When ChatGPT launched in late 2022, I watched something remarkable happen.
Within two months, it hit 100 million users, a growth rate that sent shockwaves through Silicon Valley. Today, it has over 800 million weekly active users. That launch sparked an explosion in AI development that has fundamentally changed how we build and operate the infrastructure powering our digital world.
That all happens in data centers… and lots of them. Data centers are economic engines and technical marvels, but they also come with real costs that communities are only beginning to feel.
Let me walk you through the three major pros and cons of today’s data center boom and what they mean for our future.
The pros of data centers: Why communities are rolling out the red carpet
1. They’re an economic catalyst (especially for rural areas)
I’ll start with what gets politicians and economic development officials excited: the money. And there’s a lot of it.
In 2025 alone, data center deals hit a record $61 billion globally, with the majority happening in the United States. We’re not talking about small investments. Meta is pouring $27 billion into its Hyperion data center in Louisiana after teaming up with Blue Owl Capital, expanding what it initially described as a $10 billion project. Google announced a $25 billion investment in AI infrastructure in Pennsylvania. The numbers are staggering when you step back and look at them.
According to recent economic analysis by S&P Global, data center and related high-tech spending accounted for 80% of the increase in US economic demand in the first half of 2025. That’s not a typo; these facilities are literally driving the US economy forward.
Take Texas, for example. In 2024, data centers paid an estimated $3.2 billion in taxes to state and local governments. Or look at Loudoun County, Virginia, where officials expected real and property tax revenues from data centers to hit $895 million in fiscal year 2025, almost as much as the county’s entire operating budget of $940 million.
For states and municipalities competing for these projects, the appeal is obvious. Companies bringing in significant capital investment and substantial tax revenue that can fund schools, roads, and public services. It’s why Virginia, Texas, and now more than 20 other states offer tax breaks to attract data centers to their regions.
2. They (mostly) create jobs
The promise of jobs is a powerful lure.
When Meta first announced its “Hyperion” data center in Richland Parish, Louisiana, state officials projected 500 permanent “direct new jobs” and over 5,000 temporary construction jobs at peak build.
OpenAI’s ambitious Stargate Project touts the potential for 100,000 jobs across construction and operations. While the permanent operational staff for a highly automated data center is often smaller than people hope — sometimes as few as 20 to 150 for a large facility — the construction boom is very real.
It creates demand for electricians, HVAC technicians, and engineers, a fact underscored by Microsoft President Brad Smith’s estimate that the US will need “half a million new electricians” in the next decade to meet this demand.
3. They can force infrastructure improvements (for better or worse)
A new hyperscale data center can use as much power as a small city.
To support it, utilities and developers often have to invest in local power grids, new substations, and fiber-optic networks. In theory, this upgraded infrastructure can benefit the broader community and make the area more attractive for other businesses.
There’s also a geopolitical angle here. The US government, under both the current and previous administrations, has made it clear that leading in AI is a national priority.
The Trump administration’s “America’s AI Action Plan” explicitly calls for building a vast AI infrastructure to maintain competitiveness. Having this critical infrastructure on domestic soil is seen as vital to national security and economic sovereignty, helping prevent over-reliance on foreign data processing hubs.
Love it or hate it, our digital world runs on these facilities.
Every AI query, streamed movie, cloud backup, and smart device interaction touches a data center. They enable the services that have become integral to business, education, and communication. Without the massive scale of data centers, the generative AI toolsnow weaving into everything from drug discovery to customer service simply wouldn’t exist.
The cons of data centers: The staggering, not-so-hidden costs of AI factories
1. They are voracious, grid-straining energy hogs
This is the big one.
As early as 2024, the International Energy Agency noted that the US already accounted for 45% of global data center electricity use. And AI is changing the math entirely. A traditional cloud data center might draw power equivalent to 25,000 homes. A new AI hyperscale facility can demand as much as 100,000 homes or more.
Meta’s Hyperion center in Louisiana is expected to draw more than twice the power of the entire city of New Orleans, according to reports.
The strain is real and present. The North American Electric Reliability Corporation warned that “the speed at which large load facilities are being built adds stress” on the grid due to this demand. The result? Everyone pays more.
A June 2024 study from Carnegie Mellon and North Carolina State University projected US electricity rates could rise 8% on average by 2030 due to data center demand, with hotspots like Northern Virginia facing potential increases of over 25%.
2. They consume water at an alarming rate
If the power demand doesn’t shock you, the water usage will. Data centers are incredibly thirsty. For cooling, these facilities suck in millions of gallons daily.
A medium-sized facility can consume 110 million gallons of water per year for cooling, equivalent to about 1,000 households, while the larger ones can consume up to 5 million gallons per day.
In 2023, Google’s data centers alone withdrew 6.1 billion gallons of water globally. Their Council Bluffs, Iowa, facilities — just two data centers — pulled 1.4 billion gallons from the local water supply.
Texas, for example, could see AI data centers consuming up to 161 billion gallons annually by 2030, nearly 3% of the state’s water usage, a serious concern given recurring drought conditions. In regions where water is scarce, this can affect agriculture, municipal supply planning, and local ecosystems.
3. They impose heavy burdens on local communities
The impact isn’t just resource-deep.
These are massive, often windowless complexes covering hundreds of acres of former farmland or forest. They bring noise pollution from 24/7 cooling systems and periodic testing of hundreds of diesel backup generators, each the size of a railcar.
Imagine living next to a heavy metal concert that never ends. That’s the reality for some folks. A data center’s cooling fans can crank out 92 to 96 dB(A).
Data centers are constantly updating hardware. Old servers and networking equipment become obsolete quickly due to rapid advances in chip performance, especially for AI workloads.
This creates significant quantities of electronic waste (e-waste), which often contains toxic materials and requires careful handling. This e-waste challenge spans global technology infrastructure and raises questions about long-term material sustainability.
Final thoughts
From where I stand, data centers are the infrastructure backbone of the AI era, transformative, necessary, and inevitable. But they aren’t free of trade-offs.
As we move forward, I believe we need much more transparency about the true costs of these facilities, not just the tax revenue they’ll generate, but the water they’ll consume, the electricity they’ll demand, the emissions they’ll produce, and the lasting impacts on the communities that host them.
The companies that power our future must be accountable for the full footprint they leave today.
Also read: Enterprise leaders are scaling faster than their guardrails, and AI adoption trends show why governance and upskilling now determine who wins.
