Early in the morning in northern Virginia, the lights inside a windowless warehouse complex never go off. Thousands of servers hum in tight formation, training artificial intelligence models that will write code, diagnose disease, and accelerate the tempo of global commerce. The people who built these systems track processor speed, model parameters, and inference latency with obsessive precision. But the thing that ultimately determines whether any of it runs at all is something far older and far simpler: the electron moving through a wire.
For decades, electricity demand in the United States behaved the way a mature utility sector is supposed to behave. It crept along at roughly one percent annual growth, sometimes less. Efficiency gains offset population increases and economic expansion. Policymakers treated the grid as settled infrastructure rather than competitive strategy. Investors treated it as ballast in a portfolio rather than propulsion. The switch flipped, the room lit up, and nobody gave it much thought.
That era ended quietly, somewhere around 2023, and the numbers have been telling the story ever since. Demand dipped that year, then surged roughly three percent in 2024, and continued climbing in 2025. More than fifty gigawatts of new data center capacity was announced in the space of less than two years. In certain corridors of Northern Virginia, developers began reporting multi-year waits simply for a new power connection. The bottleneck had moved. It was no longer the ability to generate electricity. It was the ability to approve it, permit it, and move it.
When everything becomes electric, electricity stops being a line item. Electricity becomes strategy.
This past weekend, our team at HarnessAI built a working energy markets dashboard at energy.harnessai.net over roughly forty-eight hours. The exercise began as a technical proof of concept and turned into something closer to a reckoning. We structured it as a layered briefing: executive summary, domestic history, domestic forecast, international history, international forecast, and investment implications. The sequence was not arbitrary. The structure forces you to separate what has already happened from what is being predicted, and it prevents a reader from mistaking a good headline for an actual trend. What the data revealed was not a cycle. It looked more like a regime change.
THE AMERICAN GRID UNDER STRESS
The domestic history of the past three years is best understood as a machine that idled for a decade and was then thrown into high gear without adequate preparation. Data centers alone consumed approximately 176 terawatt hours in 2023, representing roughly 4.4 percent of total U.S. electricity demand. Solar dominated new capacity additions, holding that position for more than two consecutive years. In 2024, roughly thirty gigawatts of utility-scale solar and more than ten gigawatts of battery storage came online. Natural gas additions slowed sharply even as reliability concerns grew louder. Renewables were expanding, but the infrastructure required to connect them to the customers who needed the power was not keeping pace.
Prices reflected this tension in the clearest possible terms. Residential electricity rates climbed from around sixteen cents per kilowatt hour in 2023 to nearly eighteen cents by late 2025. Henry Hub natural gas prices rebounded from historic lows, moving from the mid-two-dollar range in 2024 toward the mid-three-dollar range in 2025. Gas remained the swing fuel that kept the grid stable during peak load events and periods of renewable variability, which meant that its price trajectory carried consequences well beyond the natural gas market itself.
The single most revealing number in the domestic picture, however, is not a price or a capacity figure. It is the size of the interconnection queue. The United States currently has roughly 2,600 gigawatts of generation and storage projects waiting to connect to the grid. That number is nearly twice the size of the entire existing grid. Projects that would accelerate the energy transition sit idle not because the technology is unavailable or the capital is absent, but because the administrative process for approving new grid connections cannot move fast enough to clear the backlog. Electrification has become, at its core, a governance problem.
THE NEXT THREE YEARS IN AMERICA
Forecasts suggest that the recent surge is not a temporary aberration returning to a familiar baseline. Average annual load growth estimates in some projections approach five to six percent over the next five years. Peak demand could rise by more than 150 gigawatts across that period. Artificial intelligence data center demand alone is projected to expand dramatically, from single-digit gigawatt territory just a few years ago toward triple-digit levels by the mid-2030s. These are not marginal additions to a stable system. They are multipliers applied to a system already operating near its limits.
The supply response is taking shape, though unevenly. Solar additions are expected to accelerate further in 2026, with planned capacity well above 2025 levels. Natural gas capacity is poised for a modest rebound as developers seek dispatchable generation capable of providing reliability when the sun is not shining and the wind is not blowing. Battery storage continues to scale, though supply chain constraints and permitting timelines slow the pace. The technology exists. The physics are understood. The friction is institutional.
Transmission remains the decisive variable in the domestic forecast, and it may be the most underappreciated constraint in the entire energy conversation. The existing grid was designed for large centralized generators located near population centers. It was not designed to handle a distributed and highly variable renewable fleet paired with hyperscale data centers that consume electricity at industrial scale around the clock. Rebuilding or expanding that grid requires not only capital and engineering, but workforce development, environmental review, and the coordination of dozens of regulatory bodies operating under different rules. The electrons can be produced. The question is whether they can be moved.
THE GLOBAL SPLIT SCREEN
Internationally, the story diverges in geography but rhymes in structure. China has been adding renewable capacity at a scale that renders the word record almost meaningless, installing hundreds of gigawatts in a single year while simultaneously maintaining coal capacity as a reliability hedge. India’s electricity demand is rising rapidly, driven by urbanization, a growing middle class, expanding cooling needs, and industrial ambition. Europe continues recalibrating its energy mix in the aftermath of geopolitical shocks that exposed the fragility of dependence on imported fossil fuels, balancing decarbonization commitments against the immediate demands of industrial competitiveness and consumer affordability.
The International Energy Agency projects global electricity demand growing at roughly 3.6 percent annually through the end of the decade, adding on the order of 1,100 terawatt hours each year. Grid investment requirements are tracking toward 600 billion dollars annually by 2030. Interconnection queues are swelling across regions. The IEA’s Electricity 2026 report describes a world that is not short on generation ambition. What the world is short on is wires, transformers, regulatory bandwidth, and time. The common constraint is not ideological. It is physical and administrative.
Coal is gradually declining in its share of global generation, though the timeline varies significantly by country and the decline is far from linear. Renewables and nuclear are expanding. But the transition is happening against a backdrop of demand growth that is itself accelerating. The result is a global system adding clean energy at historic rates while simultaneously struggling to serve load that is growing faster than anyone projected five years ago.
THE CAPITAL ALLOCATION QUESTION
Viewed through the lens of long-term capital allocation, this situation is less about energy ideology than about identifying durable structural positions. Grid infrastructure occupies a singular place in the economy because every electron, regardless of its origin, must pass through regulated wires. That regulated passage creates a form of pricing certainty that is rare in modern markets. Natural gas persists as a bridge fuel not because investors are indifferent to decarbonization, but because reliability is not optional. A grid operator who cannot keep the lights on during a heat dome does not get credit for the percentage of their portfolio that is renewable.
Solar appears economically compelling at the generation level, but the more strategic asset in the current environment may be interconnection access itself. A queue position that allows a new data center to receive power two years before a competitor is not a minor operational detail. It is a structural advantage. The capital that flows into transmission infrastructure, into battery storage capable of bridging renewable variability, and into the engineering and legal expertise required to navigate interconnection queues is likely to earn durable returns precisely because the constraint it resolves is not going away on any near-term horizon.
We built energy.harnessai.net not to offer investment advice or to advocate for any particular energy source. We built it because we believed the story embedded in this data deserved a structured, navigable form. The dashboard moves from history to forecast, from domestic to international, and finally to implications, in the same order that a careful analyst would. It is designed to prevent the reader from reaching the conclusion before they have seen the evidence.
THE END OF INVISIBLE INFRASTRUCTURE
There is a broader cultural shift embedded in this moment that deserves acknowledgment. Electricity was invisible for most of the past century not because it was unimportant, but because it was so reliable that taking it for granted was rational. The grid was background infrastructure, like municipal water or paved roads. It was noticed primarily when it failed. The work of keeping it running was technical, unglamorous, and largely out of public view.
That condition is ending. The grid is reemerging as a central platform of economic and geopolitical competition. Nations that can build transmission infrastructure quickly, approve interconnection requests efficiently, and coordinate the financing of large capital projects across public and private actors will gain a structural advantage in the industries that define the next generation of economic growth. Nations that cannot will find themselves constrained by their own electrons, unable to run the artificial intelligence systems, electric vehicles, and advanced manufacturing facilities that require abundant, affordable, reliable power.
Flip a switch, and the room lights up. That act of faith, performed billions of times a day around the world, rests on an infrastructure system that is now being asked to do things it was never designed for, at a speed it was never built to support, under a level of public scrutiny it has never previously attracted. The question is no longer whether the electrification of the global economy will happen. The question is who builds fast enough, coordinates carefully enough, and finances patiently enough to keep the lights on in an age that demands more power than any age before it.