Ford Motor Company did not get into the energy storage business because it wanted to. It got in because the companies buying its cars also need to power the server farms training the models that may eventually design the next generation of those cars. In May, Ford launched a subsidiary called Ford Energy, backed by $2 billion, to sell battery storage systems to data centers and other large power users. Ford's stock rose to its highest level in three years. The announcement told investors everything they needed to know: electricity is now a growth sector, and almost any company with manufacturing capacity and a balance sheet can stake a claim in it.
This is not an energy story. It is a structural story about what happens when one technology — artificial intelligence — requires so much power, so fast, that it reorganizes the economy around the scramble to supply it. Ford is the most legible symbol of that reorganization, but it is not the most consequential. The consequences are accumulating in utility queues, in canceled construction projects, and on household electricity bills in communities that never asked to host a data center.
The financial numbers are genuinely staggering. Bloom Energy, which makes on-site power generation equipment, saw its stock price climb more than 1,200 percent over the past year, according to Axios. Fervo Energy, a geothermal startup that spent years being dismissed as speculative climate technology, surged after going public earlier this month. GE Vernova booked $2.4 billion in electric equipment orders for data centers in the first quarter of 2026 alone — more than its equivalent sales for all of last year, with its stock up roughly 60 percent this year. These are not incremental gains. These are numbers that rewrite a company's identity.
What is driving this? The arithmetic of AI infrastructure is brutal. A single large-scale data center can consume as much electricity as a small city. The companies building them — and the companies training the models running inside them — need power at a scale and speed that existing grid infrastructure was not designed to provide. "Utilities are inundated with applications for power and doing triage on who's real," Andy Power, president and CEO of Digital Realty, one of the world's largest data center companies, told Axios. The queue for grid connection has become, in effect, a secondary market in its own right — a chokepoint that determines which AI projects get built and which die waiting.
The original thesis the financial press is missing: the AI energy rush is not a story about innovation. It is a story about the privatization of a public resource. Electricity — historically regulated as a utility, priced as a near-commodity, and managed as shared infrastructure — is being claimed as a strategic asset by private actors whose primary obligation is to shareholders, not to ratepayers or grid stability. When Ford Energy sells battery storage to a hyperscaler, it is not building public energy resilience. It is building a private bypass around a grid that everyone else still depends on.
Brian Janous, who was Microsoft's first energy hire 15 years ago and is now co-founder of data center developer Cloverleaf Infrastructure, was direct about the risk in remarks to Axios: "A lot of people are going to lose a lot of money in this space" — not because demand for AI compute will collapse, but because so many massive projects are simultaneously chasing that demand. He pointed specifically to a troubled Texas project that bills itself as the largest data center proposal in the world, and a Utah proposal backed by celebrity investor Kevin O'Leary. These are not cautionary tales about bad technology. They are cautionary tales about capital flooding into a sector faster than the underlying infrastructure can accommodate it.
The cancellation data makes the risk concrete. The number of data center projects canceled after local pushback hit a record high in the first quarter of this year, according to data compiled by Heatmap Pro and reported by Axios. Those canceled projects accounted for more than $40 billion in planned investment. That figure deserves to be read carefully: it is not $40 billion in losses to shareholders. It is $40 billion in projects that communities — often with no formal regulatory mechanism to stop them — managed to block anyway. The opposition is coming from both sides of the political spectrum, a point our earlier reporting on bipartisan data center resistance documented in detail. Rural towns in red states and suburban counties in blue states are reaching the same conclusion: the economic benefits of a data center accrue elsewhere, and the costs — power draw, water consumption, land use, noise — stay local.
That tension is not a bug in the AI energy boom. It is a structural feature of how American infrastructure development works. The companies that win grid connection rights, secure land, and lock in power purchase agreements will extract enormous value. The communities hosting that infrastructure will absorb the externalities. This is the same pattern that has governed fossil fuel extraction, semiconductor manufacturing, and warehouse logistics for decades. The AI boom did not invent the model. It accelerated it, and dressed it in the language of technological progress.
There is a secondary accountability question that the stock-price narrative consistently obscures: who is responsible for the grid when it buckles? The companies entering the energy business — Ford, Bloom, Fervo, GE Vernova — are doing so to capture value, not to bear risk. If a region's grid becomes unstable because demand from data centers has outpaced transmission capacity, the cost of that instability falls on every ratepayer in the service area. The companies that drew down the grid's headroom are not liable for the brownouts. Regulators, utilities, and ultimately consumers absorb that cost. As the debate over federal guardrails for AI energy consumption has shown, the political will to impose meaningful constraints on that dynamic remains minimal.
The international dimension of this story is almost entirely absent from American business coverage. The electricity that U.S. data centers consume is not generated in a vacuum. Every megawatt directed toward AI training is a megawatt competing with residential and industrial demand. In regions where the grid is already stressed — parts of the American Southwest, for instance — the competition is not abstract. It shapes energy prices, reliability, and who gets prioritized in a crunch. Beyond U.S. borders, the global race for AI infrastructure is driving similar dynamics in Europe, Southeast Asia, and sub-Saharan Africa, where energy access is far more constrained and the populations bearing the cost of industrial power demand have even less political recourse. The scramble for electricity that looks like a gold rush from a Wall Street terminal looks quite different from a household in a country where data center construction is crowding out residential grid expansion.
AI model training and inference require sustained, high-density power that standard commercial real estate cannot provide. A single large-scale AI data center can draw 100–500 megawatts continuously — comparable to a small city's residential load. Utilities process grid connection requests in order, and queues in some regions now stretch years. Companies that secure early power purchase agreements or build on-site generation gain a durable competitive advantage over rivals still waiting in line.
The financial press frames this as a moment of opportunity. That framing is not wrong — the opportunity is real, and the stock prices confirm it. But opportunity for whom is a question the headlines consistently skip. The companies positioned to win are large, capital-rich, and already embedded in the technology ecosystem. Ford can spin up a $2 billion subsidiary because Ford has $2 billion to spin up. Fervo Energy can go public because it had venture backing through years of being called speculative. The electricity gold rush, like most gold rushes, is being won by the people who arrived with capital, not by the communities sitting on top of the resource.
The deeper issue is regulatory. American electricity markets were designed around a utility model that assumed relatively predictable demand growth and public oversight of major infrastructure decisions. The AI boom has broken both assumptions. Demand is growing faster and less predictably than any utility planning cycle anticipated. And the major actors driving that demand — hyperscalers, data center developers, and now automakers-turned-energy-companies — are private firms making infrastructure decisions at a scale that used to require public deliberation. The SEC's recent decision to rescind climate disclosure rules, leaving investors without reliable data on corporate energy risk, means that even the financial system lacks the information it would need to price this risk correctly.
Janous's warning — that a lot of people are going to lose a lot of money — is almost certainly correct. What he did not say, and what the Axios source material did not press him on, is that the people who lose the most money in a speculative bust are rarely the ones who made the decisions that caused it. The investors who funded the Texas megaproject and the Utah proposal will take losses. The communities that were promised tax revenue and local jobs from data centers that never broke ground will be left with nothing. The ratepayers whose electricity prices climbed because utilities prioritized industrial demand will not get a refund. The gold rush has winners and losers. The winners are already known. The losers are still being selected.
