As demand for AI tools explodes, so does demand for the physical infrastructure to support and power these tools.
“We are witnessing an infrastructure moment related to the physical building blocks required to support the AI explosion,” says Mary Pahissa Upchurch, Executive Vice President for Customer Success at Gray Decision Intelligence.
Data centers are specialized facilities that house the essential IT infrastructure that makes AI possible. They act like a high-tech warehouse for the internet, allowing us to search the web, stream videos, bank online, and use AI. Key components include:
- Hardware: High-performance computers that process billions of parameters.
- Storage Systems: Massive arrays of drives holding the world’s data.
- Networking Gear: The routers and switches that manage the flow of information.
- Power & Cooling: UPS systems, generators, and advanced cooling systems to keep equipment running smoothly.
McKinsey & Company predicts global data center demand will nearly triple by the end of the decade, with AI workloads driving 70% of that growth. This has triggered a $5.2 trillion “capital expenditure tsunami” in physical infrastructure—a figure exceeding the GDP of many mid-sized nations.
The Resource Crunch: Electricity, Water, and Land
The rapid speed of the data center build-out has created a squeeze on resources, sometimes sparking conflict with local communities, but also driving innovation.
The Largest Consumer of Electricity by 2035?
Data centers have voracious power needs and require reliable power sources that won’t go down. In 2023, data centers accounted for 3.7% of U.S. demand; by 2030, that number is expected to hit 12%. By 2035, data centers could become the largest consumer of electricity in the country.
As data center power demands accelerate, they are increasingly competing with local communities and business districts for power. Some communities are pushing back. For example, a Georgia lawmaker recently introduced a bill to halt data center projects until March of next year, because their growth is threatening the local power supply.
Data Centers: A Thirst for Water
Data centers consume a lot of water – both directly and indirectly. AI chips run hotter than traditional hardware, requiring millions of gallons of water for cooling (direct usage). Indirectly, many data centers pull energy from power plants that use water to generate electricity. While there is uncertainty about the exact amount of water used by AI data centers and divided opinions about their long-term impact, some communities are concerned that they will infringe on already scarce water resources.
Chasing Power, Not Fiber
Historically, data centers were built near major cities to be close to fiber and internet cables, reducing latency for services like Netflix and online banking. With AI, however, the needs are different. AI training prioritizes steady power over proximity to fiber. As a result, developers are migrating to rural America to be closer to energy sources rather than fiber. Rural land is also cheaper and more readily available in the quantities needed for new “mega campuses”.
Powering Energy Innovation
Grid constraints are forcing tech giants to become energy innovators. Many companies are “rediscovering” nuclear energy. Microsoft is working to restart the Three Mile Island nuclear plant, while Amazon recently purchased a data center directly connected to a nuclear station to bypass the public grid. Meanwhile, Google is investing in Small Modular Reactors (SMRs) – miniature nuclear plants that can be manufactured in factories—to power its campuses.
Alternative energy sources are also being explored. Boom Supersonic is adapting its jet-engine technology to develop “Superpower” natural gas turbines to power AI data centers. Geothermal startup Zanskar aims to generate predictive heatmaps to identify undiscovered geothermal hotspots to provide clean power to data centers.
Community Colleges: Scaling the Data Center Workforce
What does this all mean for higher education?
The US is currently facing a shortage of talent for both the construction and operation of AI data centers.
According to research by PwC, there were over 600,000 workers in the data center industry in the US in 2023, with employment growing over 50% since 2018. This number has likely risen even more over the past two years. For perspective, this is about the same number of practicing lawyers or post office workers.
Data from Gray DI’s Job Postings dashboard shows almost 13,000 postings for jobs with “data center” in the job title over the trailing 12 months, including data center technicians, operators, supervisors, facilities managers and engineers, security personnel, and more. Job postings grew 50% year over year, and the average salary over the combined 24-month period was $98,852.
For community colleges, the AI infrastructure boom represents an opportunity to launch or expand programs that address these labor-market needs.
Data Center Construction
Building a new AI data center requires hundreds, even thousands, of construction workers and skilled tradespeople. Construction workers are already in short supply. In 2024, the Associated Builders and Contractors (ABC) estimated the construction workforce shortage at roughly 500,000 workers. To compound the problem, data center construction jobs often require more technical knowledge (i.e., cooling systems, power redundancy) than traditional commercial projects. Construction companies and large contractors have resorted to moving talent across the country (like traveling nurses during the pandemic) to complete installations.
Because of the talent shortage, employers are partnering with community colleges to develop the local talent needed to complete these projects. For example, Google’s STAR (Skilled Trades and Readiness) program provides paid 5-6 week training bootcamps to prepare students for careers in trades such as construction, carpentry, mechanical, electrical, and fiber-optics. It partnered with Ivy Tech College in Fort Wayne, IN, as part of a $2 billion data center build-out to train local students in the trades.
Data Center Operations
Once a data center is up and running, these facilities need skilled technicians to maintain 24/7 uptime. Community colleges are scaling up programs related to data center operations, again often with the help of employers. Maricopa Community Colleges offers a Data Center Operations Certificate, developed with industry partners that include QTS and EdgeCore. The certificate can be embedded within the AAS in Automated Industrial Technology. Big Bend Community College partners with Microsoft Datacenter Academy to train students for roles in data center operations. Northern Virginia Community College partners with Amazon/AWS and offers certificate and associate programs related to data center operations.
A Word of Caution: Permanent vs. Temporary Jobs
While the data center boom is real and can bring hundreds, or even thousands, of jobs to local communities during the multi-year construction phase, ongoing data center operations require far fewer people. A completed center may only need 30 to 50 permanent employees to keep the lights on. Colleges must be careful not to “strand” graduates or programmatic resources in a sector that moves at lightning speed once the initial building phase is over.
Frontier Programs: What’s Next?
In addition to career and technical education, the data center boom is creating opportunities for four-year and graduate degree programs that address digital infrastructure needs and innovations. These can include programs related to data center systems engineering, nuclear and advanced energy engineering, energy and climate strategy, and policy.
As an example, the University of Chicago offers an MS in Climate and Energy Policy that blends training in data science with an understanding of climate physics, energy technology, and market economics. It prepares graduates to evaluate the viability of new infrastructure—such as the power-hungry data centers—and design data-backed, scientifically sound policies that balance energy reliability with climate goals.
Southern Methodist University offers an M.S. in Datacenter Systems Engineering. This interdisciplinary program is designed to train professionals in the design, management, and security of digital infrastructure.
The University of California, Berkley recently launched a graduate certificate in Global Digital Infrastructure. The program provides a holistic overview of the technical, economic, geopolitical, and environmental dimensions of digital systems, including data centers. It aims to equip students from all academic backgrounds—from liberal arts to engineering—with the knowledge needed to manage, build, and sustain the massive infrastructure required for modern life and the expansion of AI.
Space: The Next Frontier for Data Centers?
We may even start to see programs looking into space. According to Upchurch, “If you don’t want to look at the ground, then look up at the sky, because there is AI infrastructure in space that’s coming.”
Gray DI discussed programs centered on the Space Economy during our 2021 webinar on emerging academic programs. We might have been onto something. Tech giants such as SpaceX, Google, and Blue Origin are racing to develop space-based data centers. And higher education is lending a hand. Leading research institutions are tackling the architectural and physical challenges of off-world processing. University of Pennsylvania researchers are working on developing scalable, solar-powered designs using tethers to support AI-level workloads. Scientists at Carnegie Mellon University are pioneering orbital edge computing to enable nanosatellites to process data locally.
Meeting the Moment
The “infrastructure moment” is a chance for higher education to launch programs that address the challenges and opportunities posed by the growing demand for digital and AI infrastructure. The sector is moving at light speed; programs must be as modular and adaptive as the data centers they support.
By focusing on both practical skilled trades and longer-term policies and innovations, we can ensure our students are not just users of AI, but the literal architects of the world it inhabits.
Want to see how this program stacks up alongside the other emerging opportunities we’re tracking? Explore the full Emerging Programs webinar, access the slides, and dive into the data behind all of this year’s programs to watch.
