Florida has made it harder for hyperscale AI data centers to offload water and electricity costs onto residents, forcing a new calculus for founders planning physical AI infrastructure.
Gov. Ron DeSantis signed SB 484 into law in early May, creating a regulatory framework that shields utility customers from subsidizing the power and water needs of large data centers and preserves local control over siting decisions, while still allowing limited confidentiality during negotiations with developers, according to the bill text and contemporaneous reporting. As the law's language and press accounts make clear, the state's aim is to ensure large-load customers pay the full cost of service and not shift those costs to the general body of ratepayers.
SB 484 requires the Florida Public Service Commission to adopt tariffs and service requirements so that big data center customers shoulder the incremental costs for connections, transmission upgrades and other grid expenses, rather than spreading them across residential bills, the statute and news coverage explain. The law also tasks water management districts with tighter scrutiny of permits, including refusing permits when a proposed center's water use would harm local resources, and it preserves local governments' authority over comprehensive planning and land development rules for data center placement.
At the same time the law curtails secrecy: it bars state and local officials from entering nondisclosure agreements that hide projects from the public, while permitting municipalities and companies to keep exact site details confidential for up to 12 months during active negotiations, a compromise that was reported across local outlets.
Why founders building AI should care
This is not a narrow zoning tweak, it is precedent. Florida's statute explicitly links data-center approval to utility-cost allocation and water-permitting rules, which means startups planning on-premises clusters, private cloud buildouts or colocations need to treat local politics and utility economics as first-order risks when sizing and siting compute capacity. For founders that assumed infrastructure is purely an engineering and cost problem, Florida's approach adds a legal and civic layer: hostile or cautious municipalities can now block projects or make them materially more expensive by insisting developers internalize grid and water costs.
Practically, that shifts the startup checklist. Teams must now model not just power and cooling costs, but also potential tariff charges, transmission upgrade contributions, permitting time frames, and the possibility that a county will refuse a project on environmental or ratepayer-protection grounds. That affects total cost of ownership, fundraising narratives, and go-to-market timelines for compute-heavy products.
Where founders can look for opportunities
Regulation creates demand. If utilities and local governments insist that hyperscale loads internalize costs, founders can sell solutions that reduce that burden. Startups that deliver ultra-efficient GPUs and accelerators, liquid-immersion cooling systems that cut power draw, or closed-loop water systems that dramatically lower consumptive use will be in a stronger bargaining position with permitting authorities and utilities.
Likewise, software-level efficiency tools, including scheduler and model-compression services that pare peak demand, become practical compliance levers. Local governments will prefer proposals that demonstrate lower peak loads and minimal water withdrawals, turning efficiency into both a selling point and a regulatory requirement for site approval.
How to change your infrastructure roadmap
Founders should treat siting as a cross-functional problem involving policy, legal, engineering and community relations teams, not only real estate and ops. That means running scenario analyses that include potential tariff passthroughs, buy-in requirements for grid upgrades, and alternative designs that reduce water use or shift cooling to reclaimed water where permitted.
Negotiate early with utilities and water managers, and build contingency plans that include smaller, distributed clusters, hybrid colocation strategies, or purchases of capacity from suppliers that already internalize infrastructure costs. Those approaches reduce exposure to a single jurisdiction's political shifts and may improve time to market while keeping regulatory friction low.
Finally, document community benefits: local hiring, tax revenue, co-investments in grid resilience, or commitments to reclaimed water use can tilt local political calculus in favor of a project. Florida's law preserves local authority, which means community-facing concessions still matter.
Florida's move will be watched closely. If other states adopt similar rules, startups that baked in efficiency and community-centered siting will have a competitive edge, while those that did not will face longer approval cycles and higher effective infrastructure costs. For founders building AI products, the lesson is simple: the physical layer is now as much a political and regulatory design problem as it is a technical one.
Also read: Utah's 'Stratos' megadatacenter exposes the physical limits of AI compute • Predatory pipelines are turning high schoolers into paid perpetrators of fake ML research • Meta's mouse-tracking revolt is a warning for founders about surveillance and AI in the workplace
