Supply chain constraints are limiting revenue growth among the largest U.S. data center developers—known as hyperscalers—and threatening the United States’ progress in artificial intelligence innovation, according to recent company earnings calls. A limited supply of chips, electricity and land underlies these challenges. The data center ecosystem needs to figure out how to scale and be flexible as they work through constraints.
The U.S. is the global leader in data centers with 45% of all data centers globally by count, but these emerging constraints make maintaining that lead far from certain.
Along with data centers’ longtime real estate and infrastructure equipment needs, electricity and water are becoming key inputs to enable intensifying growth. This new age of rapid data center demand requires acknowledging and growing the whole ecosystem, with opportunity for all involved.
Recent earnings calls from multiple hyperscalers specifically called out the availability of technology (chips and motherboards), electricity and land. Below we take a closer look at those input constraints.
Technology
Semiconductor chips and motherboards are key to the scalability and processing power of data centers. However, their manufacturing is highly concentrated and highly segmented, with a few companies dominating the design, fabrication and essential equipment production. High-performance chips, though expensive, are required in large quantities to meet data center redundancy needs, resulting in significant capital investment.
Since 2020, global shifts in the geopolitical landscape increasingly favor more nationalistic production models, presenting opportunities for technology equipment companies. In response to supply chain vulnerabilities and the strategic importance of semiconductor independence, the United States introduced the CHIPS and Science Act in 2022, allocating $280 billion in funding to bolster domestic production, enhance chip performance (both in processing power and energy efficiency), and encourage the development of alternative designs.
Despite this substantial investment, building semiconductor manufacturing capacity has long lead times. Most fabrication plants funded by the CHIPS Act are not expected to be fully operational until 2028 or 2029. Given the exponential growth of data centers driven by the AI boom, the broader technology ecosystem must scale accordingly and implement efficiency measures to sustain demand. Further breakthroughs and commercialization in quantum computing would only increase demand for chips and technology infrastructure.
Electricity and water
Electricity is a critical dependency for data centers, and securing enough electricity is an increasing problem as data centers are rapidly scaling up. For example, PJM, the grid operator responsible for powering “data center alley” in northern Virginia, projects a potential capacity shortage as early as 2026. With this shortage, builders of data centers in areas that are capacity-constrained are faced with hard choices between speed to build, reliability, and carbon footprint.
Other potential energy sources for hyperscalers include:
- Solar/wind energy, which are emission-free but not yet reliable without significant battery expansion to sustain power when the sun or wind isn’t available.
- New nuclear plants, which are reliable and do not emit greenhouse gases, but the last nuclear plant (Vogtle 3 and 4) took over 15 years to build. Small modular reactors have promise for shorter timelines but are unlikely to be commercially viable until the early 2030s.
- Natural gas plants, which are reliable but moderately carbon-intensive (more than renewables or nuclear but less than coal) and can take approximately four to five years to come online.
Some data centers are also looking at direct (“behind the meter”) connections to power plants to get faster access to power and/or improve project economics, but they still must connect to the grid for resiliency against outages of their main power source.
Data centers are also adopting water-based cooling of chips as server and chip density has increased, rendering traditional air conditioning less effective. As a result, local water utilities serving large data centers are having to invest millions of dollars of capital to increase their water-handling capacity. Some local governments are including funding for such upgrades in deals to attract large data centers. Water demand is also creating opportunities for companies that provide solutions for more efficient water consumption or processing.
Real estate
Also key to growth in data centers is the availability of suitable land for facilities. Large U.S. data centers have historically been in core markets such as Northern Virginia, Oregon, Phoenix and Dallas/Fort Worth, with proximity to major internet exchanges, enough electricity and water, limited exposure to environmental risks/natural disasters, and available labor.
As these core locations become saturated, data center developers are looking to alternative locations that still meet key requirements but with more land availability and lower land and operations costs. Meanwhile, the state regulatory landscape is evolving with legislators in several states planning and proposing bills aimed at ensuring data centers pay their fair share of energy bills and in some cases setting renewable energy use goals for data center customers.
Takeaways
As data center providers and AI leaders seek to quickly build new facilities to increase capacity, it is apparent that their destiny relies on the growth and success of the entire ecosystem. Strategic planning, innovation and supply chain resilience will be crucial in meeting the growing demand and preparing for the next wave. Organizations should consider the following key actions:
- Evaluating their ability (or constraints) to rapidly scale their business to meet growing demand from data centers and identify opportunities to increase capacity, while navigating the higher-for-longer interest rate environment and higher cost of capital.
- Organizations whose supply chains overlap with data centers must review their critical supply needs (products, services, and partners) and consider alternatives in anticipation of less availability and higher cost as demand increases.
- Identifying government incentives and programs (e.g. CHIPS Act) from the federal, state, or local government that organizations can use to obtain funding or other benefits that can improve project economics.
