the engineering and environmental challenges whose solutions will be vital for optimizing and bringing new generation data centers online.
What are the biggest challenges in powering modern data centers?
Reliability is unquestionably the biggest challenge in powering today’s data centers. While building a power plant is a well-understood process, the context of data centers and their need for effectively constant power introduces new layers of complexity.
The grid is already becoming overloaded, and if clients are pulling power for a data center from the grid or even from a dedicated plant that relies on the grid for backup, there's nothing particularly novel about that. But if you're building a plant that is solely responsible for providing power to a data center with no grid connection, or a behind-the-meter/islanded approach, then reliability becomes the central issue.
This need for reliability translates into more generators, mechanical redundancy, larger and more intricate switchyards, and complex power distribution systems. Compounding the challenge is the race to be first. Projects must move fast, often without precedent to guide them.
How is innovation shaping the future of data centers?
When it comes to innovation, it’s less about new technology and more about creative application of existing tools. The innovation lies in the scale and complexity of the whole system, just one example of which includes designing what could be the largest switchyard in the world.
Projects often demand inventive design and implementation strategies. Data center developers have unique needs, with some requiring extreme reliability, others anticipating large load swings and others needing both. That variability forces one to rethink accepted wisdom and go back to first principles. You can’t take anything for granted. We’re aiming for near-perfect uptime; we’re solving a different problem than a typical power plant, which just needs to provide power most of the time.
In terms of technological innovation, there is likely a potential future role for nuclear power, either in using existing nuclear plants or future market solutions that could include small modular reactors or nuclear fusion. Advanced control systems will also play a role in optimizing design solutions in the future. Plants will need to do things they have never been designed to do before.
How do data centers achieve near-perfect power reliability?
Data center developers and operators require redundancy for high reliability, with some needing up to 99.999% reliability of power supply to ensure no outages disrupt their operations and services. Typical power plants provide 88-90% availability. Therefore, redundancy is required to meet the high reliability needs of data centers. Continuous and stable energy supply is crucial to avoid significant financial repercussions from outages, in the form of liquidated damages owed by the power plant owner.
To meet the stringent reliability requirements of data centers, the system must include sufficient redundancy. This means having additional units on standby, ready to ramp up instantly if a unit fails. In some cases, bridging technologies may also be necessary to ensure seamless transitions and maintain continuous power delivery. To help a client forecast behind-the-meter power requirements, we developed a high-level reliability simulation tool that evaluates the number of proposed generators, incorporating assumptions on scheduled maintenance and forced outages. Using a Monte Carlo Analysis, it assesses the likelihood of one or more generating units failing simultaneously, taking planned outages into account, and can be used to estimate reliability associated with a proposed configuration. Conversely, the tool can be used to find a configuration that meets a particular reliability requirement.
What factors are critical when selecting a site?
Environmental considerations play a critical role in data center development, particularly when it comes to site selection and power infrastructure planning. Effective site analysis involves evaluating proximity to key infrastructure such as natural gas and water sources, transmission lines and fiber optic infrastructure, while also accounting for environmental constraints, regulatory requirements and land availability. Water availability determines the cooling options that are available to the data center and thus is an important aspect in overall system efficiency. Factors like noise, heat generation, and air quality must be considered to ensure regulatory compliance and minimize community impact. Permitting processes are supported by detailed matrices and schedules that help guide decision-making. Additionally, environmental teams often work closely with engineering counterparts to evaluate power generation options—including natural gas, hydrogen, battery storage, diesel and solar—to ensure reliable and sustainable energy supply. Collaboration between environmental and engineering teams streamline this process, enabling faster delivery of site assessments and development strategies.
In one case, we worked with a client to evaluate sites across multiple states for a new power plant to support a hyperscale data center. Site evaluations were guided by several strategic factors, including proximity to a prominent CO₂ pipeline, availability of sufficient acreage, avoidance of non-attainment air quality zones, access to natural gas and water infrastructure, wetlands and other remediation zones, and the feasibility of transmission interconnection. Based on these considerations, the recommended approach is to select the site that best aligns with both strategic and logistical priorities—particularly those related to environmental compliance and infrastructure access.
How are developers balancing speed with sustainability?
Speed to deployment has always been important, but today it is a top priority with the tech giants all vying for AI supremacy.
Some data center developers are opting for behind-the-meter power plants for speed to market, but tying into the existing grid enhances reliability. We work with major original equipment manufacturer (OEMs) to tailor generating options that are ideal for the specific needs of each data center, ensuring both speed and reliability.
Developing strategic partnerships early on with environmental groups is crucial to fully understand land acquisition needs and minimize community impacts. Involving engineering consultants early in the process helps address reliability and meet the total megawatt requirements of data centers.