An Opportunity (Zone) to Determine Where Data Centers Go

The public is turning against data centers. Polling conducted by Heatmap over the past year shows that the percentage of Americans who oppose local data center development grew from 42% in August of 2025 to 75% a year later.¹ Furthermore, new Searchlight polling shows that voters believe Congress should prioritize issues surrounding AI and data centers at a comparable level to border security and public safety.²

Currently, operational data centers largely exist in urban areas. A Pew Analysis from April put this percentage as high as 87%³, with Northern Virginia suburbs alone accounting for 13%of the world’s data center capacity.⁴ For this reason, we often associate the data center build out with urban and suburban communities. However, the location of data centers – and the demographics of Americans who live near them – may shift following the passage of the One Big Beautiful Bill Act (OBBBA).

OBBBA introduced “rural opportunity zone funds” which, starting on January 1 of 2027, will set a lower threshold to receive tax benefits for investments in certain rural areas, while the urban and suburban investment threshold remains unchanged. We analyzed the available evidence on proposed data center construction and found that there have already been dozens of data centers being proposed in rural areas. These proposals have largely coincided with OBBBA making it easier for developers to claim tax incentives in designated rural areas, and it’s worth noting that the House Ways and Means Committee, which drafted the text, explicitly identified data centers as possible investment opportunities.⁵

Lawmakers should be aware of these developments and understand that rural opportunity zone tax credits might shift where hyperscalers build new data centers.

What are Qualified Opportunity Zones?

In 2017, the Tax Cuts and Jobs Act (TCJA) established “opportunity zones”, which identify low income communities and provide tax benefits to companies that invest in them. The purpose, according to the IRS,⁶ is to “spur economic growth and job creation in low-income communities while providing tax benefits to investors.” 

Only certain low-income census tracts are eligible to become opportunity zones. As modified in OBBBA, eligible census tracts are areas which have a median family income that is at or below 70% of the statewide median family income or metropolitan area median family income, or – in cases where the census tract has a median income of up to 125% of the statewide or metropolitan median family income – has a poverty rate of at least 20%. 

The governor can nominate up to a quarter of their state’s qualifying census tracts for opportunity zone eligibility; then, the Treasury Department gives formal approval. Historically, businesses invested less in these low income areas for reasons such as perceived workforce mismatch, lower education levels, and poor public transit and infrastructure, including utility connection and reliability. Congress enacted opportunity zones to counter these investment barriers. 

Starting in 2018, in order to incentivize investment, businesses can qualify for certain tax benefits as part of the Qualified Opportunity Zone program. Under the law, investors obtain these tax benefits by creating a special type of entity called an Opportunity Zone Fund. The fund has to have 90% of its assets in an opportunity zone in order to remain certified. To receive tax benefits, the fund has to invest enough capital to “substantially improve” the property within the first 30 months of investment. Until the end of 2026, “substantial improvement” for all opportunity zones means investing at least 100% of the purchase price back into the property. For example, a fund purchasing a $1 million parcel for development would need to invest $1 million or more into the opportunity zone within 30 months to qualify for opportunity zone tax benefits. 

If the fund meets these investment and certification requirements, they are then eligible for tax benefits:

• Five-year deferral of capital gains reinvested in the qualified opportunity fund.
• Elimination of capital gains tax on 10% of the deferred capital gain if the investment is held for 5 years or more (through step up in basis).
• Elimination of capital gains tax (on up to 30 years of gains) if the investment is held for 10 years or more (through step up in basis).

These tax benefits were then updated last year as part of OBBBA. The policy changes are set to take effect in 2027 and expected to reduce federal revenues by $40.9 billion over the next decade.⁷

In OBBBA, the newly minted “Qualified Rural Opportunity Zone Fund” was established, which allows opportunity zones in rural areas to qualify for tax benefits at a lower level of investment. The change takes the existing tracts already eligible for the opportunity zones, designates them as rural or urban, and lowers the necessary investment threshold to qualify in rural tracts. The median income qualifications are the same, but qualified opportunity zone (OZ) funds which have their investments in rural areas (a census tract that is in or next to a town under 50,000 people) have a lower investment threshold to receive benefits on the entire investment.

To understand how the rural and urban tax treatments differ, imagine two potential development parcels — one in an urban opportunity zone and one in a rural opportunity zone. Each parcel could be purchased for the same amount, say $1 million, and could house a proposed new data center. To qualify for the tax benefits in the urban area, the data center investor would need to make “substantial improvements”, meaning they need to add capital improvements of at least $1 million to the property (100% of the purchase price) within 30 months. However, in the rural area, they would only need to invest $500 thousand in the property (50% of the purchase price) within the same time in order to meet the “substantial improvement” threshold. This makes the rural parcel considerably easier to qualify, even though the underlying tax benefits available once qualified are identical.

Prior to this change, 93% of OZ investment went to urban areas.⁸ It’s worth noting that undeveloped lands (Ie. farmland) in opportunity zones will always have a distinct advantage because there is no substantial improvement requirement for them to qualify for these tax advantages. The Treasury Department acknowledged this difference in a set of regulations in 2019.⁹ The OBBBA change is intended to give an even greater incentive for investments in improved-upon rural land. This legislative emphasis, which is now permanently in statute, has implications for where a range of investments will be located including – and notably – for data centers.

New data centers are being proposed in Rural Opportunity Zones

The theory of the opportunity zones is that capital investment in these rural tracts will boost the local economy by contributing to local tax revenues, providing jobs for local residents, and indirectly uplifting other businesses. But it is unclear if data centers have these positive impacts in rural areas, or rather negative or mixed impacts.¹⁰ What is clear, however, is that data center developers are receiving guidance on how to leverage these new rural opportunity zone incentives.¹¹ As we noted previously, data centers were explicitly identified by the House Ways and Means Committee as possible investment opportunities.

Available evidence on proposed data centers suggests that developers may be taking this advice. In the April Pew study discussed above, researchers analyzed private data from Data Center Map and found that “most planned data centers” (67%) are in rural areas – a much larger share than the 13% rural share of data centers currently operating.¹²

Furthermore, available public data indicates that dozens of data centers are already being proposed in census tracts eligible to become rural opportunity zones, which constitute around 10% of all census tracts.¹³

We combined data from Compute Atlas, an open data source on data center location and capacity, with Treasury data on opportunity zone eligibility to determine where proposed projects are being placed.¹⁴ Compute Atlas has aggregated information on data centers from municipal and county portals, securities filings, and utility documents.¹⁵ The data provides some context as to planned data center build out in rural areas and potential rural opportunity zones in particular.

58 of the 328 reported or confirmed “proposed” data centers in the Compute Atlas collection are in census tracts that will be eligible for rural opportunity zone status. Of these 58 data centers, 43 (74%) were publicly proposed or announced after March 2025, when the House began deliberations on what would become OBBBA, and 39 (67%) were proposed after OBBBA passed on July 4.

It is important to note that the tax code may not be the only potential driver of the data center migration to rural areas. The land is often cheaper – and more abundant – in rural areas, and the cooler climate in some of these regions can substitute for cooling-related energy use.¹⁶ And unlike data centers used to support consumer-facing services, data centers built specifically to train new AI models do not need to be near people.¹⁷ It is possible that some proposed data centers in census tracts eligible to become rural opportunity zones will not receive the associated tax benefits, and are being placed there for another reason. The limited data makes it difficult to draw a clear conclusion.

Regardless of the primary driver, it’s reasonable to assume that companies will increasingly look to take advantage of these tax opportunities.

Deciding on a desired outcome

Whether or not these data center developments in rural areas are good or bad (or even neutral) depends on the desired outcome. On one hand, this could be an indication that opportunity zones, particularly the rural designation, are having the intended impact on investment.¹⁸ On the other hand, the stated reason for the opportunity zones was to “spur economic growth and job creation in low-income communities while providing tax benefits to investors.”¹⁹ It’s not clear if data centers are leading to these positive economic impacts.

Lawmakers will need to determine what result they want, and respond accordingly.

One option is to leave the rural opportunity zone designation as-constructed. If politicians’ main concern is moving data centers away from population centers – to where fewer people live – then the draw of rural opportunity zones is a positive. Lawmakers would still have the potential to reduce tax benefits for data centers by leaning on other policy levers, but should be careful to maintain consistent tax treatment across entities. For example, lawmakers could tighten benefits going to data centers by limiting the types of overall business spending that qualifies for tax opportunities (such as what investments qualify for bonus depreciation).²⁰ Likewise, broad reforms could improve the rules and treatment around vertical integration or corporate taxation.

However, policymakers may decide that the data centers are not delivering the intended benefits of opportunity zones to these rural areas – and the qualifying investment standards for all business development should be raised. Billions of tax dollars could be going to development that is not providing direct benefits to the local community and therefore should not receive this special type of tax treatment. For example, the exemption for unimproved land makes the administration of the zone much simpler, but it also means investors are incentivized to build on ‘greenfield’ (unimproved) land rather than improving previously built land, like unused warehouses.²¹ Furthermore, the proposed data centers may have been sited in these areas regardless of the incentives, making this an inefficient use of tax benefits.

Likewise, permanent employment directly resulting from data centers is fairly low, opening between a few dozen and a few hundred jobs per facility.²² Meanwhile, temporary construction jobs and indirect permanent jobs in sectors such as professional and business services generally go to specialized workers from outside areas.²³

There are also broader energy implications of data center build out in rural areas. Available interconnection in rural areas does not necessarily mean that these areas are best suited for the high levels of continuous power demanded by data centers. Increased energy demand from data centers may pose an increased risk of failures or price increases for consumers in rural areas if there aren’t proper upgrades to account for this large load.

Conclusion

Federal lawmakers must be clear-eyed about how rural opportunity zone tax credits may be used. With insufficient data on where data centers are going and how they precisely impact communities, it is reasonable to question if data centers should be able to leverage tax incentives like opportunity zones. But narrow bills carving data centers out of these tax benefits are not the answer – such reforms would be easy to avoid and could cause even more behavioral distortion.

If policymakers wish to minimize the incentives for data centers to migrate to rural areas, they could return the threshold to qualify for rural opportunity zone tax benefits back to its prior level, equalizing it with urban zones — although this goes against the intended purpose of creating the rural-specific opportunity zones and increasing investment in rural areas. An alternative reform option is to revise the requirements for opportunity zone benefits and extend them beyond just capital investment. This would ensure that opportunity zone investments are providing benefits to the local community rather than just lowering their federal tax payment.

In contrast, if policymakers want to encourage data center build out away from the mass public – knowing the potential downsides – then they should consider reforms other than tightening rural opportunity zone restrictions. Other tax benefits and non-tax practices used by data centers – bonus depreciation, vertical integration, the use of tax preferred trusts, bonds, or special purpose vehicles – may be better targets for reform. Broad reforms are favorable to data center specific carveouts to minimize avoidance.

No matter which route is taken, though, lawmakers should not lose focus on other, long-needed and more consequential reforms. These opportunity zone considerations are relevant, but incapable of fixing the bigger problems with our tax code. Broad corporate tax reforms are critical to raising more federal revenue, limiting tax manipulation, and subsequently enacting program expansions that benefit the broader public.

Appendix

Methodology

Data center locations are sourced from Compute Atlas, an openly available dataset of datacenter buildout.²⁴ Records were accessed from version 1.31.0 on September 1, 2026. All records with “confirmed” or “reported” confidence per Compute Atlas’ classification were retained in the sample, while “rumoured” records were dropped due to low verifiability. Full details of Compute Atlas’ collection and verification process can be found in their methodology.²⁵

The recency of data center proposals was confirmed by hand by the authors using publicly available information and verified with Claude.

Census tracts were designated as income-eligible for opportunity zone nomination using the Department of the Treasury’s “Eligible LICs for Nomination as 2027 QOZs” dataset.²⁶ This dataset identifies tracts that meet the low-income-community eligibility criteria under 26 U.S.C. § 1400Z-1(c)(1) median family income and/or poverty-rate thresholds, but does not reflect current or future opportunity zone designation. It also identifies tracts that are entirely rural. Accordingly, our analysis identifies data centers located in eligible low-income and rural tracts, not necessarily those located in tracts that have been or will be ultimately designated as opportunity zones.

Of 1,351 facilities in the Compute Atlas dataset, 1,349 were successfully matched to a census tract using the Census Bureau’s Geocoding API and each facility’s latitude and longitude; Two facilities were excluded because the census tract ID was not in the Treasury data. Two separate facilities had incorrect geocoordinate data which was hand corrected. Nine further records were excluded for being classified as “rumored,” resulting in a final dataset size of 1,340. Where precise facility-level coordinates were unavailable, locations were instead coded to the centroid of the parcel or town (see Limitations).

Summary of the data

We are confident that dozens of data centers are being proposed in these ROZ tracts, were announced after OBBBA was passed, and are primed to take advantage of the tax preference. Limitations of the dataset led us to avoid making even stronger statements, even though the numbers indicate two potential developments.

First, there is a difference between the proportion of tracts that are eligible to become rural opportunity zones and the proportion of proposed data center projects in rural opportunity zone eligible tracts. About 9% are eligible, but 17% of proposed projects are in these eligible tracts. We did not include this comparison in the body due to data confidence regarding operational data centers.

We also note the difference between where operational data centers are sited and where proposed data centers are sited: 7% of operational data centers are sited in rural opportunity zone eligible tracts versus 17% of proposed data centers. This pattern carries over to the permitted and under construction data centers as well. We did not include this comparison in the body due to data confidence regarding operational data centers. 

Source: Kubiak, E. (2026). Compute Atlas (Version 1.31.0).

Limitations

The data set from Compute Atlas is not assumed to be a perfect or comprehensive record of data centers in the United States. Compute Atlas relies on public filings, permits, and reported disclosures, which may undercount privately developed or undisclosed facilities. Verification confidence varies by record, with 541 of 1340 records “confirmed,” meaning there are multiple sources or an explicit announcement confirming the project, and the other 799 “reported,” meaning there is at least one filing or credible outlet rather than multiple independent sources of reporting. Geocoding accuracy varies by record, with 584 of 1340 records geocoded to the centroid of the most precisely verifiable parcel or town rather than to an address. We present our findings with these caveats in mind.

It should also be noted that there is a clear deficiency in the publicly-available data regarding where data centers are located and where they are being built. Researchers must either pay to acquire private data sets or use public sources with varying levels of recency and accuracy. This analysis uses data from Compute Atlas, one of the datasets used in the Brockovitch Data Center Reporting initiative, which appears to be the most robustly constructed, verified, and publicly available dataset. Still, the 1,300 data centers in that data set pales in comparison to the more than 3,000 available from paid sources such as Data Center Map and Cleanview. This could be because the Compute Atlas relies on public data, and ongoing data center business activity may not generate an announcement or public document, like a new groundbreaking or utility connection request, that would be openly findable. The shortfalls of this data highlight a much larger issue with the availability of public information on where data centers are located. Robust data is important for accurately determining the impact new and existing data centers are truly having on their communities. The current lack of accurate public data puts unnecessary barriers between accurate policy insights and the public.