Kansas’s contest for governor has exposed a disagreement that a data-center tax incentive left unresolved: who must prove that a large computing campus is safe for the community before development advances? Democrat Cindy Holscher proposes a temporary pause while stronger protections become law. Republican Ty Masterson says existing utility regulation and local decisions can protect residents without stopping suitable projects.
Both nominees voted for the 2025 incentive they now debate. Holscher says concerns raised during roughly 80 town halls changed her view; Masterson calls the reversal politically convenient. Neither position can be reduced to supporting or opposing AI. The dispute concerns the order in which Kansas offers benefits, secures commitments and obtains public consent.
A twenty-year benefit with conditions—and unresolved costs
Senate Bill 98 created a 20-year state and local sales-tax exemption for qualifying data centers. The Kansas Commerce program requires an investment commitment of at least $250 million alongside employment and operating conditions, long-term power-purchase planning, water-conservation planning and security review. Utilities cannot offer eligible facilities discounted economic-development electricity rates.
The Kansas Intelligence Fusion Center must examine equipment and software for threats to critical infrastructure before public benefits are awarded. That addresses foreign influence and cyber risk; it does not replace environmental review, utility oversight or local participation. These are genuine conditions, but not one comprehensive siting law.
The enacted legislation also carries a transparency controversy. SB 98 began as a Route 66 license-plate bill before legislators substituted data-center provisions. That procedure is allowed, but critics question whether a two-decade commitment received adequate public scrutiny.
An earlier fiscal analysis estimated $9.8 million in forgone sales tax for a $250 million project under comparable assumptions. The final law changed, and actual cost depends on qualifying projects: this is not a final bill for every campus. It nevertheless illustrates the opportunity cost of attracting investment through a long exemption.
Local disputes arrived before a statewide settlement
Emporia approved a Digital Infrastructure Overlay District in July 2026. It established zoning rules for potential infrastructure, not approval to construct an actual data center. Opponents pursued an initiative to ban high-impact facilities; the city sought judicial guidance on the petition’s legal status. A zoning district, a proposed ban and a construction permit must not be treated as the same decision.
Lawrence adopted a temporary pause while officials studied rules for large facilities. In Independence, voters recalled councilmember John Perkins by 68% to 32% after he supported a proposed Nebius development. Separately, another project supporter lost a mayoral contest after an opponent challenged the approval process and sought a public vote. Those results demonstrate local political exposure, not statewide consensus or a forecast of the governor’s race.
The underlying problem is when disclosure becomes meaningful. Residents may learn a zoning category before knowing the operator, cooling design, final water demand, computing load or construction schedule. Confidential commercial negotiations are not automatically misconduct, but asking a community to trust incomplete estimates can undermine legitimacy even when formal procedures are followed.
Smaller governments also enter negotiations with unequal expertise and bargaining power. A city assessing land use may simultaneously need to evaluate utility finance, water, cybersecurity and long-term contractual commitments. An approval can be legally valid without producing durable consent throughout construction and operation.
What Evergy’s large-load tariff does—and does not do
In November 2025, the Kansas Corporation Commission approved Evergy’s Large Load Power Service for customers requiring more than 75 megawatts of peak demand. The regulator’s announcement describes demand and energy charges designed to recover the incremental cost of serving those customers and prevent existing customers from subsidizing new large loads.
Participants included Evergy, Google, the Data Center Coalition, consumer representatives, environmental organizations, industrial customers and school districts. That breadth matters: the arrangement is more than an operator’s unsupported promise. Its intended protection must still be tested against actual contracts and investments rather than assumed to have eliminated every risk.
Electricity billing does not automatically settle water limits, drought restrictions, zoning compatibility, land restoration or disclosure. Nor does a paid monthly bill prove that all generation, transmission and distribution commitments are secured. Utilities plan years ahead; if a campus is delayed, canceled or changes its expected demand, regulators must determine who remains liable.
Employment and water claims need similar precision. Temporary construction jobs are different from permanent operating positions. Water consumption varies with cooling, weather and reuse, and direct facility consumption differs from water associated with electricity generation. A conservation plan alone does not tell residents which limits or remedies apply during drought.
Two governing models, not two technology slogans
Holscher proposes stopping new large corporate data centers temporarily until stronger statutory protections cover utility rates, water, transparency, local participation and Kansas jobs. She says development could proceed where communities support it and enforceable protections address their concerns. This is a conditional pause, not a permanent ban on all computing infrastructure.
Masterson favors keeping siting decisions primarily local within existing state and utility rules. At an August forum he said large users should meet their energy needs and avoid inappropriate use of water or land. He also links domestic AI capacity to competition with China. That national argument does not itself answer who pays for local infrastructure, but neither is his position simply an offer of unregulated development.
Their strongest arguments expose each other’s weaknesses. A pause could give communities common standards and better information, yet become indefinite if its scope and exit conditions are unclear. Distributed regulation can preserve flexibility and local choice, yet leave gaps between agencies and private agreements. Smaller cities may not be equipped to negotiate those gaps alone.
A governor can propose legislation, direct executive agencies, shape development-policy evaluations and appoint regulators when vacancies arise. The office cannot simply override every local zoning decision. If Holscher wins, Republican legislative majorities could constrain or reshape her proposal; if Masterson wins, existing opposition would still put pressure on lawmakers to demonstrate that protections work.
A national AI buildout meets local accountability
The Kansas dispute reflects a geographic mismatch: AI’s benefits can reach national users while substations, transmission lines, backup generators, water systems, traffic and land conversion concentrate near one community. Gallup found roughly seven in ten Americans opposed a local AI data center; another national survey reported 58% expected new construction to harm the country. These are different opinion measures, not Kansas-specific impact assessments.
Build American AI, linked to a pro-AI political network, began an effort focused initially on Kansas, Ohio and Wisconsin with a reported roughly $50 million for advertising, research, public engagement and related work. That is not a Kansas-only spending figure or proof the entire budget was spent. Its combination of national leadership and family protections shows the industry responding to local cost objections.
Republicans elsewhere, including Texas, have sought stronger cost controls, while Democrats differ over moratoria and investment. The relevant divide is therefore not reliably partisan. Scale and sequencing matter: a small server facility and a campus seeking hundreds of megawatts create different obligations.
The November 3 election will provide evidence about voter response, not proof that one issue caused the result. More decisive policy tests will be whether pause proposals specify covered facilities and measurable exit rules, whether hearings precede binding commitments, and whether contracts secure canceled-load costs, drought responses, decommissioning and reporting. Public summaries can expose performance without disclosing protected commercial information.
The Kansas Reflector account of the race makes those administrative questions concrete. AI capacity depends on physical infrastructure and credible agreements, not chips and capital alone. Both candidates must show that promised safeguards remain enforceable after the election and beyond the initial construction boom.
Original reporting aggregation entry—not a direct publisher link.