The Truth About Data Centers and Why AI is Driving the Real Crisis
- 4 days ago
- 8 min read
Kass James is an assistive technology specialist with a master’s in management of information systems from the University of Houston’s Bauer College of Business. Fully licensed in ADA compliance and environmental access, he’s a partner at The Spoonie Advocate Associates.
There is almost constant conversation about data centers, but much of it seems one-sided and rooted in preconceived ideology. It is incredibly useful to start these conversations from a place of shared understanding and truth about the function, necessity, and problems surrounding big data.
It’s important to talk specifically about which kind of data center you mean
There are five main kinds of data centers: enterprise, colocation, hyperscale, cloud, and edge.
Enterprise facilities are owned and operated exclusively by an organization. Although they are expensive to own and operate, organizations that need to store sensitive data or proprietary information often choose to invest in the control afforded by an enterprise data center. Government agencies also run these centers for the same reason.
Colocation facilities are professional-grade facilities that rent space to other businesses. These facilities are operated by real estate companies that rent space to tenants who use the cooling, power, and network infrastructure provided by the larger facility. It is important to note that this is a physical real estate agreement, not a digital real estate lease. The infrastructure required to run these systems is expensive, and sharing resources lowers operating costs.
Cloud data centers are the digital real estate option for businesses that want to operate entirely on another company’s infrastructure. Unlike colocation, where only the infrastructure and space are rented, and the tenant owns the physical equipment, cloud data centers are fully owned and operated by a single business that rents digital space to other businesses. This is best for smaller companies or those that need to scale infrastructure rapidly. Amazon and Google own the bulk of these facilities, but many businesses operate cloud data centers.
Edge centers are smaller facilities designed to reduce service delays. These facilities are why media streaming services have become so much faster and why the internet operates better every year. While that’s partially due to increased internet speeds and improved personal devices, much of it is due to these edge centers supplementing local traffic. These are also rarely discussed, as they are typically the size of a shipping container and require relatively few resources to run compared with big data centers.
Hyperscale facilities are typically what people mean when complaining about data centers. These centers aren’t built for a single purpose but are massive sites covering millions of square feet and incurring high resource costs. These sites are typically run by one of the “big three,” Amazon Web Services, Microsoft Azure, or Google Cloud, and can be larger than cities. These facilities typically provide a combination of cloud storage, supercomputing services, and AI data infrastructure. Before the modern AI boom, they were built in areas with cheap resources or tax incentives, which is why so many are located in areas with cold weather, access to cheap hydroelectric systems, or proximity to Northern Virginia’s critical internet exchange point.
Today’s society would be badly crippled without big data
Everyone uses these systems daily to function. Mobile data, GPS, financial transaction systems, utility management systems, and traffic pattern management software all run on big data centers. Amazon Web Services, or AWS, is the largest of these platforms, running everything from Netflix to Airbnb. However, almost every major manufacturer uses this system to back up data, manage websites, and keep its systems running.
In 2025, humans produced a bit more than 402 terabytes of data per day. However, that number is expected to increase by 50% each subsequent year, possibly more. Some of that is “junk” data, meaning things we create but never use. However, the current estimate is that at least 300 terabytes of that daily data will be kept for at least 10 years. We send over 16 million texts every minute and 361 billion emails every day, and we spend an hour on social media every day. Data centers store and manage all of it.
Companies require data centers to function
While the internet runs on data centers, they are also essential to day-to-day functionality in today’s business environment. PayPal alone processes more than $22,000 per second in digital transactions, and that’s a relatively small number compared with ecommerce giant Amazon. In 2025, the top users of AWS were Sony, Adobe, Facebook, Johnson & Johnson, and 3M. So, while the public sees the PlayStation Network and Facebook, it’s research, manufacturing, and digital media production that make up the majority of AWS usage. These data centers provide necessary infrastructure while reducing infrastructure costs. It is, after all, more efficient to build massive sites optimized to provide these services than to have each business build and maintain smaller, independent sites.
In many ways, this is the most efficient model
Colocation and cloud centers are so popular and profitable because they are significantly more efficient than enterprise operations. The resource cost of running a hyperscale center is massive, but it’s minuscule compared with meeting each business’s needs through individual systems. Owners of these sites are motivated to optimize them to lower resource costs and maximize efficiency.
The loss of data centers would be catastrophic
Beyond the financial infrastructure, many people rely on data centers to live. Healthcare systems, air traffic control, accurate weather prediction, emergency dispatch, resource management, and basic utilities rely on these systems to operate. Local governments use them to manage traffic on our roads and operate 911 dispatch systems. Hospitals rely on cloud services like Epic to manage everything from prescriptions to patient health records. We need these data centers to keep the power on and people healthy.
AI is where it all goes wrong
Our reliance on these systems and their optimization are key to their value. However, the recent AI boom has flipped the narrative. AI is NOT a business necessity for anyone other than AI businesses. No company is reliant on AI to provide necessary functionality, and AI is not keeping people alive. Predictive systems that run in data centers are not the same as AI, and while many keep the world running, they are not the same as artificial intelligence.
The resource cost of AI is astronomical
While the systems that run AI consume massive amounts of water and electricity for cooling and require specialized hardware, this problem is compounded by the ever-increasing complexity of AI models. These systems don’t have a static infrastructure cost. Unlike other areas of software engineering, which prioritize efficiency, statistics show that these AI models’ resource requirements rapidly increase over time.
Beyond utility requirements, these systems also have an environmental impact. Water consumption is a major concern, but the heat and other pollutants they generate are a problem as well. Many municipalities require hyperscale systems to have alternative power plant facilities, many of which lack EPA standards because they are not regulated as utility providers. They also generate massive amounts of heat and noise as part of their daily operations. Noise and heat pollution don’t just affect wildlife and the local environment. Many neighbors have successfully filed lawsuits against data centers under nuisance laws because of their impact on property values and daily life.
The average consumer also shares the financial burden. Utility infrastructure upgrades are typically shared across customers. However, when a new data center uses the same utilities as 100,000 homes but pays only one share of the infrastructure development costs, utility costs skyrocket. The cost of RAM and microchips has also risen by over 300% in the last year due to the development of these centers. The cost of consumer electronics has gradually fallen since the 1980s relative to inflation, but it has surged by over 18% in the last year. The average customer is paying more for development costs every day.
There is also the environmental impact of systems going out of date every 4 to 8 years. While electronics recycling facilities exist, many of the heavy metals and pollutants will end up with local waste disposal services, many of which are unprepared for the rapid expansion of necessary hazardous materials treatment facilities. Since most of these facilities are publicly funded, that’s another cost taxpayers have to cover through the construction of these systems.
The lie of employment
Local municipalities are sold a pipe dream of financial stability and job opportunities, yet these jobs exist only during construction. At the time of writing, the Switch TAHOE RENO 1 data center in Nevada is the largest by square footage and power consumption, with 1.3 million square feet and 130 MW of consumption, with a total of 650 MW planned. It employs about 750 people across the entire campus. Given that nearby Reno’s downtown is only slightly larger and employs over 22,200 people while using an estimated 8.5 MW of power, the relationship between infrastructure costs and employment is poor. Also, much of the highly skilled and specialized workforce is brought in from Silicon Valley and other sites to run data centers.
The problem of politics
As with any new industry, the data center market has few regulations, and companies use money to lobby local and federal governments for the best opportunities. Politics is lagging, and local municipalities are paying the cost. Because there’s very little regulation, the environmental and financial impact can be destructive. Within months of the development of Amazon’s Canton facility, local residents started experiencing health problems stemming from the use of diesel generators and massive water consumption. There are multiple reports of pollutants entering Georgia’s water supply and of the Flint River ecosystem being strained by the high rate of consumption in that hot climate. Some international estimates put AI water consumption at over 1.5 trillion gallons by 2027. That’s just AI, not every other use of data centers.
Georgia Power, a for-profit utility company, is also considering using eminent domain to seize and demolish 330 parcels of privately owned land to make way for a new power line to supply that state’s growing data center market. Yet state and local lawmakers seem confused about the necessity of this action, since eminent domain is intended for public benefit and requires a valid public purpose rather than private gain.
Currently, local municipalities are fighting back, fueled by residents’ outrage. The only options available to these communities are bans or moratoriums on building such facilities, yet these facilities will eventually be constructed. State and federal legislation is lagging behind modernity, and the issue will not be quickly resolved.
Point your rage at the problem
If you, like me, are concerned about the financial, health, and environmental impacts of these facilities, educate yourself and get involved in the solution. Most politicians don’t know what kind of data center is being proposed and don’t consider the long-term tax implications of infrastructure costs. For a hyperscale facility, it’s best to think of it as a new city that produces almost no taxable income, consumes tremendous resources, and requires decades of public and private infrastructure improvements. It’s the unrestricted and rapid expansion of these facilities that is the cause of the problem. Previously, companies would carefully plan and design the most efficient facility possible in a part of the world with the best available resources. Now, AI fuels a desire to build where there is available land and access to construction services. It’s this unlimited expansion, with no regard for resources, that is the problem.
Utilities, water, and public infrastructure function perfectly well without AI, and AI depends on them. Therefore, utilities, water, and public infrastructure are more important than AI and should be prioritized. This is not to say that we shouldn’t build new and better data centers, but they should not come at the expense of public well-being.
Read more from Kass James
Kass James, Healthcare Business and Disability Specialist
Kass James is a forerunner in the field of disability rights, corporate responsibility, and healthcare business. Having been physically disabled for most of his life, Kass was acutely aware of the lack of accessibility in the workplace. His work focuses on restructuring healthcare to increase profitability while benefiting patients, as well as doing patient assessments for ADA compliance and assistive technology. He’s a partner with the Spoonie Advocate Associates, an organization pushing for increasing value and patient outcomes through common sense and responsible change.
