As demand for data centers accelerates, so does the debate around their impact on communities, infrastructure, and the economy. Len O’Donnell, Chairman and CEO of our strategic partner, Affinius Capital, examines five common misconceptions and how the industry is evolving to address them.
Download the full piece or continue reading below.
Data centers are becoming critical to AI, cloud services, and the broader digital economy. As demand grows, so do questions about community impact, infrastructure needs, economic competitiveness, and America’s technological leadership.
Many concerns stem from past mistakes and poor community engagement. The five myths below revisit these concerns in light of advances in technology, practices, and standards.
Data centers consume significant amounts of water.
Reality: Closed-loop, air-cooled systems use just 1% of the water required by evaporative systems, while overall data center water consumption has fallen by more than 90%.
Data centers place added strain on the power grid.
Reality: Under the 2026 Ratepayer Protection Pledge, data center and AI firms commit to paying for the power infrastructure their growth requires, sparing residential ratepayers.
Data centers raise electricity costs for residents.
Reality: Self-funded infrastructure and steady long-term demand from data centers can spread fixed utility costs over more sales, possibly reducing per-unit delivery costs for current customers.
Data centers generate disruptive levels of noise.
Reality: Greater setbacks and sound-control measures can limit noise levels to 45–50 dB at the property line, comparable to a quiet suburban neighborhood.
Data centers provide limited community benefits.
Reality: In one U.S. market, data centers occupy less than 3% of county land but generate 38% of county revenue, supporting property tax cuts of nearly 30%.
Most jurisdictions are expected to favor pragmatic legislation over blanket restrictions, given potential revenue at stake and addressable environmental, cost, and noise concerns. While greater scrutiny may hinder new projects, it may reinforce the value of sites already positioned to move forward. Affinius Capital believes its portfolio remains well positioned geographically and from an entitlement, zoning, and permitting perspective.
Download or continue reading for the full commentary from our strategic partner, Affinius Capital.
Throughout American history, transformative technologies have often been met with a mixture of excitement, skepticism, and concern. Railroads, electricity, automobiles, the internet, mobile phones, autonomous vehicles, and now artificial intelligence all sparked vigorous public debate about their impact on society.
This paper will focus on the current debate over the rapid development of data centers that is raging across parts of the United States as companies such as Microsoft, AWS and Google, as well as frontier models, increase their spending to meet the explosive demand for this critical infrastructure. I will attempt to reflect on some of the causes of the debate including:
deconstructing some of the early missteps taken by communities and the data center companies,
how the industry has moved to correct those items, and
the lingering false information or "myths" that continue to fuel much of the current opposition.
As the United States seeks to maintain its global technological and economic leadership, data centers have become one of the most important forms of infrastructure in the modern economy. They support everything from hospitals, emergency services, and financial systems to scientific research, manufacturing, education, and the digital tools used by billions of people every day; and it is critical that the infrastructure needed to support the modern digital economy is developed at scale.
While there are appropriate debates about whether artificial intelligence and related technologies will create or eliminate jobs, there can be little debate that these technologies will enhance productivity, accelerate the progress in healthcare and increase economic growth. It is well documented that the investment in the AI and data center build-out has represented as much as 35% of GDP.
There is also a more recent debate about how rapidly AI models should be developed, but my focus herein is on the build-out of cloud data centers as well as the current generation of AI facilities focused on much-needed compute capacity for large learning and inference models.
What seems to be lost entirely in both the local opposition and recent political deliberation is that maintaining America’s leadership in this area is a matter of national security. Opposition intended to slow down the technology is both naïve and risks an erosion of America's leadership and security, because our adversaries face no such opposition and are continuing their expansion at a massive scale.
Where perhaps our view of this discussion differs from others that support data center development in an unfettered fashion, is that we believe the public discussion is both healthy and appropriate. The conversation surrounding data center development is an important one. Communities are right to ask questions about power, water, noise, land use, and economic impact, and it is equally important that those questions be answered with current facts rather than outdated assumptions.
Like many growing industries or in past deployments of new technology, there were some missteps in the early stages of data center advancement. For instance, early data centers did consume massive amounts of water, they did generate unacceptable levels of noise and, in rare instances, citizens did experience increases in their electric bills.
In my opinion, much of the current resentment is the result of a data center industry that did little to engage with the local communities in which they were developing their projects. There can be no doubt that some communities permitted some early data centers to be developed without giving proper consideration to how they might impact their neighbors. These mistakes could have been avoided through greater community engagement, and we are pleased to see that this is now happening across the industry.
Developers, hyperscalers, local governments, utilities, and community stakeholders have significantly changed how projects are planned, approved, and constructed. Today's facilities are more efficient, consume substantially less water, produce less noise, and are developed with far greater transparency and collaboration than their predecessors.
The challenge is that public perception often reflects yesterday's realities rather than today's facts. Stories and images from early projects continue to shape opinions even as the underlying technology and development practices have changed dramatically. The fact is most of these early missteps have been largely corrected or eliminated. The data center industry has matured and is now actively engaged with local and state governments as well as communities through active and collaborative public relations campaigns.
For example, we are experiencing the most significant period of new manufacturing construction that this country has seen in 40 years, yet we are not seeing similar protests over the location of these factories. Why is that? I believe it is because zoning codes long ago established where such projects should go and residents are familiar with the nature of these facilities.
Early missteps by the data center industry should not be surprising to anyone, and they were not the result of “dishonest” or adversarial behavior as some have suggested. At the start of this build-out, no one, not even the hyperscalers, could have anticipated how demand would explode over the coming five-year period. In addition, communities were highly solicitous of these projects because they produce significant property tax revenue while placing little strain on community resources.
Consider that, in 2020, a data center lease between 30 and 75 megawatts was considered large. By 2023, tenants were seeking 100 to 300 megawatts and, by 2024, 300 to 500 megawatts. Today, just six years later, it is typical to see deals of 1,000 megawatts (a gigawatt) or more. Suffice it to say a gigawatt campus requires more careful planning and collaboration than a single 75-megawatt building might have five years ago.
However, the scars left from the past have left us with a false narrative, or what we refer to as the “myths” that too often fuel the current debate. In America in 2026, these myths spread rapidly across social media, where misinformation too often overwhelms the facts.
The data center industry, led by developers and owners in full cooperation with hyperscalers, is now actively engaging with local governments and community leaders in advance of pursuing data center projects. At the same time, local governments have made important improvements to zoning codes to ensure projects are developed in a more responsible manner. These include limitations on water consumption and noise ordinances that place limitations on the decibel levels permitted through appropriate setbacks and other noise attenuation features.
The industry has also responded in a positive manner by utilizing technologies that significantly reduce noise and water consumption. In addition, the newest data centers consume power far more efficiently. Indeed, power consumption is still growing, but at a much slower rate because less power is needed today to produce an increment of compute capacity than just a few years ago.
With this in mind, it is instructive to address some of the many myths that are either outdated or simply incorrect:
Earlier generations of data center data centers consumed as much as two liters of water for every kilowatt of power deployed. In simpler terms, a 100-megawatt campus could consume as much as 10 million gallons of water per year depending on the local climate. However, most of this consumption was driven by the use of evaporative cooling systems in the data halls.
Most of the industry has greatly reduced its reliance on this type of cooling technology and data centers can now utilize closed-loop systems that circulate the same coolant (water, water/glycol, or another fluid) through a sealed piping network such that the cooling fluid is not open to the atmosphere and subject to evaporation. A closed-loop, air-cooled system, which has been part of our base design since 2021, will consume 1% of the water consumed by evaporative systems, and overall data centers have reduced water consumption by more than 90%. The reality is that most water consumption in the current generation of data centers will come from traditional uses such as bathrooms, water fountains, and landscape irrigation, which is no different than a lightly populated office building.
Data centers do require an enormous amount of electricity. This is needed to power the servers required to run our digital economy ranging from the large learning models that are advancing the knowledge and effectiveness of virtually every industry, but also to facilitate the ongoing growth of the cloud that is so critical to our national security and the financial sector. They also store our photos and videos, support every app on our phones, enable our hotel, restaurant and airline reservations, our on-line shopping and, of course, this is how we access social media.
Despite this ever-growing demand, the need for power is not unsustainable. First, efforts are underway to expand the grid. This was needed regardless of data centers to support growth in manufacturing, but data centers have escalated the focus on modernizing the grid. The White House recently identified this as a national priority.
Further to this, the White House Ratepayer Protection Pledge, announced in March 2026, establishes a framework for aligning AI and digital infrastructure investment with consumer protection and grid reliability. The pledge commits hyperscalers, AI companies, utilities, and data center developers to fund the generation and transmission infrastructure needed to support their growth rather than shifting those costs to residential ratepayers.
Since its launch with initial signatories, including Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI, the initiative has expanded to include hundreds of organizations across the energy and digital infrastructure ecosystem. We are proud that Corscale is among the companies supporting this effort, reflecting our longstanding belief that responsible infrastructure development should deliver economic growth, strengthen local communities, and protect consumers. The pledge's rapid adoption across the industry demonstrates a growing recognition that data center expansion and grid affordability can advance together when the beneficiaries of new demand assume responsibility for the infrastructure required to serve that growth.
In addition, power authorities and power companies have come to the realization that gigawatts of power have been previously committed to consumers that are not utilizing it, and efforts are underway to free up this capacity. As part of this effort, power companies are increasingly imposing “curtailment” and minimum demand requirements on data centers and other large consumers of power.
Essentially this allows the power companies to reduce the load sent to these facilities during periods of peak usage for the grid or guarantee a minimum revenue level for power allocated. To explain this further, the power companies have historically sized capacity and contracts to meet peak demand, but the reality is this level of demand exists for only a small percentage of time each year, and today the power industry is much more capable of forecasting these peak loads and managing the flow of power. Over time, this will allow the industry to use the grid’s available capacity far more efficiently.
Another technology that will become increasingly important is battery energy storage, commonly referred to as BESS. Put simply, these are large rechargeable batteries that store power and can deliver it almost instantly when needed. This allows data centers to respond to rapid changes in power demand, reduce sudden impacts on the grid and significantly reduce reliance on diesel generators, thereby making them greener. This is yet another example of how technology is allowing the industry to manage its power requirements more efficiently.
As a result, several countries now require or encourage the use of BESS in new data center projects; and the recent Executive Order from the Governor of Pennsylvania encourages deployment of BESS for various reasons throughout the body of the order.
What most don’t recognize is there is more capacity in the grid than public discussions and the media suggest. According to EIA data and other sources, as of 2025, the total capacity of the United States electrical grid is approximately 1,400 gigawatts, and that figure is growing with about 500 gigawatts in development. By contrast, peak summer demand was 745 gigawatts in 2023 and 2024, and average demand through 2025 was 506 gigawatts, indicating that, if power companies can properly manage allocations and demand, there is ample capacity for data centers for the foreseeable future.
Some governments now mandate that data center projects “bring their own power”, whether this is accomplished by acquiring new production within the grid or by various means of self-generation. As demand dictates, we will see more power produced both from gas-fired turbines and sustainable sources such as wind, solar and hydro power. Texas, for example, offers access to abundant low-cost reserves of natural gas, much of which is currently burned off at wellheads; but what may be less known is that Texas is also the largest producer of wind and solar power in the country. While this is indeed an encouraging trend, it takes time to design, permit and construct these power plants, so the industry needs to continue utilizing existing grid power, as evidenced by the temporary mortarium in Texas which is in place pending an audit of the grid.
Longer term, adding nuclear capacity to our grid will be necessary and highly efficient, but this is likely a decade or more away. Going back as far as 2010, the United States had plans to significantly expand its nuclear footprint, but much of this was put on hold following the Fukushima disaster in Japan. Future nuclear facilities will be safer, cleaner and SMR’s (small modular reactors) will be more easily deployed halls.
There were instances of higher electric bills early in the expansion, but the data center industry and governments have been taking constructive steps to ensure this can be avoided in the future. For example, in some markets, power authorities have implemented a surcharge on data centers such that they will bear the cost of new power infrastructure transmission. In other communities, data center developers are now required to pay for their own power infrastructure, including substations and, in some cases, transmission, adding hundreds of millions of dollars of cost to the projects. As mentioned above, some data centers will also produce their own power.
Often overlooked in the current debate is the idea that if a data center pays for the infrastructure, it requires and provides stable long-term load, it can lower transmission and distribution costs per unit by spreading fixed utility costs over a larger sales base. Existing customers in the community then benefit from improved utilization of infrastructure that would otherwise be underused. This has already played out in existing data center markets.
It is also important to note that the persistent inflation environment of the past several years caused all costs to go up, power certainly being a key component. The natural political reaction is to blame data centers for this increase – but the facts support a different conclusion. In Northern Virginia, the data center capital of the world, energy costs have increased on par with the national average. Conversely, states with much smaller data center footprints, such as California, Massachusetts and Connecticut, have seen increasing energy costs well above the national trendline.
This one is the easiest to address. In the early days data centers operated at approximately 80 decibels at the property line, which happens to be about the same as urban traffic noise. While older models of generators contributed to this, their impact has been overblown because they run less than 50 hours per year, either during severe weather events (where it seems data center noise is not a high priority) or during maintenance that is scheduled during weekdays to avoid disturbing neighbors.
AI data center campuses, the largest of the facilities being constructed, may be developed without relying on generators as their primary source of back-up power for data halls because they do not have the same needs for uninterrupted service as traditional cloud data centers. Generators are essential to cloud data centers because these facilities run the necessary systems for governments (military, intelligence community, etc.) and the global financial markets. In other words, if these buildings did not have full back-up power source during a power outage, it would cause considerable chaos and economic damage. However, these tend to be smaller facilities within existing cloud networks and using new techniques, sound can be managed effectively.
As an alternative, AI campuses of the future will increasingly rely on BESS for short-term backup power, while generators are used to provide long-term back-up to only critical components of the data center. This approach has the added benefit of further reducing noise.
Today, the average data center deploying proper setbacks and sound attenuation now operates at around 45 to 50 decibels at the property line, well below the 85 dBA that the CDC/NIOSH treats as hazardous. By comparison the average dishwasher produces 45-60 decibels, “quiet models” run closer to 35-44 decibels, and a normal conversation is said to be about 60 decibels.
Once again, the industry and the local governments identified a concern from residents and successfully addressed it in a variety of ways including greater setbacks, sound attenuation and scheduling of maintenance. In other words, there is no credible argument that a modern data center, properly sited, deploying full sound attenuation, and operating within lawful limits, will create noise that will impact the quality of life in communities.
The facts are as follows: A typical cloud facility might produce 50 jobs and perhaps 200 to 250 for an AI campus. These are high paying jobs attracting highly trained and educated people to these communities. In addition, industries emerging to supply and support data centers create hundreds of additional jobs. As indicated in a recent Oxford Economics economic impact study, there will be a 5.3x jobs multiplier in the region.
Those that argue against data centers simply ignore the thousands of jobs, including union labor, that are generated by data center construction. A large AI campus can require as much as $10 billion in construction and equipment contracts and might employ 1,500 people in high paying jobs on site for as much as three to five years as campuses are built in phases. The same Oxford Economics study indicates a project of this scale will support nearly 14,000 jobs throughout the region during the construction phase.
Then there is the tax revenue that can rise into the billions for states and communities. Consider Loudon County Virginia, the county that has benefited from data centers more than any other. In 2013, the county received $60 million in tax revenue from data centers, and in 2024, that had reached $875 million (roughly half of the county’s property tax revenue), and that figure is expected to be $1 billion in 2026.
As a result, county property taxes have dropped by nearly 30%, all because of facilities that occupy less than 3% of the county’s land. To highlight this, a county representative, speaking to recent tax cuts, explained that the county could afford them largely because of increased revenue from the data center industry, which now generates 38% of all general fund revenue!
The county has also seen property values rise steadily, although this is likely less due directly to data centers than to the high quality of life all this tax revenue has enabled the county to create, including great roads, schools and public facilities. The same official described the $4.7 billion budget as adding 236 positions in 30 different county departments to open new facilities, maintain or enhance current programs, support the county’s Capital Improvement Program, and pursue the Board’s strategic initiatives, thereby further reflecting the positive impact of data centers. As to the tax incentives, it was estimated that the county’s return on data center investment has been 26 to 1.
Despite all this success and the extraordinarily positive impact on the community, even Loudon County local officials are seeking to implement a pause on further approvals of data centers. Such a measure will surely bring legal action, and that alone will delay the advancement of these critical technologies.
Another example, Virginia embraced data centers more than any other state and JLARC, the nonpartisan oversight and research arm of the Virginia General Assembly, estimates the data center industry in Virginia contributes 74,000 jobs, $5.5 billion in labor-related wages, and $9.1 billion in additional GDP each year. Perhaps this is why America’s labor unions recently came out in support of data centers by threatening to withdraw support from any candidate that opposed them. If labor unions, developers, and Silicon Valley are aligned on an issue, it would suggest the wide-ranging benefits must be compelling.
Despite boisterous public and political opposition in some communities, other communities have gotten the message and are now actively negotiating community benefit packages as part of the data center approval process. One recently secured $125 million in the form of new fire stations, community facilities, vehicles, etc. This is all prior to the tens of millions of dollars in tax revenue the project will produce annually. At the same time, fewer projects are pursuing tax incentives.
Despite all this data, we continue to see citizens and government officials in some communities oppose data centers, even when these projects would dramatically enhance the quality of life in these communities. This is particularly true amid the current election cycle.
Thomas Mann is credited with saying, “Everything is Politics,” and the current data center debate is no exception. Elected officials that have been outspoken supporters of data centers have recently withdrawn this support, produced executive orders, or permitted temporary or permanent moratoriums in response to popular opposition to these projects. As one would expect, these elected officials are simultaneously being criticized because those opposed believe the actions insufficient or perhaps think they are disingenuous and, at the same time, those that support economic growth (such as labor unions) and those that are concerned with economic and national security are now becoming outspoken and politically active.
The irony is not lost that much of the opposition to data centers expresses itself through social media, which would not exist if not for data centers. To make this point, the Pennsylvania Chamber of Commerce somewhat sarcastically just proposed a “No Technology Pledge” urging supporters of data center moratoriums to also swear off the everyday technologies that depend on data centers.
Unfortunately, the build-out of this critical infrastructure has been caught up in the partisanship that prevails today, and this will remain in place through the November elections. However, it seems clear many elected officials do understand the economic benefits to their constituents and the critical nature of these facilities, and I am hopeful that they will resume their support once it is clear that community concerns and power have been addressed through constructive collaboration responsive to the concerns of residents.
Despite partisanship, I am an optimist. I believe that public debate is healthy and can optimize outcomes, but for this to be true, it is important that decisions be guided by current data, objective analysis and a clear understanding of both the costs and benefits of modern data center development. The digital build-out will continue, and the only questions are where will this occur, which communities will benefit from the economic impact of these projects, and which ones will lose out because they relied upon dated or false information?
In closing, I believe that America’s economic competitiveness, technological leadership, and national security increasingly depend on digital infrastructure. Sadly, the current wave of moratoriums, executive orders, and other methods of slowing the advancement of data centers and AI is already threatening our economic leadership and national security.
The challenge is not whether we should build data centers, but how to build responsibly. If we remain committed to transparency, accountability, and community partnership, there is no reason we cannot achieve both. Failure to do so will do lasting damage to our country.
The good news is this is already happening. Across the country, developers, hyperscalers, utilities and local governments are implementing stricter standards, better planning processes and more robust community engagement than existed only a few years ago. Yet public perception continues to be shaped by isolated examples, outdated information, or incomplete facts. This needs to change.
This Letter is for informational purposes only and should not be considered a recommendation to purchase any investment product sponsored by Affinius Capital or Accordant Investments. Past performance is no guarantee of future results. The selected examples of types of investments and asset classes were selected for illustrative purposes only and are not representative of all investments purchased, sold, or recommended for Affinius Capital or Accordant Investments, and it should not be assumed that types of investments identified were or will be profitable or are representative of investments that will be made in the future.
This Letter contains forward-looking statements which include statements, express or implied, regarding current expectations, estimates, projections, opinions and beliefs of Affinius Capital or Accordant Investments. Such statements are forward-looking in nature and involve a number of known and unknown risks, uncertainties and other factors. Certain information contained in this document has been obtained from published and non-published sources. Recipients should understand that any such information may not have been independently verified. Except where otherwise indicated herein, the information provided herein is based on matters as they exist as of the date of preparation and not as of any future date and will not be updated or otherwise revised to reflect information that subsequently becomes available, or circumstances existing or changes occurring after the date hereof.
For informational purposes only and not intended as investment advice or a recommendation. Past performance does not guarantee future results. All investments involve risk, including the possible loss of principal.