Skip to Content

Don’t Be Fooled: The Truth About the Tax Implications of Data Centers

Data center engineers overlooking a bank of servers.

Bottom line up front: Data centers generate the highest positive net fiscal output of any real estate asset type. Period. Any other conclusion is simply incorrect. And despite reports stating otherwise, the tax revenue they generate justifies any number of tax incentives local jurisdictions use to entice developers to build them in their markets.

As an elected official, I evaluated the fiscal and economic impact of a wide variety of commercial and economic development projects. As a real estate professional, I have 20 years of expertise advising on state and local tax matters. And as a partner and co-founder of Oasis Digital Properties, a data center development company, I’m fully immersed in the economics of data centers. So, while I may personally and professionally benefit from more data center development, I am duly qualified to critique studies on this subject as well. So, I dug deep into the most frequently cited study on data center development and corresponding tax implications to understand why these developments have created such a firestorm and what is really being analyzed in that study.

Spoiler alert: My conclusion from reviewing this study in question, and others like it, is that the studies grossly misrepresent and omit data related to the larger fiscal picture surrounding data centers. Regardless, these studies are incomplete at best.

Unfortunately, several studies by think tanks have sparked a different narrative that has been circulating widely throughout mainstream media as of late. One study from December 2022 led to a rash legislative decision that threatened to choke off data center development and its local tax benefits in Georgia. The University of Georgia study concludes that the state will forgo more than $307 million in taxes between 2024 and 2030. As a result, the incentive is posited as a money loser for the state.

Off the heels of that study, Georgia state legislators attempted to force a two-year pause of the data center sales tax exemption. Fortunately, Governor Brian Kemp vetoed that effort. And rightfully so as that type of state-level tax exemption has become a fairly standard tax incentive across most states over the last decade. In general, the incentive exempts companies from paying state sales tax on the purchase of data center equipment, assuming those companies hit certain job creation and capital investment thresholds.

And while that exemption may receive the most press, it must be made clear that it is only one small part of the fiscal equation related to data centers. That study is now being used as a weapon of the broader anti-data center community to attempt to discredit the fact that data centers bring substantial fiscal benefits to local taxing jurisdictions.

However, these studies miss the nuances and many of the data inputs that are necessary ingredients of any worthwhile fiscal impact analysis associated with commercial and economic development initiatives or policy. Data centers are not built or taxed the same as other types of property and therefore warrant a more nuanced analysis. As a result of these incomplete studies, popular opinion is shifting due to a combination of misinformation and misunderstanding of those studies. The decisions now being made based on these studies are proving detrimental to the counties, cities, and towns whose lifeblood depends on capital investment by the real estate industry.

One concern when analyzing this type of exemption is how the nature of the exemption is framed. Most incentives via exemption are structured as simply forgoing the collection of taxes that “but for” that tax incentive wouldn’t be on the table to begin with. If you are abating taxes on something that hasn’t happened yet, you’re not “losing money.” You’re simply forgoing collecting certain taxes that didn’t exist already.

The Georgia study focuses on a state-level tax, but real estate investment and development benefit local taxing jurisdictions far more than the state tax coffers. States may offer state-level incentives to encourage development, but the most significant tax benefits real estate and data centers in particular create are disproportionately concentrated at the local level (i.e., counties, cities, and towns). 

A vast majority of the taxes collected by local governments to operate schools, maintain roads, fund police and fire departments, and every other local public service comes from taxing the value of real estate (AKA real property) and personal property (e.g., equipment, furniture, tools, etc.).

The valuation and taxable profiles of data centers are unlike any other type of taxable property. For one, a data center will cost on average $800-$1000 per square foot to build and average 100,000 square feet in size, with some individual buildings exceeding 500,000 square feet. And that’s just the physical structure or building taxed as real property. To put that into perspective, most well-appointed single-family homes, even in high-cost areas, can be built for less than $350 per square foot, and average 2,480 square feet.

Even so, local property taxation of data centers goes well beyond the steel and concrete used to construct the building. The land itself, if zoned for data center development, commands far higher land values than any other type of real estate, and that land value is also taxed as real property.

But perhaps the most striking benefit of the local taxation and subsequent revenue that data centers generate is what’s inside the data center. The equipment inside data centers costs more than $2,500 per square foot. Such equipment includes the CPUs and GPUs housed in racks that essentially make up the hardware running applications supporting our modern digital economy. While not all jurisdictions tax personal property, most do.

The taxes on real estate (buildings and land) are based on a respective taxing jurisdiction’s opinion of the value of a building and the land. The taxes on personal property are based on the original cost of that equipment (as reported by the taxpayer). And for data centers, the servers/computers are taxable as personal property. Overall, the value of the real and personal property in data centers far exceeds other typical types of real estate and personal property. Therefore, the taxes are almost always disproportionately higher than any other property type. 

And even though certain jurisdictions offer favorable tax rates and depreciation schedules for data center equipment, that still doesn’t negate the substantial taxes produced by that equipment. When any kind of personal property is taxed, the jurisdiction depreciates the reported value of that property based on its age. This means that the original cost you paid for that personal property isn’t taxed at 100% of that cost; it’s almost always depreciated by some amount to account for the value lost over time. That taxation structure applies to almost all personal property reported by a personal property taxpayer.

But here’s what is distinctly different about the equipment in data as personal property: Not only is the original value of that personal property vastly higher than other personal property types, but the data center equipment itself is also refreshed or replaced on average every 3-5 years. That means that the taxable value of that equipment also resets every 3-5 years.

Apples-to-apples, data center equipment drives significantly higher taxes as taxable personal property than other types of personal property. Almost anywhere that data center equipment is taxed, other types of personal property other businesses own are also taxed. That includes everything from machinery and tools to desks and chairs. Most office furniture and tools are not replaced every 3-5 years and they certainly don’t cost as much on a comparative basis to data center equipment.

A more comprehensive analysis would show that even with purported “losses” due to the state sales tax exemption, the direct local tax revenue produced by data centers is substantial, far outpacing any other type of real estate. So, while state taxes related to the purchase of data center equipment may be abated, the tax revenue produced by the real estate and equipment in data centers generates sizable tax revenue for local taxing jurisdictions. Accordingly, considering that the public service costs associated with supporting this type of real estate are minimal, the net fiscal output of data centers usually results in $15 of tax revenue for every $1 in public service support. Compare that to other real estate asset types such as mixed-use projects that generate on average only $3 in local tax revenue for every $1 in public service support. 

Ultimately, even with tax revenue aside, considering the breakneck speed of advancements in technology and AI, we need more data centers – not less. At the current pace of progress, if we do not keep pace with the present demands for data processing and storage by building more data centers, our march towards technological progress and our country’s position as a leader in technological advancement may come to a grinding halt.

The demand for data storage and processing is staggering. If our collective position is to stymie the development of data centers then we should all be willing to trade in our smartphones for a used Nokia 3310, cancel the Netflix account, unsubscribe from Amazon Prime, break up with our internet provider, and buy our next car from a vintage dealer who only sells vehicles built before 1980. That’s a non-starter for most, and a bygone era for many.

In conclusion, the data center industry needs more transparency and education around its fiscal and economic benefits. Looking at a single incentive or taxation component of data centers is not sufficient to measure the true cost and benefits of data centers. Whether you like it or not, data centers are here to stay. And as we’ve just discussed, the data is there to support any number of tax incentives to build them, which is a winning proposition for local economies throughout the United States.

BACK TO ARTICLES
Back to top