An effective energy procurement strategy has never been more critical as AI data centers expand across the United States and transform electricity demand. These facilities require enormous power to support high-performance computing, increasing stress on regional grids and reshaping price dynamics. In New England, where infrastructure and supply constraints already challenge stability, the rise of AI data centers could contribute to faster upward pressure on electricity prices.
Organizations that wait for the market to adjust risk facing higher energy costs and tighter procurement windows. A proactive approach can help maintain cost predictability and shield operations from unexpected volatility. This includes combining fixed-price hedges with flexible contract terms and pursuing energy efficiency measures that offset peak consumption. Such a strategy provides financial clarity while supporting sustainability goals and regulatory compliance.
Leaders must make informed decisions grounded in data, not speculation. Data-first energy procurement strategies track regional power trends, analyze market signals, and build models that translate industry shifts into measurable outcomes. With a clear understanding of where AI-driven demand is headed, decision-makers can act with confidence to lock in favorable rates today to protect tomorrow’s budgets.
The national demand for AI data centers
AI-driven data centers are emerging as some of the fastest-growing sources of electricity demand in the United States, with AI workloads adding intensity and unpredictability to traditional data center load profiles. As more computing shifts toward high-performance GPUs and large language model training, these facilities increasingly draw steady around-the-clock power while also creating sharp local demand spikes where clusters are concentrated. This shift changes how utilities and system operators think about capacity needs, grid reliability, and where to invest in new infrastructure.
An effective energy procurement strategy must account for this changing load shape, not just total consumption. AI-centric data centers are less “peaky” than some traditional loads, yet their high, sustained demand can drive capacity expansion and influence when peaker plants run, which affects overall price formation. In some cases, operators can pair AI workloads with demand response or flexible scheduling to support peak shaving, but the feasibility of these tactics depends on operational tolerance for shifting non‑critical workloads and on local market rules.
National vs regional effects
Nationally, data centers currently represent a modest share of total U.S. electricity use, but that share is projected to grow rapidly as AI adoption accelerates. Studies indicate that data center electricity demand could roughly double in the coming years, suggesting a higher baseline load for regional grids even after accounting for efficiency gains in hardware and facility design. Federal and academic analyses also highlight large uncertainties, with outcomes highly sensitive to efficiency improvements, AI deployment pathways, and policy decisions that influence generation and transmission build‑out.
Regional effects will not be uniform. States that already host significant data center clusters, such as Virginia and parts of Texas, are projected to see a much larger share of their total electricity demand tied to these facilities than the U.S. average, creating localized stress on infrastructure and potentially sharper price responses. New England faces a different set of constraints, including limited transmission import capability, older generation fleets, and strong clean energy policy targets, which can magnify the price impact of incremental load growth even if the absolute number of AI data centers is lower than in some other regions.
For executives, one critical implication is that prices respond to a combination of factors, not AI data center growth alone. Wholesale and retail rates reflect the balance between demand and supply, the pace of new generation and storage development, transmission bottlenecks, and policy signals that affect fuel mix and capacity markets. In practice, this means an energy procurement strategy should stress-test different scenarios for AI-driven load growth alongside variables such as gas prices, renewable build-out, and transmission upgrades, rather than assuming a single, linear path for future energy costs.
Price dynamics: AI-driven demand and electricity prices
AI-centered data centers are beginning to reshape wholesale electricity price formation in markets across the United States, especially where clusters of new facilities connect to already constrained grids. These sites add large, often continuous blocks of demand that participate in wholesale markets through standard demand bids, which can raise clearing prices in the hours and locations where supply is tight. When transmission into those areas is limited, increased flows toward data center hubs can increase congestion rents, widening price differences between nearby nodes and contributing to higher bills for customers who share the same grid.
An effective energy procurement strategy needs to account for how AI-driven demand interacts with scarcity pricing mechanisms in organized markets such as ISO and RTO regions. When system operators anticipate sustained load growth from data centers, they may clear higher prices during peak hours or activate scarcity pricing more often to attract additional generation or reduce demand, which can raise average costs over time. In capacity markets, higher forecasted demand and the need for new infrastructure can increase capacity prices, which then flow through to retail tariffs and contract structures.
Different AI growth paths
AI demand growth does not have a single trajectory, and that uncertainty matters for price risk. Under a steadier adoption path, data centers increase total load but give grid planners more time to add generation, storage, and transmission, which can moderate upward pressure on prices even as capacity needs rise. A rapid, nationwide expansion scenario, in contrast, compresses the timeline for new investment and can create more frequent periods where demand approaches available capacity, amplifying congestion and scarcity pricing effects in certain regions. In that higher-growth case, an energy procurement strategy that relies heavily on index exposure may see larger swings between low and high-cost periods, especially in markets already facing supply or transmission constraints.
Key price signals to monitor
“To manage this uncertainty, decision-makers should track several market indicators that reflect how AI data centers are influencing price dynamics over time,” said Justin Vissat, Kb3 Advisors managing partner. “Forward curves for power in key hubs provide an early view of how market participants expect sustained AI-related load to affect future prices and can guide decisions on when to lock in portions of load versus maintain index exposure.” Additionally, Vissat said day-ahead versus real-time price spreads reveal how well the system is handling forecast risk and operational variability; persistent, wide spreads can indicate growing stress on the grid in regions with heavy data center buildout.
Capacity prices and regional price dispersion, including nodal or zonal differences within a single ISO, offer additional insight into where constraints are binding and where AI-driven demand may be contributing to localized cost pressure. “Integrating these signals into an energy procurement strategy helps organizations respond to AI-related demand growth with informed timing, product mix choices, and risk limits rather than reactive, short-term decisions,” he said.
AI data center effects on New England energy prices
New England faces a distinctive set of risks as AI-driven electricity demand grows, and those risks matter directly for how an energy procurement strategy is structured. ISO New England (ISO‑NE) has documented an ongoing shift away from older coal and oil units toward natural gas, renewables, and storage, along with significant investment needs for aging transmission infrastructure.
The ISO’s draft 2025 Regional System Plan projects more than an 11 percent increase in annual regional electricity use between 2025 and 2034, driven largely by electrification and state climate policies, even before layering in more aggressive AI data center growth. In a region with limited interconnections, winter gas constraints, and ambitious renewable portfolio standards, additional large, relatively constant AI loads can heighten sensitivity to capacity shortages, congestion, and policy outcomes.
Regional structure and constraints
New England’s resource mix continues to rely heavily on natural gas, complemented by nuclear, hydro, and growing amounts of wind, solar, and storage, while many legacy fossil units are at risk of retirement. ISO‑NE and state policymakers are pursuing long‑term transmission projects to increase transfer capability, including proposals to move more clean energy from northern Maine and neighboring regions into load centers, but those upgrades take years to plan and build.
“AI data centers sited near existing bottlenecks or in areas with limited local generation can therefore contribute to higher congestion costs and more volatile nodal prices, even if total regional capacity remains adequate on paper,” said Vissat.
Procurement profiles and exposure
Within this environment, different procurement structures in New England face distinct exposures as AI-related demand grows. Fixed-price full-requirements contracts can provide strong budget certainty. However, long terms may embed higher premiums if markets anticipate sustained load growth and transmission constraints. Block forwards tied to specific shapes or hours can target known exposure yet leave residual load subject to ISO-NE wholesale prices, which may be more volatile in regions with data center clusters or winter reliability concerns.
Hybrid structures that blend fixed and index components, and that are informed by granular load forecasts, allow organizations to calibrate how much volume is hedged versus exposed, and in which time blocks.
Guardrails and practical takeaways
Concerns about skyrocketing prices in New England should be framed with clear guardrails. ISO‑NE’s current planning outlook suggests that, given expected new resources, the region can meet demand through at least the early 2030s, although the ISO also highlights significant uncertainty tied to policy, fuel supply, and infrastructure conditions.
This indicates that abrupt, sustained price spikes are more likely in specific locations or seasons, such as winter periods with tight gas supply, rather than as a simple, linear effect of AI load across every hour of the year. Liquidity in forward markets, evolving capacity market rules, and state-level clean energy and energy efficiency investments also act as stabilizing forces, even as they introduce new variables that need to be incorporated into planning.
What this means for decisionmakers
For decision-makers, the practical implication is that an energy procurement strategy in New England should prioritize early hedging and robust demand forecasting as AI data centers expand. Locking in a portion of expected load through multi-year fixed or block forward products while maintaining some flexible exposure allows organizations to benefit from periods of lower prices without leaving their full portfolio vulnerable to potential AI-driven congestion or scarcity events.
Detailed forecasts that account for operational changes, efficiency projects, and possible AI-related load growth at nearby facilities provide the foundation for sizing those hedges and timing procurement decisions. This approach reduces downside risk, supports more accurate budgets, and positions organizations to respond thoughtfully as New England’s resource mix, transmission system, and policy environment continue to evolve.
Procurement strategy implications for organizations
Organizations facing rising AI-related demand need an energy procurement strategy that treats volatility as a core planning variable rather than an afterthought. A structured approach that links forecasting, hedging, load shaping, and governance helps decision-makers connect market signals with internal budget and risk priorities in a consistent way. This approach also supports clearer communication with executives and boards that expect traceable logic behind procurement decisions.
“An effective framework begins with robust demand forecasting that combines metered data, operational plans, and realistic assumptions about future electrification and digital workloads,” said Vissat. “Forecasts should distinguish between baseload, seasonal peaks, and discretionary or shiftable loads so that procurement teams can match products to specific risk profiles.” Within that context, hedging decisions focus on how much volume to fix over what terms, while load shaping strategies consider efficiency projects, demand response, and operational changes that reduce or shift usage away from the costliest hours. Governance ties these elements together through clear roles, decision thresholds, and documentation standards so that procurement actions align with financial and risk policies.
Short-term actions to build resilience
In the next 6 to 12 months, organizations can strengthen their energy procurement strategy with a set of practical steps. Establishing a reliable baseline of consumption at each facility, including interval data where available, gives teams a clearer picture of how load aligns with wholesale price patterns. Scenario analyses that test steady versus accelerated AI-related demand growth, as well as changes in fuel prices or policy, help quantify potential exposure and inform the mix of fixed-price, block forward, and index-linked products. With that foundation, decision-makers can define target hedge ratios by term (for example, one, three, or five years) and structure solicitations that invite suppliers to propose combinations of products that address specific risk tolerances.
Long-term strategy and scalability
A scalable energy procurement strategy requires a repeatable playbook rather than one-off decisions. Such a playbook can describe standard analysis steps, preferred contract structures, approval workflows, and communication protocols for different market conditions. Supplier diversification helps reduce counterparty risk and can create competitive tension in pricing, especially when organizations operate across multiple states or ISO footprints. Where corporate sustainability goals or investor expectations call for emissions reductions, sustainability-linked pricing or clean energy attributes can be integrated into procurement decisions, provided premiums and benefits are assessed with the same rigor as traditional cost metrics. This makes it easier to demonstrate how procurement supports both financial and ESG objectives over time.
Measuring ROI and aligning with regulation
Any procurement framework built around an energy procurement strategy must translate into measurable outcomes. Key ROI considerations include the net cost of hedging versus expected savings, improvements in budget certainty (for example, reduced variance between actual and budgeted energy spend), and quantified reductions in exposure to extreme price events. Organizations can also track metrics such as avoided cost from efficiency and load management projects, or changes in the share of consumption served under preferred contract structures.
From a regulatory and sustainability perspective, procurement teams should monitor evolving state and federal requirements related to emissions reporting, renewable portfolio standards, and reliability, and confirm that contracts and reporting practices support compliance. This alignment not only reduces regulatory risk but also strengthens ESG signaling to stakeholders who increasingly scrutinize how energy decisions support long-term resilience and climate commitments.
Getting ahead of the structural shift in power demand
AI-driven data centers are accelerating a structural shift in power demand, and organizations that prepare now will be better positioned to manage cost and risk. An intentional energy procurement strategy gives decision-makers a way to translate that shifting demand picture into concrete actions on pricing, hedging, and long-term resilience that can withstand board-level scrutiny.
Kb3 Advisors helps organizations turn complex market signals into clear, defensible decisions. The team provides market intelligence, demand forecasting, price trajectory modeling, and scenario planning so leaders can see how different AI adoption pathways, policy changes, and regional constraints may affect future costs. This insight informs procurement program design, including the mix of fixed-price hedges, structured products, and index exposure, along with implementation support and ongoing performance reporting tailored to executive needs.
We align your energy procurement strategy with financial targets, operational requirements, and sustainability commitments. Transparent reporting, data-backed recommendations, and clear ROI projections help decision-makers communicate the case for action to CEOs, finance committees, and boards. For organizations that want to stay ahead of AI-driven load growth rather than react to it, Kb3 Advisors offers risk-adjusted strategies and executive-ready materials that support timely, confident decisions.
To position your organization for the next decade of AI-related demand growth, connect with Kb3 Advisors to review your current portfolio, stress-test your procurement approach, and design a strategy that protects budgets while supporting long-term strategic goals.
Sources
- Projecting the Electricity Demand Growth of Generative AI Large Language Models in the US. energypolicy.columbia.edu. Accessed December 18, 2025.
- DOE Releases New Report Evaluating Increase in Electricity Demand from Data Centers. energy.gov. Accessed December 18, 2025.
- 2025 Regional System Plan identifies future needs of New England grid. isonewswire.com. Accessed December 18, 2025.
