13 U.S. States Move to Repeal Data Center Tax Incentives, Potentially Adding $3 Billion to AI Compute Costs per Gigawatt
Complete. Here is the key summaryAs of early August, four U.S. states have repealed or suspended data center tax incentives, with nine others considering similar measures. Assuming a 7% sales tax, the equipment procurement cost for a single 1GW AI data center would increase by approximately $3 billion. The policy shift in key markets like Texas, combined with constraints on power and water resources, is transforming tax incentives from a "default bonus" into a "hard cost" for AI infrastructure
U.S. states are launching a wave of repeals on data center tax incentives, a policy shift that is driving up the cost of AI infrastructure construction and creating new uncertainties for the large-scale expansion plans of tech giants.
On August 2, The Information reported that since this summer, four U.S. states have canceled or suspended data center tax incentives, while nine other states are studying similar policies. For companies relying on massive capital investment to build AI computing power, this means that tax benefits once taken for granted are gradually loosening.
The impact of these tax changes is significant. IT equipment represents the largest capital expenditure for data centers. Calculated at a 7% state sales tax, the equipment procurement cost for a single 1GW-level AI data center will rise from approximately $40 billion to $43 billion, an increase of billions of dollars.
Since servers and AI chips typically need updating every five years or so, sales taxes will continue to push up future operating and expansion costs, and have already begun affecting financing terms for some data center projects.
Bipartisan Shift: AI Boom Weakens the Case for Tax Incentives
The reasons behind the policy shift are straightforward.
In recent years, states have introduced preferential policies such as sales tax exemptions to compete for AI investments from tech companies like Microsoft, Google, Meta, Amazon, and Oracle. However, as AI investment continues to heat up, more state governments believe that these incentives are eroding fiscal revenue, and that tech companies will not easily abandon local investments even without tax breaks.
Meanwhile, the public opinion landscape is also changing. More voters are beginning to question the actual contributions of large data centers to electricity, water resources, and employment, demanding that tech enterprises bear the same tax burden as other businesses. This issue has gradually formed a bipartisan consensus.
Texas Becomes the Industry Focus
The market is currently most focused on Texas.
According to forecasts by real estate services firm JLL, Texas's total data center power capacity is expected to surpass Virginia's by 2030, making it the world's largest data center market. However, in June this year, Governor Greg Abbott asked the state legislature to study the repeal of sales tax exemptions for data centers and other "outdated or unnecessary" incentive policies.
This statement is particularly sensitive.
Currently, dozens of gigawatts of AI parks are under construction or announced in Texas, and the local grid has received interconnection applications for data centers totaling hundreds of gigawatts. Dan Diorio, Executive Vice President of the Data Center Coalition, recently traveled to Austin specifically to attend hearings, hoping to preserve the tax incentives implemented since 2013. The Texas Legislature is expected to revisit the relevant bills in 2027.
Diorio stated: "If these incentives suddenly disappear, your entire business plan will be disrupted."
More States Begin Recalculating the "Subsidy Bill"
In addition to Texas, several other states have begun taking action.
Washington State last month repealed the sales tax exemption for data center equipment upgrades and renovations; the state government expects this to generate approximately $207 million in additional fiscal revenue by 2029. Arizona has suspended the sales tax exemption for data centers for three years.
Louisiana did not directly cancel the incentives but raised the threshold for qualifying. Meta is currently building a data center in the state scalable to 5GW. Governor Jeff Landry requires data center companies to bear all additional power demand themselves to continue enjoying tax incentives.
Virginia adopted a more moderate approach. The state government ultimately did not cancel the sales tax exemption but instead imposed an electricity consumption tax on data centers. According to Diorio's estimates, this will add approximately $600 million in annual tax burden to the industry; if the sales tax exemption were directly canceled, the annual additional tax burden would exceed $1 billion.
AI Infrastructure Cost Models Are Changing
Faced with tightening policies across various states, the industry is seeking new compromise solutions.
Diorio stated that the industry is willing to accept more additional conditions, such as linking tax incentives to metrics like job creation and local investment. He believes that since 40 states nationwide provide tax exemptions for manufacturing capital equipment, data centers, as a capital-intensive industry, should receive similar treatment. Studies commissioned by the Data Center Coalition from PwC, as well as independent reports from the Virginia Legislative Audit Agency, all suggest that the tax and economic benefits created by data centers exceed the sales tax revenue forgone by the government.
Nicholas Miller, a policy researcher at the National Conference of State Legislatures, expects that more states will adopt approaches similar to Louisiana's in the future, attaching more conditions to retained tax incentives rather than directly canceling programs.
However, some states are beginning to take a harder stance. New York State recently passed a moratorium on data center development.
For tech giants still racing to build AI infrastructure, tax incentives are no longer policy bonuses that can be taken for granted. As states reassess the cost of subsidies, tax policy is becoming an important variable influencing AI data center site selection and investment returns, alongside power supply, transmission capacity, and water resources.
