Bernstein: The key obstacle to data center and energy construction is not “resistance to change,” but fear of real losses

Overview

This is a thematic study on data center construction, the energy transition, and community opposition. Bernstein argues that the root of the dispute is not a generalized rejection of change, but rather concerns among different groups about losing existing benefits, control, or security across different time horizons; consequently, AI data center construction will continue, but conditions for a consensus that enables rapid capital reallocation are currently absent.

Core views

The report begins with Kahneman and Tversky’s prospect theory, arguing that what people truly resist is “loss,” while change is merely the vehicle through which loss usually arrives. Prospect theory holds that people evaluate gains and losses relative to their current state as a reference point; broadly speaking, the pain from an equivalent loss is about twice the pleasure from an equivalent gain. The report illustrates this with two choices: when faced with a certain gain of $100 versus a 50% chance of gaining $250, people typically choose the former even though the latter has an expected value of $125; when faced with a certain loss of $100 versus a 50% chance of losing $250, people often choose the latter to avoid a certain loss. The resulting status quo bias means that change is first perceived as a potential loss before it is understood as a potential gain. Bernstein summarizes energy policy and capital flows around four forces: affordability, reliability, sustainability, and national security. Most of the time, all four can coexist; but when one is threatened, abstract objectives become urgent, and capital and regulatory resources flow toward the area under greatest pressure. The report uses history to illustrate this mechanism: the oil crises of the 1970s primarily created a security threat, followed by the Strategic Petroleum Reserve, CAFE standards, and nuclear power construction; in the 2000s and 2010s, sustainability relied more on top-down mechanisms such as renewable portfolio standards and subsidies because the associated losses were slow-moving and dispersed, making it difficult to generate political urgency organically; when Russia reduced gas supplies to Europe in 2022, reliability and security were simultaneously impaired, and capital reallocation toward LNG import terminals, coal-power restarts, and nuclear power occurred within months, but it also required governments to rapidly reverse years of climate commitments and created new perceptions of loss. The report believes that AI data center construction currently threatens both affordability and reliability as perceived by consumers. Surcharges tied to incremental data center load have already increased electricity prices, while interconnection queues and capacity constraints are viewed by consumers as risks to supply reliability even before actual outages occur. More broadly, some consumers see the AI race as a national-security agenda in which they are compelled to participate without consent, while also worrying that AI may cause job losses. The dominant federal narrative is that falling behind China in AI competition constitutes an existential national-security threat and can therefore override local rate and permitting disputes; however, communities adjacent to substations and project sites face the direct losses of construction and operational noise, potentially higher bills, and changes to local life. The report does not consider either side irrational, but argues that both are responding to loss aversion, merely anchored to losses of different scope and on different clocks, making their arguments difficult to reconcile. For electricity affordability, the report adopts a “share of wallet” perspective. According to EIA data, the average U.S. household uses about 10,500 kWh annually, or roughly 875 kWh per month; at an average U.S. retail electricity price of approximately $0.17/kWh, this equates to a monthly bill of about $150 and an annual bill of about $1,800, representing 2.3% of average annual household spending of $78,535. If retail electricity prices rise by 20%, the monthly bill would increase to about $180. For reference, housing accounts for roughly 33% of household spending, transportation about 17%, and food about 13%. The report does not use this to judge whether higher electricity prices have a material impact on consumers; rather, it notes that electricity is a relatively small component of total spending, and consumers’ strong reactions also reflect a loss of control and being drawn into a race they did not actively choose. At the same time, the report acknowledges that price increases in some regions are far higher than in others. The report distinguishes between opposition that can be mitigated through economic arrangements and opposition that is difficult to monetize. Many data center siting disputes concern affordability and reliability, which can in principle be negotiated through better rate structures, community benefit agreements, or closed-loop water systems. By contrast, a mother and daughter in Kentucky rejecting a $26 million offer for their 1,200-acre farm, which had been in the family for 200 years and four generations, or roughly $48,000 per acre, illustrates that some opposition is defending not against financial loss but against the loss of values and heritage, and thus may not be resolvable through compensation. Bernstein believes historical disputes have often been resolved quickly only when the different forces ultimately pointed to the same type of loss. Such convergence has not yet emerged in current data center construction: the federal government is responding to security threats that residents may not directly experience in daily life, while residents are responding to electricity-price, reliability, and community risks that corporate narratives have not adequately addressed. Unless AI competitiveness becomes a household-level concern, or electricity-price and reliability pressures escalate into a nationwide issue, this is more likely to be a slowly draining debate than the rapid capital rotation seen when all four forces align. The report also emphasizes that this does not mean construction will not occur; from rural transmission lines to interstate transmission networks, most contested energy infrastructure is ultimately still built.

Analysis framework

The report first uses loss aversion and status quo bias from behavioral economics to explain decision-making mechanisms, then reviews historical cases of energy policy and capital allocation through four forces—affordability, reliability, sustainability, and national security—and finally applies this framework to electricity prices, interconnection, siting, and community disputes surrounding U.S. data center construction.

Methodology notes

• Event-Driven Games and Behavioral Finance

  Prospect Theory and Loss Aversion

  The report uses the asymmetry whereby the pain of losses exceeds the pleasure of equivalent gains to explain differing reactions by communities, consumers, and governments to data centers and the energy transition.

• Event-Driven Games and Behavioral Finance

  Status Quo Bias

  The report argues that change is first perceived as a potential loss, so established habits, community arrangements, and energy choices create resistance.

Key data

• Average annual electricity consumption of U.S. householdsApproximately 10,500 kWhApproximately 875 kWh/month; the report cites EIA data.
• Average U.S. retail electricity priceApproximately $0.17/kWhUsed to estimate average household electricity bills.
• Average household electricity billApproximately $150/month, $1,800/yearApproximately 2.3% of average annual household spending of $78,535.
• Monthly electricity bill after a 20% price increaseApproximately $180/monthIllustrative calculation in the report.
• Kentucky farm offer$26 millionApproximately 1,200 acres, or about $48,000/acre; the report uses this to illustrate that non-monetary losses may not be resolved through compensation.

Impact & implications

The report believes the core of resistance to data center construction lies in the inconsistency of losses perceived by different stakeholders. Affordability and reliability issues can be partly mitigated through rates and community arrangements, but losses involving local control, living environments, and heritage may not be negotiable; before national-security objectives and community concerns converge, projects may continue to face permitting, rate, and siting friction.

Risks

• Electricity-price increases associated with incremental data center load may intensify consumer affordability concerns.
• Interconnection queues and capacity constraints may be perceived by communities as risks to supply reliability.
• Local concerns over construction, operational noise, community changes, and loss of heritage may slow data center siting and construction.