Data center projects have been expanding rapidly across the U.S., driven by surging artificial intelligence (AI) demand, and have become a major topic of conversation. What that discussion frequently leaves out, however, are the significant risks and challenges that designers and contractors must confront and overcome to bring these structures from concept to completion. This article explores those challenges and risks designers and contractors must face when working on these projects, along with the risk management and practical contract strategies they can use to help limit their risks.

AI runs on data center projects. These sites are massive facilities packed with servers, power equipment, and cooling systems. Hyperscale data centers can cost $15-$20 million+ per megawatt (MW) to design and construct, while occupying at least 10,000 sf of physical space. A single AI-focused data center in the U.S. can use as much electricity as roughly 100,000 households. Larger campuses can require up to two gigawatts (GW) of power, which could provide sufficient energy for around 1.5 million homes. Demand is expected to grow by about 10% a year until 2030, with more than 1,400 new projects already planned or under construction and over 3,000 already operating. The U.S. government is moving to speed up permitting, including on federal land, to accelerate the buildout.

For the designers and contractors who create these facilities, this boom is a major opportunity. Yet this potential brings significant risks. Understanding those risks, and how the industry is attempting to manage them, is useful for all participants involved in the projects.

Why Data Centers Are a Different Kind of Project

The industry has seen boom-and-bust cycles before, in the dot-com era and the early cloud-computing years, with both ending in significant losses from overbuilding. Even today’s AI leaders, like OpenAI, have posted billions in losses despite selling popular products, and cheaper open-source AI models are cutting into revenue for paid services. If demand cools, some projects could become financially insolvent before they’re even finished.

Data centers require huge amounts of power and water, while facing increasing zoning fights, permitting delays, and organized opposition over noise or strain on local utilities. One proposed Missouri project was blocked entirely after a citizens’ group lobbied for a zoning change. In addition, New York has recently imposed a moratorium on qualifying environmental permit applications for large data centers while the state develops new energy use standards.

These projects can take three to five years to complete, but the computer chips and cooling systems needed for construction can become outdated in a fraction of that time. An acceptable design contemplated at the start of a project may match neither the technology available nor the customers’ needs by the time it’s finished. Each part of the design is interlinked with another, and any small issue could trigger a copious number of failures across the entire system.

Critical components like transformers and cooling systems can take over a year to obtain, straining schedules that owners want compressed rather than extended. Data center operators often promise their customers 99.99% uptime or better, backed by contracts with financial penalties for outages. If a facility goes down, the resulting lost revenue, reputational damage, and contractual penalties can significantly outweigh the building’s cost.

What This Means for Designers and Contractors

Designers and contractors on these projects face liability risks that are unusually concentrated and severe, arising from a handful of recurring issues. Requirements are often unclear at the outset or change midstream (such as increases in computing capacity, shifts in redundancy targets, or changes in the project’s purpose) after the design is already under way. Responsibility can also be fragmented across many specialized disciplines working in parallel, which makes it difficult to pinpoint accountability when failures occur. At the same time, owners frequently seek assurances that a facility will achieve specific outcomes.

Data center projects carry a wide range of external risk factors that largely fall outside the control of designers and contractors. Ambiguous or shifting owner expectations, the scale of megaprojects, and compressed schedules are constant issues. Site-level challenges such as labor availability, equipment procurement, and power or water constraints add further complexity. Financial pressures like cost overruns compound these risks.

Regulatory hurdles, environmental compliance, and political or community resistance frequently cause schedule delays, while affordability concerns can even trigger contractual exit clauses. Additional complications include tenant expectations misaligned with the owner’s original program, phased construction where parts of a facility remain operational while others are still undergoing construction, and constrained insurance or surety capacity that shifts risk onto project participants.

Underdeveloped local infrastructure and severe weather are also important risks to consider. Collectively, these external factors illustrate that risks in data center project development are shaped as much by forces beyond the control of designers and contractors as they are by the technical work itself.

How the AEC Industry Should Respond to Risks in Data Center Project Development  

Contracts should now disclaim guarantees of uptime, tier certification, or operational performance, clarifying that commissioning participation is limited to professional observation and review and that operational outcomes remain with the owner, contractor, vendors, and commissioning authority.

Implement phased-design freeze milestones aligned to procurement, lock owner project requirement assumptions at each freeze, and require a written change-control protocol for post‑freeze modifications that documents scope, schedule, and fee impacts. Value engineering and substitutions should be treated as directed decisions with express allocation of responsibility and no warranty of equal performance.

Risk allocation should address suspension, delay, and cancellation realities common to these projects. Contracts should limit exposure to data loss and cyber incidents by stating that data protection, integrity, recovery, and cybersecurity are outside the design scope unless expressly included.

Given heightened business interruption exposures, parties should incorporate mutual waivers of consequential damages and coordinate these allocations with professional liability insurance expectations on data center projects.

Mutual waiver of consequential damages essentials:

• Mutual waiver should bar recovery of consequential, incidental, special, or indirect damages arising out of the project, including loss of revenue, profits, use, business opportunity, and customers or tenants.
• The waiver should expressly include loss, corruption, compromise, or unavailability of data, as well as downtime or service interruption damages.
• Service level agreement–related remedies such as penalties, service credits, and uptime credits should be identified as waived consequential categories between the contracting parties.
• Reputational harm and missed business opportunities should be identified as waived to avoid attempts to re‑characterize business interruption losses.
• Consequential‑damage allocations should be coordinated with insurance and risk transfer programs given the severity of data center business interruption exposures.

Data centers are one of the fastest-growing, highest-stakes categories of construction today. For the designers and contractors involved, success depends not only on technical skill but also on contracts that realistically reflect what a designer and contractor can and cannot control. These contracts must protect against a market, a technology curve, and a set of expectations all moving faster than the building of any single data center project.

About the Authors
David J. Hatem is a nationally recognized attorney whose practice specializes in representing engineers and architects and advising professional liability insurance underwriters on project-specific insurance matters. David teaches Legal Aspects of Civil Engineering at Northeastern University Graduate School of Engineering and has continuously served as counsel for the American Council of Engineering Companies of Massachusetts (ACEC/MA) since 1988. He has authored and edited numerous articles and publications in his field. 

Dillon Aisenberg focuses his practice on evaluating, managing and resolving claims asserted against engineers, architects, and other design professionals in the construction industry. He brings a disciplined and strategic approach to navigating the complex liability issues that arise in design professional disputes, guiding clients through every stage of the claims process with clarity and precision. In addition, Dillon also provides comprehensive guidance and support to design professionals in the drafting, reviewing, and negotiating process for design and construction contracts, helping clients structure agreements that protect their interests and minimize exposure from the outset. 

Jack Baschwitz focuses his practice on general liability, toxic tort, consulting, design professional, and asbestos matters. Before joining MG+M, Jack worked at an Am Law 100 firm and later as a litigation paralegal at a Boston-based law firm. He joined MG+M as a summer law clerk in 2024 before returning as an associate.