The construction of AI data centers continues unabated, with energy demand that in the United States alone could reach 106 gigawatts by 2035, according to a recent BloombergNEF report. This represents a 36% increase over the company’s previous forecast, published just seven months earlier. It is another sign of the existing pressure to manufacture these infrastructures, key to feeding artificial intelligence systems.
However, there are serious limitations in the availability of energy, materials, equipment and, perhaps most importantly, the lack of engineers, technicians and qualified personnel that could make data centers fail, they explain from IEEE Spectrum where they analyze this problem.
The electrical grid engineering workforce is currently shrinking and data center operators are suffering from a shortage of trained personnel. Laura Laltrello, director of operations at Applied Digital (a data center developer and operator), says that demand for civil, mechanical and electrical engineers, as well as construction management and supervision positions, has accelerated in recent months. The growing demand for qualified workers has forced your company to expand the hiring perimeter.
Lack of qualified personnel to build AI data centers
According to the State of the Data Center Report 2025 from the Association for Computer Operations and Management (AFCOM), 58% of data center managers identified multi-skilled data center operators as the top growth sector, while 50% noted growing demand for data center engineers. Security specialists are also a critical need.
Over the next decade, the U.S. Bureau of Labor Statistics projects the need for nearly 400,000 additional construction workers by 2033. By far the greatest needs are in electrical infrastructure, electricians, plumbers and HVAC systems, and approximately 17,500 electrical and electronic engineers. These categories relate directly to the skills required to design, build, launch and operate data centers for AI.
«The challenge is not just the absolute number of workers available, but the timing and intensity of demand«says Bill Kleyman, author of the AFCOM report and CEO of AI infrastructure company Apollo. “Data centers are expanding at the same time that utilities, manufacturing, renewable energy, electric grid infrastructure and construction compete for the same skilled workforce.” AI, as demanding as it is insatiable, is intensifying this pressure.
Large data center developers and construction companies face Huge demands to build faster, larger, more energy-dense facilities. That increase in construction pace requires tens of thousands of additional engineers. The engineering talent shortage is accompanied by persistent staffing shortages in data center operations and facilities management professionals, electrical and mechanical technicians, high-voltage and power systems engineers, qualified HVAC technicians with experience in liquid or high-density cooling, and construction specialists familiar with the complex integration of mechanical and electrical systems.
Technical schools and applied education programs are among the most effective drivers of workforce growth in the data center industry. They focus on practical skills, facility operations, power and cooling systems, and preparation for the real world of work. “Technical schools are driving the incorporation of new talent into an industry that is experiencing exponential growth and an almost infinite appetite for qualified workers”says the AFCOM manager.
Suppliers and industry associations also are actively addressing the talent gap. The Microsoft Data Center Academy is a public-private collaboration involving community colleges in regions where Microsoft operates data centers. Google supports local nonprofits and universities by offering training in IT and data center operations, while Amazon has its own data center apprenticeship program.
understanding that “The best way to predict the future is to build it”universities are also considering this challenge and adapting their curriculum to prepare students for a future of digital infrastructure.
We can’t feed the insatiable AI at this rate
The problem is that the needs of AI go ahead of the ability of humans to feed it. The lack of qualified personnel to build data centers for AI joins another even greater problem, such as energy and environmental sustainability. As artificial intelligence systems expand faster and faster, their energy footprint grows and some reports indicate that it even threatens the global sustainability of the planet.
Another example of excessive consumption is that Water use for cooling has increased by almost 10% as server density increases. And the geographic lottery of energy sources means that the same AI query can generate twice as many emissions in coal-dependent regions as in those using renewable energy. The need for cheap energy has generated renewed interest in nuclear energy and specifically in small modular reactors.
The CEOs of Microsoft and OpenAI already indicated that the big problem for the deployment of AI did not lie in computing capacity, but in the lack of energy to power the thousands of accelerators that are being installed in data centers.
