The Real Cost of AI: Higher Bills, Lost Jobs, and What's Actually Safe

You've probably heard two things about artificial intelligence lately. One, that it's driving up your electric bill. Two, that it's coming for your job, or maybe already took someone else's. Both of those things are true. But the details matter, and most of what gets said about this online is either exaggerated or missing the point entirely.

Here's a plain English breakdown of what's actually happening with AI data centers and AI jobs right now, based on the most current numbers available.

What Is a Data Center, Anyway?

Think of a data center as a giant warehouse packed with computers, running around the clock. That's basically what it is. Every time you use an AI chatbot, every time a company processes data using AI, and every time a cloud service stores or moves information, it's running through a facility like this somewhere.

These systems take an enormous amount of electricity to run. And all those computers running at once generate heat, so the buildings need constant cooling, usually with water or industrial air conditioning, to keep from overheating. That combination, huge power draw plus huge cooling need, is where the trouble for regular families starts.

Your Electric Bill Is Paying for This, Even If You Don't Live Near One

Nationally, residential electricity prices have climbed more than 36% since 2020. Data centers aren't the only reason, but energy experts increasingly point to them as the biggest new source of demand. The U.S. Department of Energy estimates data centers could eat up somewhere between 6.7% and 12% of all U.S. electricity by 2028, up from just 4.4% in 2023.

Here's why this shows up on your bill even if you're nowhere near a server farm. When a data center demands huge amounts of power, utilities either build new power plants and power lines, or pay a lot more to buy electricity during high demand periods. In many states, regulators allow those costs to be spread across every ratepayer, not just the tech company that needed the power in the first place. So your bill can go up to help pay for infrastructure built for somebody else's server farm.

Some places are getting hit a lot harder than others. In Virginia, home to the biggest concentration of data centers in the world, these facilities now make up roughly 40% of the state's total electricity use. Dominion Energy, the state's main utility, proposed its first base rate increase since 1992 this year, adding about $8.51 a month to a typical family's bill. In the regional power market covering 13 mid-Atlantic and Midwest states, wholesale capacity costs jumped 174% in one year, largely blamed on data center demand. One analysis found some communities near data centers have seen electricity costs rise as much as 267% compared to five years ago.

Even the tech industry seems to know this is a real problem. Microsoft announced this year it will ask to pay higher electricity rates in areas where it builds data centers, specifically so local families don't get stuck with the bill. When a company that size offers to pay more, it's a sign the backlash is real.

The Water Problem Is Just as Serious

The cooling side is hitting hardest in places already short on water. A single large data center can use up to a billion gallons of water a year, and as much as 2.7 million gallons in one day during peak summer heat. That's enough to fill about 180 swimming pools.

Multiply that across the country's more than 5,400 data centers and the totals get staggering. One federal estimate put the indirect water use of U.S. data centers, meaning water used by the power plants that supply their electricity, at 211 billion gallons in 2023 alone. Northern Virginia's data centers used close to 2 billion gallons that same year, a 63% jump from just four years before. In Texas, data centers pulled more than 50 billion gallons of water in 2024, enough to supply a city the size of Austin for months.

This has hit at the worst possible time in states already dealing with drought. This year in North Carolina and Virginia, cities asked residents to take shorter showers and only water their lawns on certain days, while in many of those same areas, data centers faced no special restrictions at all, because most states haven't passed laws requiring it. The pushback has been real. More than 20 states are now weighing bans or limits on new data center construction, and more than $130 billion in planned projects were delayed or scrapped in just the first three months of 2026, more than all of 2025 combined.

Bottom line: the AI boom is real and it isn't slowing down. The question everyone is fighting over right now is who pays for it. The companies making billions off it, or everyone else's utility bill.

Jobs AI Has Already Replaced

This part isn't a prediction. It's already happened.

Customer service, tier one. Basic questions like order status, password resets, and account lookups are now handled by AI chatbots at most major companies. Klarna, the payment company, built an AI system that did the work of about 700 human agents. Salesforce's CEO has openly said the company cut roughly 4,000 support roles through AI. Companies using AI chatbots now report handling 70 to 80% of customer questions with no human involved at all.

Data entry. This is considered the most exposed job category in the country right now. Software tools can read and process thousands of documents an hour with a fraction of the errors a person makes. Estimates suggest as many as 7.5 million data entry and administrative jobs could disappear by 2027.

Telemarketing and outbound sales calls. AI calling systems can now hold a real sounding phone conversation, handle objections, and book appointments. Several companies now run call operations almost entirely through AI voice agents that work around the clock for a fraction of the cost of a human team.

Basic bookkeeping and routine paperwork. Software now handles a lot of the repetitive number crunching that used to require a person, especially processing invoices and flagging errors.

Corporate layoffs tied directly to AI. Amazon eliminated 14,000 corporate roles in the past year, citing AI as a reason leaner teams are possible. Workday cut about 1,750 jobs, roughly 8.5% of its workforce, to shift resources toward AI investment. Nationally, AI attributed layoffs are running about nine times higher this year than last, according to a survey of corporate financial officers conducted with the Federal Reserve.

Here's the honest twist worth knowing: this isn't a one way street. Klarna is now rehiring human agents after customers complained the AI couldn't handle complicated or emotional situations. One survey found 29% of companies that laid off workers after bringing in AI ended up rehiring for similar roles. That doesn't undo the job losses that already happened, but it's proof companies are still figuring this out, not executing some flawless plan.

Jobs Still on the Chopping Block

Based on the most credible research available, including a widely cited Microsoft study that scored hundreds of occupations by how much of their daily work overlaps with what AI can already do, these are the jobs facing the highest exposure going forward:

1. Interpreters and translators

2. Customer service representatives

3. Writers and authors

4. Sales representatives (services)

5. Telemarketers

6. Data entry clerks

7. Bookkeeping and accounting clerks

8. Ticket agents and travel clerks

9. Technical writers and editors

10. Market research analysts

Notice what isn't on that list. No doctors, no nurses, no electricians, no plumbers. The common thread among the jobs above is that the work is mostly digital, repetitive, and doesn't require being physically present with another person. That's exactly the kind of work AI is good at right now.

Setting the Record Straight on Healthcare

There's a popular idea floating around that AI is coming for medical jobs. Based on the actual research, that's mostly backwards, and it's worth being precise about, because a lot of people are making career decisions based on the wrong assumption.

The roles genuinely at risk in healthcare are the administrative ones sitting behind patient care. Medical coders and billers, health information technicians, and other roles built around processing paperwork and structured data. The World Economic Forum estimates up to 25% of healthcare administrative roles could be automated within a decade. If you or your kid is looking at a career specifically in medical coding or billing, it's worth knowing that field is likely to shrink.

But direct patient care is a completely different story. Multiple independent studies, including research from Goldman Sachs and the World Economic Forum, consistently rank hands on healthcare roles among the most AI resistant jobs that exist.

Phlebotomists, the people who draw your blood, were ranked the single most AI resistant occupation in one major 2026 study, because the job requires fine motor skill, sterile technique, and reading a nervous patient in real time, things AI simply can't do.

Nursing assistants, home health aides, physical therapists, and paramedics show similarly low risk. Nurse practitioners are projected to grow by nearly 46% over the next several years, one of the fastest growing careers in the entire economy. Even physicians are classified by researchers as facing what's called "high augmentation," meaning AI will help them work faster, not replace their judgment or their presence in the room.

If someone in your family is worried about AI wiping out a future in medicine, that worry is mostly misplaced if they're headed toward direct patient care, like nursing, therapy, or EMS work. The worry is legitimate if they're headed specifically toward medical billing, coding, or back office health administration with no clinical license attached.

What This Means for You

AI is not some far off, theoretical thing anymore. It's already showing up as a bigger number on your electric bill, tighter water restrictions during a drought, and real jobs that no longer exist the way they did two years ago. At the same time, the doom and gloom headlines claiming entire professions are about to vanish overnight don't hold up against the actual data.

The honest picture is this. AI is eliminating specific tasks and specific roles right now, mostly ones built around repetitive, predictable, digital work. It is not, at least not yet, replacing the jobs that require being physically present, using hands on judgment, or building trust with another human being. If you or someone you love is weighing a career move, the safest ground right now is skilled trades, direct healthcare, and any work where showing up in person and using real world judgment is the whole point of the job.

We'll keep tracking this as it develops. This isn't a one time story. It's the new shape of the economy, and it deserves ongoing attention, not a single headline.

---

Sources: U.S. Energy Information Administration, U.S. Department of Energy, Environmental and Energy Study Institute, Bloomberg analysis of data center electricity costs, TechTimes and CNBC reporting on AI data center power costs, WRAL and Virginia Mercury reporting on data center water use and drought conditions, Goldman Sachs Global Investment Research, World Economic Forum automation estimates, Duke/Federal Reserve Banks of Atlanta and Richmond CFO Survey, Microsoft Research occupational AI applicability study, and reporting on corporate AI layoffs at Amazon, Workday, Salesforce, and Klarna.