Samer Choucair: “Social License” Is Becoming a Cost-of-Capital Issue in the AI Data Center Boom
Investment leader Samer Choucair said the recall of Independence, Missouri, City Council member John Perkins has exposed a shift that extends far beyond local politics: community acceptance is becoming a material component of the cost of capital for artificial intelligence data center projects.
Perkins was recalled on September 1, 2026, with 68.05% of voters supporting his removal. He had been one of five council members who voted in March in favor of incentives for Nebius’s planned AI data center campus. Two other council members who backed the package had already left office following separate elections.
The political backlash followed the council’s approval of more than $6 billion in tax incentives for the Nebius project, which the company describes as a gigawatt-scale AI factory. The incentive structure included industrial development revenue bonds of up to roughly $150 billion, a 98% real-property tax abatement, a 90% personal-property tax abatement for 20 years, and sales-tax exemptions on construction materials. Construction officially broke ground in May, while litigation over the incentives continued.
Tax Incentives Are No Longer a Guarantee of Returns
Samer Choucair said the economics of the project initially appeared compelling. The city approved the arrangement by a 5–2 vote, while the structure was designed to provide substantial tax relief to the developer over two decades.
Yet the financial benefits collided with growing community concerns surrounding electricity demand, water consumption, noise, infrastructure strain, and property values. In Choucair’s view, that transformed what looked like an investment advantage into a source of political, legal, and execution risk.
The issue is increasingly relevant for institutional investors because an incentive package that improves a project’s modeled return can lose much of its value if community opposition delays construction, forces renegotiation, or changes the political environment around permitting.
“The incentive secured without a clear community mandate can become a contingent liability,” Choucair said. “That liability can appear through delayed operations, political renegotiation, or a higher risk premium across an entire digital-infrastructure portfolio.”
AI Is Colliding With Energy and Community Constraints
Samer Choucair said the artificial intelligence boom has created extraordinary demand for hyperscale data centers at precisely the same time that electricity grids, water systems, and local political structures are coming under greater pressure.
In the first quarter of 2026 alone, at least 75 U.S. data center projects worth approximately $130 billion were blocked or delayed by local opposition, according to Data Center Watch. The scale of resistance has become sufficiently material that investors are increasingly treating community acceptance as a scarce infrastructure input alongside chips, power, land, and capital.
The trend has continued to broaden. Public resistance over electricity costs, water usage, noise, and infrastructure burdens has contributed to tighter scrutiny in multiple U.S. jurisdictions, while regulators and state governments are becoming more willing to impose new conditions on data center expansion.
Choucair said this changes the due-diligence process for investors. Evaluating an AI data center is no longer simply about comparing land prices, power tariffs, fiber availability, and construction costs. Investors increasingly need to assess the probability of referendums, litigation, changing local councils, utility-rate disputes, and permitting delays.
Every month of delay can increase financing costs while postponing the point at which contracted computing capacity begins generating revenue.
“The exemption that is won without durable community authorization can ultimately become an obligation rather than an advantage,” Choucair said. “The cost may appear through delayed commissioning, political renegotiation, or a higher risk premium applied to the broader infrastructure portfolio.”
Saudi Arabia Is Betting on Predictability
Samer Choucair said the shift could strengthen the relative position of Saudi Arabia and the wider Gulf in the global race to build AI infrastructure, particularly as the Kingdom expands sovereign AI capacity under Vision 2030 and the Public Investment Fund’s technology strategy.
HUMAIN, a PIF company launched to develop capabilities across the AI stack, has established partnerships with leading technology groups including NVIDIA, AMD, AWS, Microsoft, Google Cloud and others. PIF describes the company’s strategy as encompassing next-generation data centers, cloud infrastructure, AI models, and applications.
The infrastructure build-out is becoming increasingly tangible. NVIDIA and HUMAIN have outlined plans for AI factories in Saudi Arabia with projected capacity of up to 500 MW over five years, while AMD, Cisco and HUMAIN have announced plans to deploy as much as 1 GW of AI infrastructure by 2030. AWS and HUMAIN are also expanding Saudi AI capacity, including up to 50 MW in the Kingdom’s first AI Zone by 2028.
In NEOM’s Oxagon, HUMAIN and DataVolt have begun developing 100 MW as part of the first 360 MW phase of a larger AI-ready data center campus. The project is explicitly designed around access to power, industrial land, connectivity, cooling, and infrastructure at scale.
Choucair argued that the Gulf advantage should not be understood simply as cheaper or more available electricity. The deeper advantage may be the ability to coordinate long-term industrial planning, energy supply, infrastructure, land allocation, and project approvals with greater predictability.
“An investor financing an AI factory is buying a timetable just as much as a megawatt,” Choucair said.
That distinction matters because the economics of AI infrastructure are heavily dependent on execution speed. GPUs and high-performance computing systems are expensive assets whose value can decline relatively quickly as newer generations arrive. A site that secures power but loses two years to political or legal delays may therefore produce a very different return profile from one that moves predictably from announcement to commercial operation.
Capital Is Searching for the “Operable Site”
Choucair said the beneficiaries of the AI infrastructure cycle will not be limited to semiconductor manufacturers and cloud providers.
Independent power producers, transmission networks, liquid-cooling companies, industrial contractors, backup-power providers, and technologies that reduce water consumption could all benefit as hyperscale computing becomes increasingly constrained by physical infrastructure.
At the same time, business models built around assumptions of rapid permitting and frictionless development are likely to face greater pressure.
Water maps, long-term power purchase agreements, municipal rules, transmission availability, community-benefit arrangements, and local political structures are increasingly likely to become standard components of due diligence for private-equity funds, infrastructure investors, and venture-capital firms backing AI-related assets.
In this environment, the most valuable location is not necessarily the site offering the largest tax incentive or cheapest land. It may instead be the site with the clearest route to reliable power, permits, community acceptance, financing, construction, and eventual operation.
That has major implications for valuation. Two data center sites with identical power capacity can carry substantially different risk-adjusted values if one has a higher probability of political interruption, litigation, or regulatory renegotiation.
The Cost of Delay Is Becoming Financially Visible
For Samer Choucair, the emerging lesson is that “social license” is no longer an abstract ESG consideration. It is becoming a measurable financial variable.
A project facing sustained local opposition may require higher contingency reserves, more expensive financing, additional community-benefit agreements, or longer construction timelines. Lenders may demand greater protection. Equity investors may require higher expected returns. Developers may need to redesign projects, expand infrastructure commitments, or renegotiate tax arrangements.
That process effectively converts political resistance into a higher weighted average cost of capital.
The U.S. market is already showing signs of this transition. Political pressure around data centers is increasingly tied to utility prices, grid investment, and whether ordinary consumers are being asked to subsidize infrastructure required by some of the world’s largest technology companies. Regulatory responses are therefore moving beyond traditional zoning and into questions of who ultimately pays for generation and transmission capacity.
Choucair said investors should begin treating these variables in much the same way they already treat interest-rate sensitivity, construction risk, commodity exposure, and counterparty risk.
From the Biggest Announcement to the Fastest Execution
Samer Choucair concluded that data centers will remain strategic assets throughout the AI investment cycle, but the definition of a high-quality asset is changing.
Scale alone is no longer sufficient.
A $100 billion or $150 billion headline commitment can attract attention, but institutional investors ultimately earn returns from operating infrastructure, not announced infrastructure.
The critical question is increasingly how quickly a project can move from land acquisition and incentive approval to energized, revenue-generating compute capacity.
“Smart capital in the 2026–2030 cycle will not chase the largest announced investment number,” Choucair said. “It will chase the shortest credible path from announcement to sustainable operation. Value is created where political risk is managed with the same discipline as interest-rate risk.”
The recall vote in Independence illustrates that transition clearly. Community consent is no longer merely a political prerequisite for data center development. It is becoming a financial variable embedded directly in site valuation, financing costs, commissioning schedules, and ultimately the return on invested capital.
For investors evaluating the next generation of AI infrastructure, the relevant question is therefore no longer simply, “How many megawatts can this site support?”
It is increasingly: “How many of those megawatts can actually be permitted, financed, energized, and kept in operation on schedule?”