The Most Expensive Electric Grid in the World and the Social Capital No One Is Auditing
Duke Energy has just announced one of the largest infrastructure investment programs in U.S. corporate history: $220 billion to modernize its electric grid, incorporating self-healing technologies, storm protection, and the capability to handle demand from artificial intelligence data centers. The plan includes a 13.7% increase in its five-year capital budget, reaching $83 billion, alongside a 20-year extension of the Oconee nuclear plant license, approved by the Nuclear Regulatory Commission on March 31, 2025.
The numbers are staggering. The company serves 8.2 million electric customers across six states, operates 50,259 megawatts of installed capacity, and maintains a commercial renewable energy segment with over 3,554 MW distributed across 22 states. From Charlotte, North Carolina, CEO Harry Sideris describes a company fully executing its strategy: new investments in infrastructure, agreements with GE Vernova for U.S.-made natural gas turbines, and four battery storage projects in Florida promising $843 million in savings over their lifespan.
But there’s an analytical layer conspicuously missing from any financial report, and it’s precisely what will determine whether this $220 billion program delivers on its promises or devolves into a costly infrastructure riddled with foundational flaws.
When Scale Hides the Blind Spots in Design
An electric grid that must anticipate failures, manage the volatile demand from data centers, and withstand hurricanes in Florida is not merely an engineering problem. It is a matter of distributed collective intelligence. And that’s where standard analysis falls short.
The self-healing technologies Duke plans to implement, systems that autonomously detect and isolate faults, make real-time decisions on millions of points in the grid. These decisions are encoded by human teams. The assumptions those teams operate under about how electricity is used, when, and from where, directly reflect the makeup of those who designed them. When these teams are homogeneous in origin, training, and perspective, the resulting system is technically sophisticated but socially narrow.
This is not an ideological argument. It’s an operational risk mechanic. The U.S. electric industry has a documented history of underestimating demand in low-income communities and rapidly growing peri-urban areas — precisely the locales where vehicle electrification and housing densification are generating the most unpredictable load spikes. If the demand models inherit the same perceptual biases as previous teams, the $220 billion could merely construct a network optimized for the past.
The cost of a minute of downtime in a data center is, according to industry data, $73 million. Duke Energy knows this. What is not always measured with the same urgency is the accumulated cost of building infrastructure based on incorrect assumptions about real demand behavior.
Social Capital as Invisible Infrastructure
Duke Energy operates over vast territories: 91,000 square miles across the Southeastern and Midwestern United States. That geography is not homogeneous. It includes rural communities in North Carolina, rapidly developing industrial corridors, seasonal coastal areas in Florida, and urban pockets with high concentrations of variable loads. Each of these segments has local actors — municipal governments, electric cooperatives, chambers of commerce, real estate developers, logistics infrastructure operators — who possess insights on future demand that no central model captures by default.
What makes a network of this magnitude function effectively in the long term is not just the financial capital flowing into transformers and wires. It’s the density of trust relationships that the company builds or erodes with those peripheral stakeholders. A well-capitalized network that operates without that density of local ties is fragile in a specific way: it can withstand physical storms but may not survive cycles of regulatory distrust, community opposition to substations, or friction in local permitting that systematically delays infrastructure projects in the United States.
Harry Sideris himself mentioned at the annual shareholder meeting on May 1, 2025, that the advancement of the strategy hinges upon collaborations with stakeholders and “constructive regulatory outcomes.” That phrase, in corporate language, is a euphemism for something very concrete: if state commissions do not approve the recovery of capital costs, the $83 billion in the five-year plan will not yield the projected returns. And state commissions ultimately respond to political pressure built from the ground up.
Thus, social capital is not a soft asset. It is the condition of financial viability for the program.
What SMEs in the Southeast Must Read Between the Lines
Duke Energy serves millions of small and medium-sized commercial and industrial customers across its six operating states. For this universe of businesses, the modernization of the grid has concrete implications that extend beyond supply stability.
First: electricity rates are going to rise. Investment programs of this scale recover costs through regulated tariffs. Duke needs the approval of state commissions to transfer those costs to customers, and when that happens, small business customers absorb increases that can affect already tight margins. SMEs in the industrial corridor of Florida, North Carolina, and Ohio need to model their energy costs now under scenarios of progressive rate increases between 2026 and 2030.
Second: the four battery storage sites in Florida, projected to be completed by the summer of 2026, signal Duke’s effort to build real-time demand management capacity. This eventually opens the possibility for demand response programs where commercial customers that reduce consumption during peak loads receive tariff compensation. Businesses that install energy measurement and management capacity before these programs become available will be better positioned to capture that benefit.
Third, and more structurally: the industrial electrification Duke is financing with these $220 billion will generate competition for specialized technical talent in electrical maintenance, charging system installation, and energy infrastructure management. SMEs in the services and manufacturing sectors that start building those internal capabilities before the market drives costs higher will have a cost advantage that compounds over time.
Homogeneity at the Design Table Comes with a Price, Eventually
Duke Energy’s program is genuinely ambitious and is financially backed. The 13.7% increase in the capital budget, the nuclear extension at Oconee, agreements with GE Vernova, and the battery timeline in Florida form a coherent strategy. I recognize that without reservation.
However, executing a $220 billion program on physical infrastructure that interacts with complex and diverse communities has a failure point that no accounting balance reflects in time: the architecture of who makes design decisions and with what mental map of the territory they operate.
Distribution networks that fail to anticipate the consumption patterns of rapidly growing communities, or that generate accumulated regulatory frictions due to a deficit of local trust, do not collapse in a quarter. They degrade slowly, delivering returns below projections and generating litigations and delays that erode the program's value in the fourth, fifth, and sixth years of the five-year plan.
Modernizing the most expensive electric grid in U.S. corporate history will depend, to a degree that financial models underestimate, on the quality of social capital built at the periphery of the system. That cannot be bought with gas turbines or nuclear extensions. It is built with the diversity of perspectives that work together from the outset to design the system.
Any executive reading this who thinks it doesn’t apply to their industry should take a look at their own table at the next board meeting: if everyone sitting there learned the same models, from the same institutions, and accumulated experience in the same types of markets, they inevitably share the same blind spots, making them predictable victims of the first disruption that was not in their shared manual.
