Duke Energy AI Grid Expansion, $83 B Capex Plan, GE Vernova 11 Turbine Deal, and Meta PPA (2021 to 2025)

Grid Expansion Risks, Duke Energy’s $83 B Plan for AI Demand

The explosive growth of artificial intelligence is forcing a radical and rapid recalibration of long-term capital planning at major utilities, with Duke Energy emerging as a primary case study in the immense execution, regulatory, and financial risks involved. The company’s strategy in 2025 shifted from a measured energy transition to an urgent, large-scale infrastructure build-out designed to capture the historic load growth from AI-powered data centers. This pivot is a direct response to a fundamental change in electricity demand, which the company’s CEO noted is growing at a rate 10 times faster than the historical pace, driven almost entirely by the technology sector.

The 2025 AI Demand Shock

Prior to 2025, data centers were a component of load growth forecasts, but not the single largest determinant of system planning. The commercialization of generative AI models created a step-change in power requirements, moving beyond linear projections. By early 2025, Duke Energy‘s planning assumptions were invalidated, forcing a complete overhaul of its capital expenditure strategy. This demand shock is not an incremental increase but a structural shift, compelling the utility to secure generation capacity on a scale and timeline not seen in decades.

The $83 Billion Infrastructure Response

In response to the new demand profile, Duke Energy increased its five-year capital plan by 13.7% to $83 billion in February 2025. This capital is aimed squarely at accommodating new industrial and data center loads. A key component of this strategy is the focus on new, dispatchable natural gas generation to provide the 24/7 reliable power required by data centers. The pivot also puts other capital-intensive projects, such as the company’s prior explorations into offshore wind, into a different strategic context, as immediate, reliable capacity takes precedence.

Regulatory and Ratepayer Hurdles

Committing over $83 billion in new capital creates substantial regulatory risk. Duke Energy must now navigate state-level public utility commissions to gain approval for these expenditures and, critically, the associated rate increases needed to pay for them. The utility has publicly pledged to ensure “fair bills” for all customers, but the scale of investment makes rate hikes for residential and commercial customers almost unavoidable. The primary challenge will be to create rate structures that allocate costs appropriately to the large industrial users driving the need for the new infrastructure, a process that is politically and regulatorily complex.

$83 B Capital Plan, Duke Energy’s Investment in Generation

Duke Energy‘s 2025 investment strategy is defined by a significant increase in its five-year capital plan to $83 billion, reallocating resources to build new generation capacity specifically to serve the immense power needs of data centers. This is not a speculative investment but a direct reaction to concrete load requests from hyperscalers operating in its service territories. The financial commitment reflects a new reality where utility investment is directly tethered to the growth trajectory of the technology sector.

Capital Plan Increased by 13.7% for Data Centers

The announcement in February 2025 to raise the five-year capex plan from previous forecasts to $83 billion was explicitly linked to accommodating the surge in electricity demand from data centers and advanced manufacturing. This 13.7% jump represents a material change in corporate strategy, shifting the company’s focus toward large-scale generation and transmission projects. This move positions Duke Energy to attract and retain major technology companies by signaling a firm commitment to providing the necessary power infrastructure.

Projecting Future GW Needs for The Carolinas

To quantify the scale of the required build-out, Duke Energy stated in March 2025 that it would need to supply power for at least 1.5 gigawatts of new data centers in its Carolinas service territory alone by the year 2033. This single projection, equivalent to the output of a very large power plant, illustrates the magnitude of the new infrastructure required. The utility’s subsequent filings and procurement activities throughout 2025 are direct consequences of this long-range forecast.

Table: Duke Energy Strategic Investments and Projections (2025)

Duke Energy’s GE Vernova and NVIDIA Alliances (2025)

In 2025, Duke Energy forged critical partnerships on both the hardware supply side with GE Vernova and the software development side with NVIDIA and the Electric Power Research Institute (EPRI). This dual-pronged partnership strategy is designed to secure the physical generation assets required to meet AI-driven demand while simultaneously developing the grid intelligence needed to manage an increasingly complex and strained system.

Securing Generation with GE Vernova

The most significant infrastructure partnership of 2025 was the April 24 arrangement with GE Vernova. This agreement provides Duke Energy with a clear procurement path for up to 11 American-produced 7 HA.03 natural gas turbines. This move directly addresses the supply chain risk for critical generation equipment, ensuring the utility can execute its build-out plans to meet the reliable, continuous power profiles demanded by data center customers.

Developing Grid Intelligence with NVIDIA

Beyond physical hardware, Duke Energy joined the Open Power AI consortium in March 2025, a collaboration led by EPRI and NVIDIA. This initiative aims to build open-source AI models for the power sector. By participating, Duke Energy signals its intent to be a key player in developing the advanced software needed for grid optimization, outage prediction, and efficient integration of new loads, rather than simply being a passive consumer of third-party AI tools.

Managing Demand with Hyperscalers

Duke Energy also finalized a 20-year power purchase agreement (PPA) with Meta in 2025 to provide 1, 121 MW of power. This type of long-term contract directly with a hyperscaler is becoming a key commercial tool. It provides revenue certainty for the utility’s new generation assets and allows the tech company to secure the massive blocks of power it needs for future growth while meeting its own clean energy targets.

Table: Duke Energy Key Partnerships (2025)

The Carolinas and Florida, Duke Energy’s Data Center Hubs

Duke Energy‘s strategic focus in 2025 concentrated heavily on its legacy service territories, particularly the Carolinas and Florida, which are rapidly becoming primary destinations for AI data center development. This geographic concentration is not accidental but a direct result of the utility’s proactive, albeit risky, strategy to build out generation and transmission infrastructure ahead of demand, making these regions more attractive to hyperscalers than areas with constrained grids.

• In contrast to the pre-2025 period where load growth was more distributed, the Carolinas emerged as the epicenter of Duke Energy‘s data center strategy, underpinned by the projection of 1.5 GW of new demand by 2033.
• The company’s September 2025 filing with Florida regulators to prepare for future data center development demonstrates a clear intent to replicate the Carolinas strategy in another key growth state, proactively addressing infrastructure needs before they become critical constraints.
• This regional focus is a calculated decision to leverage existing infrastructure, regulatory relationships, and large land parcels suitable for both power plants and data centers, creating a self-reinforcing development cycle.
• By committing to supply power in these specific areas, Duke Energy is actively shaping the geography of AI infrastructure in the United States, directing multi-billion dollar investments from tech companies to its territories.

AI for Grid Management, Duke Energy’s Patented Technology

In 2025, Duke Energy demonstrated a maturing internal capability by moving beyond the simple application of third-party AI to developing and patenting its own proprietary technologies for grid management. This shift indicates that the utility views AI not just as an operational efficiency tool but as a core competency required to manage the grid of the future, characterized by large, complex loads like data centers.

• The most significant validation of this strategy came in January 2025, when the company announced it had created and patented new grid simulation technology. This tool allows for proactive planning by modeling how the grid will operate under future scenarios, including the impact of massive, concentrated data center loads.
• The utility’s use of a hybrid AI model, which combines traditional engineered analytics with machine learning to identify grid vulnerabilities, shows a nuanced and practical approach. This method achieves higher accuracy than either technique alone, enhancing the resilience of the power system.
• Prior to 2025, AI applications in the utility sector were often focused on standardized tasks like vegetation management or basic predictive maintenance. Duke Energy‘s development of proprietary simulation and vulnerability-spotting tools represents a significant step up in technological sophistication and strategic importance.
• Participation in the Open Power AI consortium alongside NVIDIA further reinforces this trend. It shows a commitment to co-developing industry-wide standards, a hallmark of a technology sector moving from isolated experimentation to scalable deployment.

SWOT Analysis for Duke Energy’s AI Infrastructure Plan

Duke Energy‘s immense capital plan to meet AI demand positions it for significant long-term growth but also exposes it to considerable execution, regulatory, and competitive risks. Its established market position and access to capital are formidable strengths, but the speed and scale of the required build-out introduce new vulnerabilities that were less prominent before 2025.

• Strengths are rooted in its scale as one of the largest U.S. utilities, providing access to capital markets for its $83 billion plan and an incumbent position in key growth territories.
• Weaknesses include a heavy reliance on natural gas for new generation, which carries long-term environmental and commodity price risk, and the inherent complexity of executing multiple large-scale construction projects simultaneously.
• Opportunities are centered on becoming the premier energy provider for the AI economy, locking in decades of revenue growth from a new class of high-demand industrial customers.
• Threats are intensifying, including regulatory opposition to rate hikes, public resistance to new fossil fuel infrastructure, and emerging competition from decentralized, behind-the-meter power solutions.

Table: SWOT Analysis for Duke Energy’s AI-Driven Strategy

Duke Energy Scenario: Watch Rate Cases and Demand Flexibility

The success of Duke Energy‘s AI-driven infrastructure strategy over the next 18-24 months hinges on two critical variables: its ability to successfully navigate regulatory rate cases to fund its investments and the potential to implement demand flexibility programs with data centers to mitigate peak grid stress. The outcomes of these two factors will determine the pace of the build-out, the cost to consumers, and the utility’s long-term financial health.

• If state utility commissions in the Carolinas and Florida approve the full scope of Duke Energy‘s capital requests and proposed rate structures, then watch for an acceleration of construction timelines for new gas plants and transmission lines. This would signal regulatory alignment with the utility’s growth strategy and likely strengthen its position as a preferred location for data center investment.
• If new power contracts with data centers begin to incorporate demand flexibility, as explored in a recent Duke University study, then these could be happening: pilot programs where data centers curtail non-critical workloads during peak demand hours. Successful implementation could reduce the need for some new peaking power plants, lowering the overall system cost and mitigating some regulatory and public opposition.
• If the onsite generation model demonstrated by the Bloom Energy and Brookfield partnership gains more traction with other data center developers, then watch for Duke Energy to potentially develop its own behind-the-meter service offerings or, alternatively, face increased pressure to lower its industrial rates to remain competitive.

The questions your competitors are already asking

This report covers one angle of Duke Energy’s strategic response to AI-driven power demand. The questions that matter most depend on your work.

• Other US utilities building power for data centers
• Duke Energy rate increases for data center buildout
• Data centers using onsite power generation
• Renewable energy contracts for AI data centers

This report does not answer these. Enki Brief Pro does.

Your question, your angle, your framework. SWOT, PESTL, scenario modelling. The same niche depth, built around the decision your work actually depends on.

Run your first brief in Enki Brief Pro