Data Centers
Sections: Overview Project activity Key considerations
Data Centers
Sections: Overview Project activity Key considerations
As the North American data center sector undergoes rapid transformation, hyperscalers and developers are reevaluating their strategies to meet the rising demands for power, cooling, and specialized hardware driven by next-generation AI workloads.
Michelle Rayburn Director, Americas
Overview
The US is set to become the fastest-growing market for data centers, with demand expected to rise from 25 GW in 2024 to over 80 GW by 2030.¹
This growth is fueled by an increase in data volume, computational power, and connectivity, driven by digital transformation, cloud adoption, and emerging technologies like artificial intelligence (AI). In response to rising infrastructure demands, capital expenditure for data centers reached a record US$31.5bn in 2024, largely due to the AI boom.² Notably, the capex-to-revenue ratio for major hyperscalers increased from the historical average of 10% to 12%, highlighting the urgent need to expand.³ New AI models, such as DeepSeek, are making AI more affordable, leading to uncertainty about the future of AI investments. While lower costs could lead to more widespread adoption of AI, the market is still in flux, and its long-term direction is not yet clear.
Some AI data center projects in the US have recently been put on hold, indicating a strategic shift in capital investments to meet changing demands. While AI is gaining momentum, non-AI applications still account for the majority of data center demand, expected to be 55% by 2028 with a steady 7% average annual growth rate from 2023.⁴ GenAI is the fastest-growing segment, with training workloads projected to grow at a 30% average annual growth rate and a 122% increase in demand for inference between 2023 and 2028.⁴
The data center sector is quickly adapting to changes in demand and supply dynamics. With the US government investing in AI infrastructure, there is increasing pressure due to low vacancy rates coupled with rising demand for cloud computing. As demand continues to grow, power needs will also increase significantly, which could hinder future expansion. To tackle this issue, the industry is looking into sustainable energy solutions like small modular reactors (SMRs) and carbon capture technologies, ensuring that growth is both environmentally friendly and manageable. At the same time, there is a rise in natural gas solutions, as a means of diversification, supported by strategic partnerships with utility providers.
Tariffs present a major challenge for the data center industry. While semiconductors are currently exempt, rising costs for power and cooling systems remain an issue. These systems rely on imported components that are affected by tariffs. Although a 90-day pause on reciprocal tariffs offers some relief, high tariffs on China still hinder its role as a key supplier of essential equipment. This situation is likely to create supply chain instability and uncertainty for critical infrastructure projects, leading to delays in data center expansions. To counter these challenges, data center developers may consider diversifying their supply chains.
AI related investments
One of the most ambitious projects driving growth in the data center sector is the Stargate Initiative, which plans to invest approximately US$500bn over the next four years to develop advanced AI infrastructure. The initiative will start with a US$100bn investment to build large-scale 5 GW data centers and feature 30m ft² of facilities equipped with 2m GPUs.⁵ ⁶
Power and sustainability
As demand for data processing grows, S&P Global Ratings forecasts that US data centers will need an additional 150-250 terawatt-hours (TWh) of electricity annually by 2030.⁷ Meeting this demand will be challenging, primarily because of the need to upgrade the grid infrastructure. This could lead to higher electricity prices, a shift from the previous two decades of stable demand. This increased need for power will also drive up demand for natural gas, with an expected rise of 3 to 6bn ft³ by 2030.⁸
To address these challenges, some data center operators are adopting a "Bring Your Own Power" (BYOP) strategy to secure long-term power purchase agreements with renewable energy providers, including geothermal, nuclear, solar, and wind. The industry is also exploring new technologies like small modular nuclear reactors (SMRs) to ensure a stable power supply. Power availability is a risk, especially in competitive regions like Alberta and Ontario. The Canadian power market is generally stable, but timing and planning remain crucial.
In terms of sustainability, carbon capture and storage (CCS) technology is gaining popularity in the US. Several energy providers are planning large-scale power generation projects that integrate CCS to capture and store a significant portion of CO₂ emissions. According to ExxonMobil, data centers could represent up to 20% of the market for CCS by 2050.⁹ Digital twin technology can enhance energy efficiency in data centers by simulating operations to optimize energy usage. A recent development in the US demonstrates how digital twins can be used to improve the efficiency and management of data centers by creating virtual replicas that simulate mechanical, thermal, and electrical systems, enabling real-time management and optimization.
Evolving Data Center site selection
Moving beyond latency concerns: While low latency is crucial for sectors like high-frequency trading and gaming, AI applications are not as sensitive to it. This shift allows AI data centers to be built in more affordable, less populated areas, significantly reducing operational costs. For example, Utah is becoming a popular location due to its cooler climate and affordable energy costs. Shifting to industrial areas and renewables: Data centers are increasingly being built in industrial zones or remote locations to avoid the energy limitations found in urban centers. These sites are often near renewable energy sources, which helps alleviate the strain on energy infrastructure and meet sustainability goals. Repurposing existing facilities: There is a growing trend of converting industrial buildings and warehouses into data centers. This approach addresses the need for flexibility and offers the opportunity to use existing infrastructure. Hybrid models, which combine new constructions with repurposed sites, are becoming more common, and companies are collaborating with developers early to meet their specific needs.
Procurement and contracting strategies
The traditional delivery model approach of managing a project through a general contractor or construction manager is evolving. As the industry continues to seek ways to overcome labor and talent acquisition challenges, further complicated by rising data center construction demands, an alternative for data center delivery is project management consultancy (PMC). With this approach, the client owns the contract where multiple vendors may be involved, but the PMC team eliminates the need to manage changes through a general contractor, saving time and money.
Companies are increasingly using AI tools to speed up procurement processes and address material shortages in supply chain management. While traditional models like Guaranteed Maximum Price (GMP) and Design-Build are still popular, there is a noticeable shift toward build-to-spec models for more tailored solutions.
Project activity
Data center development is booming across the US, with key regions leading the way. Northern Virginia stands out, accounting for about 25% of the country’s data center capacity.¹⁰ Other notable locations include Silicon Valley, Dallas, Chicago, and Phoenix, all experiencing significant growth.
Northern Virginia remains particularly active, especially along the I-95 corridor, and significant growth is expected in Central Virginia. Arizona is another hotspot, with Mesa being the most active city while Goodyear and Phoenix are also seeing steady progress on various projects in the planning and execution stages.
California shows a diverse range of project stages, from early planning to advanced execution. Santa Clara is a major hub, showcasing a mature development pipeline with a higher concentration of ongoing projects.
West Texas is emerging as an important player due to its low electricity prices, driven by its access to renewable energy and natural gas resources. This has attracted interest from the data center sector and supported growth in cryptocurrency mining, which is now being adapted for AI and high-performance computing, thanks to the region's access to power and cooling infrastructure. Notably, Texas has been selected as the initial location for Project Stargate.
The US Department of Energy has identified 16 federal sites, including prominent national laboratories, to develop data centers aimed at advancing AI. Leveraging existing energy infrastructures at these locations will help speed up construction and meet the increasing computational needs of AI technologies.¹¹
The Canadian Market remains significantly busy. Alberta and Ontario are emerging as hotspots for data center development, while Quebec has seen limited activity due to government restrictions on new power procurement.
However, recent trade policies, including high tariffs on imported tech equipment from countries like China are affecting data center investment strategies. These tariffs may increase costs and delay expansion projects, leading companies to reassess their infrastructure plans.
Key considerations
01
Labor issues
Managing complex data center builds is challenging due to a shortage of experienced professionals, which impacts project execution. Concerns about the quality of workers are rising as the market shifts from residential to commercial and data center construction. In the US, recruiting for remote sites remains a significant hurdle, and retaining skilled talent will be difficult in the coming years. Stricter immigration policies also pose challenges, as about 18.6% of construction workers are foreign-born.¹² A reduced workforce could delay data center projects and lead to higher competition for skilled labor, increasing costs.
02
Material shortage
Lead times for key materials, particularly electrical (like transformers and switchgears) and mechanical equipment, are a major concern, with delays of 2-3 years in the US and 3-4 years in Canada. Tariffs on certain imports may complicate procurement, and contractors are starting to include tariff clauses in contracts to manage related risks.
03
Contractors availability
In high-demand markets like Ohio, Utah, and Louisiana, securing quality contractors is challenging due to limited availability and strained resources. This issue varies significantly by region.
04
Impact of tariffs
AI data center projects could face delays due to new trade policies and tariffs that increase costs for highly imported critical materials such as steel. This may affect project costs and timelines. Some tech imports, including semiconductors, are exempt from tariffs. This exemption is significant because semiconductors power nearly all key components in data centers, from servers to networking equipment. Their availability at stable prices can ensure uninterrupted development and deployment of AI systems. However, it’s unclear if all chip suppliers will be protected. Some hyperscalers are scaling back or rethinking their projects, which could lead to an 8-12 month delay in data center planning and construction.
Sources
- How data centers and the energy sector can sate AI’s hunger for power, McKinsey & Company, September 17, 2024
- 2025 United States Data Center Market Outlook, Newmark
- Accelerated Growth of Digital Services Driving Hyperscale Operators to New Highs, Synergy Research Group, April 10, 2025
- Breaking Barriers to Data Center Growth, Boston Consulting Group, January 20, 2025
- Infrastructure is Destiny Economic Returns on US Investment in Democratic AI, Open AI, September 2024
- Tech giants are putting $500bn into 'Stargate' to build up AI in US, BBC, January 22, 2025
- Data Centers: Can Infrastructure Developments Keep Up With The Increasing Demand, S&P Global Ratings, Dec 04, 2024
- Data Centers: Surging Demand Will Benefit And Test The U.S. Power Sector, S&P Global Ratings, October 22, 2024
- Steel, ammonia and AI? Oh my! What can’t our CCS help decarbonize?, Exxon Mobil Corporation, December 11, 2024
- Data Centers in Virginia, Joint Legislative Audit and Review Commission
- DOE Identifies 16 Federal Sites Across the Country for Data Center and AI Infrastructure Development, US Department of Energy, April 03, 2025
- Foreign-Born Workers: Labor Force Characteristics - 2023, Bureau of Labor Statistics
Disclaimer
This report contains information, data, and analysis related to the construction industry. While we strive to provide accurate and up-to-date information, it is important to note that the market is subject to various factors, uncertainties, and changes that may impact the accuracy or reliability of the report's contents. By using the report, you acknowledge and agree that Linesight assumes no responsibility or liability for any inaccuracies, errors, omissions, or losses that may arise from your reliance on the information presented in the report.
It is essential that users exercise their independent judgment, conduct their own research, and seek professional advice before making any decisions based on the information contained within the report.
Any user of this report, by their acceptance or use of this report, releases Linesight, its parent corporation, and its and their affiliates from any liability for direct, indirect, consequential or special loss or damage whether arising in contract, warranty, express or implied, tort or otherwise, and irrespective of fault, negligence and strict liability.
No section or element of this document produced by Linesight may be removed from this document, reproduced, electronically stored or transmitted in any form by parties other than those for whom the document has been prepared without the written permission of Linesight.
© Linesight
© Linesight