[Editor’s note: This is part 2 of a 3-part HousingWire special series on the impacts of data centers in housing. Part 1 is here.]
In 1998, when Steve Alloy succeeded his father, Martin, as president of Northern Virginia-based Stanley Martin Homes, buying land wisely was, as it is now, both a point of pride and a business necessity.
A cardinal rule of engagement and survival, as any homebuilder competing in the NVR- and Ryan Homes-dominated Mid-Atlantic region knew, was to counterpunch on land acquisition rather than compete head-to-head with the nation’s most efficient and profitable public homebuilder. Being able to “zag” when NVR “zigged” meant getting good at finding residential development locations that NVR land strategists would regard as too “hairy,” too time-consuming, or too fraught with engineering, permitting, or zoning complexities to fit into NVR’s hyper-efficient, asset-light land take-down system.
“We pursue different sites many more times than our national competitors do, just because our strategies are different,” Alloy told HousingWire/TBD. “We see many of them on a variety of sites, but in a lot of places, we would be kind of the go-to home builder on a difficult zoning case.”
For Alloy and his growing team, as part of Japan-based Daiwa House‘s decade-long investment in building a top-tier U.S. homebuilding platform, that “secret sauce” – detecting opportunity where competitors see operational complexity – suddenly took on a completely different dimension in late 2025.
Business partners approached Stanley Martin about a 189-acre tract the builder had painstakingly assembled, engineered and entitled for a 516-home community. The prospective buyer wasn’t another homebuilder. It was Amazon, which wanted to reverse years of residential entitlement work, rezone the property for industrial use, and build a hyperscale data center campus instead. Stanley Martin had invested roughly $51 million in the property. The transaction ultimately generated roughly $700 million.
For Stanley Martin and Daiwa House, it was a once-in-a-career capital allocation windfall.
For the homebuilding industry and the nation’s wider housing ecosystem, however, the deal immediately raised a more unsettling question.
If the world’s largest technology companies are prepared to spend hundreds of millions – even billions – of dollars securing land for artificial intelligence infrastructure, what happens to an industry already struggling to acquire residential land cheaply enough to build homes that American households can afford?
Artificial intelligence has triggered one of the largest capital investment races in modern business history. The Bank for International Settlements estimates hyperscaler capital expenditures alone will exceed $700 billion this year, with aggregate AI infrastructure investment expected to climb into the trillions over the coming years as companies compete to secure computing capacity before their rivals do.
Those investment dollars translate into an unprecedented wave of data center development across the United States.
Local governments are debating zoning ordinances. Electric utilities are racing to expand transmission capacity. Communities are wrestling with questions about water consumption, tax revenues, noise, environmental impacts and electrical reliability. Realtors are beginning to see pricing distortions in neighborhoods located near proposed campuses, themes explored in HousingWire’s earlier Part I analysis.
For homebuilders, though, the issue is different. The question isn’t whether AI infrastructure is reshaping commercial real estate. It’s whether hyperscale developers are poised to become the next structural obstacle to market-rate homeownership affordability.
The answer, at least based on today’s evidence, appears considerably more nuanced than much of the current political rhetoric suggests.
Northern Virginia’s cautionary tale
If there is one housing market where data centers have unquestionably begun changing land economics, it is Northern Virginia.
Chris McGrath, Mid-Atlantic Division President for K. Hovnanian Homes, has watched the competitive landscape morph dramatically over the past several years.
His division operates in the nation’s largest concentration of data centers – an area often referred to as “Data Center Alley,” where estimates suggest roughly 70% of the world’s internet traffic passes through server farms spread across Loudoun, Fairfax and neighboring Virginia counties.
The effects have evolved from hypothetical to starkly real over the past 24 months. According to McGrath, K. Hovnanian has lost multiple residential opportunities after landowners received substantially richer offers from data center developers.
In Spotsylvania County, McGrath says, sellers or brokers went radio silent on preliminary residential negotiations after companies such as Amazon entered the picture seeking sites for future facilities.
In Loudoun County, roughly 700 residential lots effectively disappeared from the housing pipeline after a seller elected to pursue data center rezoning rather than complete negotiations with the builder.
In Fairfax County, McGrath says the approval of a nearby data center also reduced the attractiveness of adjacent residential land, illustrating another way AI infrastructure can reshape development patterns.
“It’s hard to turn around here and not hear of another story of, especially large parcels, going to a data center,” McGrath said.
He doesn’t argue that data centers alone explain Northern Virginia’s dramatic increase in land prices. Land values, he notes, have risen sharply since before COVID for a host of reasons. Still, he has little doubt the AI build-out has added another powerful bidder to an already expensive market.
“Every land seller now thinks that they’re gonna make hundreds of millions of dollars on their [property], because Amazon’s gonna go buy it, right?” McGrath said. “That just complicates every discussion, because every seller thinks that the land is worth a lot. This adds another layer because there are these groups out there who are paying astronomical amounts. The overall effect of that is the parcels that do work for residential housing become much more expensive, and overall it limits supply.”
The experience has forced K. Hovnanian to rethink how it competes for land.
Rather than relying heavily on fully entitled acquisitions, McGrath says the company increasingly pursues raw land directly from farmers, business owners and other landholders, assuming responsibility for entitlement and rezoning itself.
“We’ve countered this by leaning almost exclusively into rezoning and entitlement deals,” McGrath explained. “Instead of looking for deals that are going through the rezoning process with a seller or that are already completely entitled, we’re going out there and sourcing our own deals. We try to find the raw land and meet with the farmer, business owner or building owner. That way, we can control the fate of the project.”
If that strategic pivot rings familiar, it may be because it mirrors the strategy Steve Alloy began refining nearly three decades earlier – creating value through engineering, entitlement expertise and local relationships rather than outbidding better-capitalized competitors for finished residential land.
That parallel also hints at a more sweeping conclusion.
Northern Virginia offers compelling evidence that data centers can materially affect local residential land markets under the right conditions. Whether it represents the future of residential development across America is a different question entirely.
Data center development also competes for everything else
If Northern Virginia demonstrates what happens when residential development collides head-on with hyperscale data center investment, it may also illustrate why the rest of the country is unlikely to follow precisely the same path.
The Stanley Martin transaction was extraordinary because it was, by almost any measure, extraordinary. It involved one of the country’s most sophisticated residential land developers, one of the world’s largest technology companies, one of the nation’s most mature data center markets, and a parcel already carrying years of entitlement work that suddenly became exponentially more valuable for an entirely different use.
Most residential development doesn’t look like that. Nor do most data center projects.
This gets to the central question confronting homebuilders, developers, land investors and policymakers: Is America’s accelerating AI infrastructure build-out becoming the next structural obstacle to producing market-rate housing?
Evidence today suggests a complicated – and, in many respects, less five-alarm-fire-freaking-out – discerning response. The broad-brush reality is that most residential communities and hyperscale data centers are pursuing fundamentally different real estate.
Homebuilders’ trade group sees a growing concern
The National Association of Home Builders has become increasingly vocal about what it views as an emerging conflict between America’s need for more housing and the explosive demand for AI infrastructure.
NAHB argues that data center developers, backed by the financial resources of the world’s largest technology companies, can outbid residential builders for strategically located land while simultaneously competing for construction labor, utility capacity and local infrastructure. The organization has warned that the trend could make it more difficult – and more expensive – to bring new housing to market in communities already struggling with affordability.
Those concerns are hardly theoretical in markets such as Northern Virginia, where builders like K. Hovnanian have documented losing residential opportunities to hyperscale developers and where landowners increasingly see data centers as an alternative path to dramatically higher valuations.
At the same time, the question facing builders nationally isn’t whether these pressures exist.
It’s whether they represent a broad new structural shift in residential land economics – or whether they remain concentrated in a relatively small number of markets with unique combinations of power infrastructure, fiber connectivity and hyperscale demand.
Different businesses, different dirt
Ting Qiao occupies a uniquely useful vantage point. As co-founder of Texas homebuilder Wan Bridge and manager of W Land Development, Qiao has experience acquiring and developing sites for both residential communities and data centers.
Rather than seeing the two uses routinely competing for the same land, he sees two businesses with fundamentally different site-selection criteria.
Housing succeeds where families want to live – close to employment, schools, transportation corridors, retail amenities and community services. Data centers do best where electricity, fiber connectivity, environmental buffers and massive tracts of land intersect.
“Based on our development experience, these two product types generally have limited site overlap because their land, infrastructure, electrical and environmental requirements are fundamentally different,” Qiao said.
That observation echoes what Anita Verma-Lallian sees every day in Arizona.
As founder and CEO of Arizona Land Consulting, Verma-Lallian specializes in acquiring and assembling sites for large-scale data center campuses. Last year, her firm partnered with billionaire investor Chamath Palihapitiya to acquire approximately 2,100 acres west of Phoenix for what could eventually become a $25 billion AI infrastructure development.
She says the industry’s rapid evolution is actually pushing many projects farther away from traditional residential growth corridors.
Artificial intelligence requires facilities dramatically larger than earlier generations of cloud computing infrastructure. Increasingly, developers are pursuing campuses capable of supporting more than one gigawatt of electrical capacity.
Those requirements reshape the search process.
“Ideally you don’t want data centers where residential development could go, just because that means you’re probably in a more densely located area,” Verma-Lallian said. “So we’re being intentional about putting our data centers far away from development in areas that are a little bit more industrial, where you’re not impacting residents and neighbors.”
Increasingly, success isn’t determined by proximity to rooftops but by proximity to power. That’s why the current public conversation risks conflating two very different land markets.
Residential land continues to derive its value primarily from housing demand. Data center land is increasingly being valued according to something entirely different. Electrical capacity.
“You have all these tech companies that are being forced to build data centers, so there is quite a bit of capital behind that demand,” Verma-Lallian said. “If you have land with power, it’s really unlimited what you could sell that for, because there’s such a big need for it.”
Unlike traditional real estate transactions priced by acreage, some data center sites are now being valued according to the amount of electrical capacity they can deliver.
“The pricing you’re seeing on land with power, it’s just something that we’ve never seen before,” she said.
That does not necessarily translate into higher prices for conventional residential development parcels.
Instead, it creates a premium market for a relatively small universe of sites possessing characteristics most residential builders neither need nor necessarily want.
The affordability equation hasn’t changed
None of that minimizes the challenges builders face. If anything, it sharpens where those challenges actually lie. The homebuilding industry entered the AI era already confronting a structural affordability problem.
Mortgage rates remain elevated. Household budgets remain stretched. The refrain among so many of the homebuilding business leaders right now is, “We’re buying our sales.”
Builders across virtually every product category continue searching for ways to reduce asking prices while protecting margins. Doing that almost inevitably requires reducing land basis, shortening cycle times, improving operational efficiency and extracting cost throughout the production process.
Those economic realities existed before hyperscalers began their AI infrastructure arms race.
They remain the industry’s defining challenge today.
That’s one reason John Burns Research & Consulting Senior Research Analyst Dillan Krieg urges caution before drawing broad conclusions from a handful of high-profile transactions.
While the firm’s research shows residential parcels being converted to data center uses in markets such as metropolitan Washington, Chicago, Indianapolis, Dallas, Austin and San Antonio, Krieg says there is not yet evidence demonstrating that data center development broadly is driving residential land appreciation.
Regional housing markets continue to behave primarily according to their own supply-and-demand fundamentals.
Indeed, Burns researchers found little evidence of similar residential conversions across California or the Pacific Northwest, where entitlement constraints already make residential land exceptionally scarce and valuable.
Residential land markets remain, first and foremost, residential land markets, and in the current uncertainty-dominated environment, many homebuilders’ outlook and appetite for land is “flat to down.”
Where builders may feel AI’s impact
That doesn’t mean the AI infrastructure boom will leave homebuilding unscathed. Quite the opposite.
It simply means the most significant impacts may emerge somewhere other than the land acquisition department. Dallas-area builder Bloomfield Homes provides one example.
Chief Executive Officer Don Dykstra says Bloomfield has not found itself routinely bidding against data center developers for residential property.
Instead, he sees the competition emerging through shared infrastructure.
In one instance, a municipality allocated significant sewer capacity to a data center project, forcing Bloomfield to redesign a nearby residential development around septic systems instead. More broadly, Dykstra acknowledges that data centers consume enormous quantities of water, sewer capacity and electricity, even if those impacts have not yet materially affected demand for his communities.
That observation may prove more consequential over the next decade than individual land transactions.
- Utilities.
- Electrical substations.
- Transmission lines.
- Water infrastructure.
- Wastewater treatment.
- Heavy civil contractors.
- Electrical subcontractors.
- Mechanical trades.
- Concrete crews.
These are resource pools shared by both industries.
Competition there may ultimately prove more significant than competition for dirt.
A familiar challenge in unfamiliar form
There is another reason homebuilders can and should resist viewing hyperscale developers as an existential new adversary. Competition for capitalized land is hardly new.
For decades, builders have adapted to changing competitive landscapes created by regional developers, master-planned community operators, Wall Street-backed land funds, institutional single-family rental investors and increasingly concentrated public homebuilders pursuing market share.
Each wave has altered land values in certain markets. None has fundamentally rewritten the economics of homebuilding. Data centers are likely to become another variable in that equation rather than its defining feature.
The industry’s long-term competitive advantage will continue to rest where it always has: identifying opportunities others overlook, navigating increasingly difficult entitlement environments, engineering value into complex sites and maintaining a land basis that allows homes to reach buyers who can actually afford them.
Steve Alloy’s $700 million transaction wasn’t ultimately a story about AI. It was a story about residential land expertise. Stanley Martin spent years creating value through entitlement, engineering, local relationships and patient capital before a hyperscaler ever appeared.
Amazon merely revealed how much value that work had created. For most homebuilders, that’s probably the lesson. Artificial intelligence may permanently reshape commercial real estate, infrastructure investment and regional economic development.
But the residential development and homebuilding industry’s biggest affordability challenge remains what it has been all along: finding, creating and delivering residential land at a cost that makes the next new home attainable for the next American household.
HousingWire’s Tyler Williams contributed to this report.
