Who Pays for AI?

It's Thursday, February nineteenth, and Welcome to HigherIQ. Last July, sixty data centers in Northern Virginia disconnected from the power grid in the same millisecond. What almost happened next explains a lot about why the economy feels broken even when the numbers say it isn't. On July tenth, twenty twenty-four, a lightning arrestor failed on a high-voltage transmission line in Fairfax County, Virginia. It was a minor equipment malfunction — the kind that happens all the time. The voltage dipped by about eight percent, well within normal operating range. But Northern Virginia isn't normal anymore. It's home to Data Center Alley — a thirty-square-mile corridor hosting over two hundred data centers that route roughly seventy percent of global internet traffic. These facilities run servers for Amazon, Microsoft, Google, and Meta. And they're programmed with hair-trigger protection systems — set to disconnect at voltage dips as small as five percent — because even brief power fluctuations can destroy millions of dollars in hardware and corrupt the data of millions of users. When the voltage dipped, sixty data centers simultaneously switched to backup generators. Not in seconds. In milliseconds. Faster than any human could react. One moment, those facilities were pulling fifteen hundred megawatts from the grid. The next moment, they were gone. Fifteen hundred megawatts is the output of three medium-sized power plants. Imagine shutting down three power plants in under a second — except in reverse. All that electricity was suddenly still being generated with nowhere to go. The power surged. And if the operators at PJM Interconnection and Dominion Energy hadn't acted fast, scrambling to cut generation before systems overloaded, the result could have been cascading blackouts across the region. They contained it. Barely. The lights stayed on. Most people in Virginia had no idea how close they came. There was no press conference, no emergency alert. Just grid operators working frantically behind the scenes while everyone else watched TV and checked their phones — on servers in those same data centers. Federal regulators were shaken enough to launch a task force specifically to study this vulnerability. The North American Electric Reliability Corporation called it a wake-up call. But here's what should worry you: it keeps happening. Texas has logged thirty-four similar incidents since twenty twenty. In one event, four hundred facilities disconnected simultaneously, shedding seventeen hundred megawatts. The grid held. That time. Alison Silverstein, former senior adviser to the Federal Energy Regulatory Commission, put it bluntly: "The behavior of data centers has the potential to cause cascading power outages for an entire region." And the problem is accelerating. Dominion Energy reported that its contracted data center power doubled in just six months in twenty twenty-four. Virginia's total electricity consumption is projected to double over the next fifteen years — almost entirely because of data centers. The grid wasn't built for this. Seventy percent of American electrical infrastructure was constructed between the nineteen fifties and seventies. It was designed for a world where demand grew slowly and predictably. That world is gone. So now we're rebuilding — new transmission lines, new substations, new generating capacity. Tens of billions of dollars in infrastructure investment. The question is: who pays? If you're a homeowner, check your electric bill. You probably already know the answer. Near major data center hubs, electricity prices have increased by as much as two hundred sixty-seven percent. In Oregon, residential customers pay around twenty cents per kilowatt hour. Data centers pay eight cents. Same grid. Same wires. Different rules. In Virginia, Dominion Energy disconnected three hundred thirty-nine thousand households for nonpayment in twenty twenty-four alone. Three hundred thirty-nine thousand families who couldn't afford to keep the lights on. In the same year, the same utility was racing to connect data centers that consume more power than entire cities. Energy analyst Cathy Kunkel put it plainly: "It's almost inevitable that ordinary people are going to end up subsidizing the wealthiest industry in the world." This is the part nobody in Washington wants to say out loud. The economy looks great on paper — GDP growing three percent, corporate profits booming, stock market near all-time highs. But the gains aren't reaching people. Economists have a name for it: the jobless boom. White-collar jobs peaked in November twenty twenty-two. That's the same month ChatGPT launched. In the three years since, finance, insurance, tech, and professional services have been quietly cutting headcount — even as revenue grows. These sectors represent forty percent of GDP. They're supposed to be the engine of American prosperity. And they're shrinking their workforce while the economy expands. This has never happened before. Every previous jobless recovery followed a recession. Companies would lay off workers during the downturn, then slowly rebuild. But this time there was no recession. The economy kept growing. The jobs didn't. Here's where the money went instead. By some estimates, AI-related investment — chips, servers, data centers, electricity — has become one of the largest drivers of US GDP growth over the past two years. Hundreds of billions of dollars flowing into infrastructure. But that capital didn't go to hiring. It went to the data centers we just talked about. The gains are real. They're just being captured by capital, not labor. To be clear: this is primarily a white-collar phenomenon. If you're an electrician or a plumber or an HVAC technician, demand for your skills is at record highs. The trades are booming precisely because someone has to build and wire all those data centers. But if you're in finance, tech, consulting, marketing — the jobs that used to be the fast track to the upper middle class — the math has changed. You see this most starkly in what's happening to young people entering the workforce. Entry-level job postings have dropped twenty-nine percentage points since January twenty twenty-four. That's not a rounding error. Nearly one-third of starting positions have vanished from the American job market in under two years. Only thirty percent of twenty twenty-five college graduates found full-time work in their field. The unemployment rate for recent grads has climbed to nine point seven percent — now equal to the rate for people with only a high school diploma. For the first time in decades, a college degree offers no statistical advantage in finding work. A degree used to be a ticket to the middle class. Now it's barely a ticket to the lottery. And here's what makes this structural, not cyclical. The tasks that used to train young workers — research, drafting, analysis, the grunt work that teaches you how an organization actually functions — are increasingly handled by AI. Companies aren't just hiring fewer entry-level workers. They're eliminating the entry level itself. The first rung of the ladder is being automated before anyone can grab it. So what do we do with this? The honest answer is that this is a policy question, not a self-help question. The way the gains from AI get distributed isn't written in stone. It's being decided right now. Oregon passed something called the POWER Act last June. It creates a separate rate class for data centers so infrastructure costs get charged to the companies that need them — not to families trying to keep their lights on. Amazon lobbied hard against it. It passed anyway. In January, thirteen governors across the mid-Atlantic signed a statement pushing data centers to cover their share of grid costs. Microsoft announced a "Community-First AI Infrastructure" pledge — higher electricity rates, no property tax breaks, a commitment that building a data center won't raise your power bill. Whether these promises hold remains to be seen. But the fact that the largest companies in the world feel the need to make them tells you something. Here's what gives me hope. Vanguard's research found that the hundred occupations most exposed to AI are actually outperforming the rest of the job market. Wages in those roles grew three point eight percent over the last two years. All other occupations? Less than one percent. Now, some of that is people building AI infrastructure — the engineers and technicians wiring up those data centers. But a lot of it is people in ordinary jobs who figured out how to use these tools before their employers mandated it. The technology isn't replacing everyone. It's creating a widening gap between those who adapt and those who don't. The playbook is changing. The thirty-year path — degree, job, ladder — is being rewritten in real time. That doesn't mean it's dead. It means the rules are shifting. And the people who understand what's happening have an edge over those who don't. One thing you can actually do: find out if your state has a separate rate class for data centers. Oregon does. Most states don't. That's a policy choice your legislators made — or didn't make — and you're paying for it every month. Next time you see a headline about record profits or GDP growth, ask: where did the money go? Because right now, it's going to chips. To servers. To data centers that can vanish from the grid in milliseconds, leaving everyone else to absorb the surge. Stay informed, Stay curious, and we'll see you tomorrow.