Through September, U.S. employers announced 573,195 job cuts. Artificial intelligence was cited in 120,136 of them, more than any other reason this year, according to outplacement firm Challenger, Gray & Christmas. Tech companies alone announced 165,925 cuts, up 54% from the same period last year.

To be fair, total announced cuts are down 39% from the first nine months of 2025, and AI was cited in only about 9% of September’s cuts. The bigger effect is on hiring. Using ADP payroll data, Stanford economists Erik Brynjolfsson, Bharat Chandar and Ruyu Chen found that as of June, employment among 22- to 25-year-olds in the most AI-exposed occupations was about 19% below where it would be if it had kept pace with young workers in less-exposed jobs. A year earlier, the gap was 15%. They write that the adjustment works “primarily through reduced hiring of young workers rather than increased separations.”

Companies aren’t firing junior workers; they are simply not hiring them. They’re using efficiency as the excuse.

A great economy for whom?

On paper, the economy looks powerful. Real GDP grew at a 3.4% annual rate in the third quarter, up from 2.2% in the second, according to the Bureau of Economic Analysis. Spending on computer equipment, a soft landing for capital expenditure, was a main driver.

But for the common citizen, this year has been anything but a soft landing. Consumer prices rose 3.7% in the year through September, and gasoline rose about 34%. Since the Iran war closed the Strait of Hormuz in March, oil has gone through what the World Bank calls “the largest oil market shock in history.”

In addition, workers now pay an average of $7,471 a year toward family health coverage, up 9% from 2025, according to KFF. Over roughly the same period, wages increased only 3.7%. Effectively, the people took a pay cut.

The looming electricity and water crises

The AI buildout is also competing with households for the same electricity. In PJM, the largest U.S. grid operator, the independent market monitor estimated that data centers caused 63% of the price increase in 2025/2026, about $9.3 billion in costs, according to the Institute for Energy Economics and Financial Analysis. Pepco residential customers in Washington, D.C., saw bills rise about $21 a month starting in June 2025. Roughly $10 of that came from capacity prices.

This is what makes me angry. None of these issues requires new technology to be solved. The models exist, the compute exists, and the capital is already being spent. What’s missing is a decision about where to point it.

The American Society of Civil Engineers estimates a water main breaks every two minutes in the U.S., and about 6 billion gallons of treated water is lost daily. We could use AI to forecast load and balance the grid so data centers pay for the strain they add. We could use AI to find leaks in water systems. We could model drought before it reaches a farm, inspect bridges, harden supply chains, and cut diagnosis costs instead of raising premiums.

In June, in the U.K., Northumbrian Water reported that a six-month trial using AI-powered acoustic sensors to help crews locate hidden leaks was saving nearly 6.4 million liters of water a day. This is the technology we must invest in and deploy immediately.

San Jose Water installed 13,600 fixed acoustic sensors across its 2,400-mile network, and the system found 483 leaks and saved more than 400 million gallons of treated water. Plus, New Mexico’s LeakTracer satellite-plus-AI program identified 216 leaks across 181 miles of pipe in its first year. Completing the needed repairs could recover 328 million gallons and save about $505,000 a year.

This must become the industry standard.

Four steps to secure the future

First, we must treat energy resilience as a national security issue. Data centers must be required via litigation to pay for the grid upgrades and capacity they drive. Families shouldn’t have to subsidize the infrastructure required by multibillion-dollar tech companies. AEP Ohio already requires new data-center customers to reimburse 100% of buildout costs if they cancel before energization. The company creating the demand, carrying the risk, must be fundamental.

In my home state, New Jersey’s Data Center Fair Share Act was signed in July 2026, creating a separate rate class for large data centers. This requires them to cover their own infrastructure costs, demands financial guarantees covering at least 85% of requested load for 10 years, and lets them offset capacity obligations by paying for demand reductions elsewhere. This could save ratepayers more than $1 billion a year. Other states are moving in the same direction, with large-load tariffs imposing high minimum-demand charges and contribution-in-aid rules so residential customers aren’t charged if a project underdelivers or exits.

Next, go after health costs at the source. Use AI to reduce the volume of unnecessary tests, procedures, and late-stage interventions rather than simply shifting higher premiums onto households. Start by capping excessive hospital prices in public employee health plans, with safeguards for vulnerable hospitals. Oregon capped in-network hospital payments at twice Medicare’s rates and saved an estimated $107.5 million over 27 months. Then require publicly funded health care AI projects to demonstrate lower costs without compromising care. Efficiency isn’t helping if the patient’s bill remains the same.

Third, deploy AI to retrain people faster than it replaces them. By tying public incentives for AI investment and data-center build-out to measurable entry-level hiring and apprenticeship commitments, AI will create efficiency and jobs. As entry-level roles disappear for the benefit of only AI owners, the coming generations are being steered toward economic ruin. Tie public AI incentives to net-new paid apprenticeships and entry-level jobs, with payments after six and 12 months of verified employment and not simply after a training course.

This isn’t some pipe dream; the blueprint exists. In July, the Labor Department awarded nearly $162 million through an apprenticeship program that links payments to verified retention and progression milestones. Reward employers for being job creators and builders. Going forward, any firm receiving tax credits, accelerated permitting, or public power-purchase support for AI capacity must fund or host a proportional number of paid apprenticeships or entry-level roles, with progress reported publicly.

Finally, give public infrastructure the same AI investment private data centers get. One of the most impactful improvements AI could make is in utilities and water. As governments embed AI into power grids and water systems to stabilize national infrastructure, we should use AI-driven material evaluation of roads and highways and continuous monitoring to predict repair needs before learning from injuries and accidents. Funding can be mandated by dedicating a share of data-center tax revenue to local infrastructure technology. Virginia began taxing data-center electricity consumption at 1.1 cents per kilowatt-hour in July. States should direct a portion of that revenue toward leak detection, grid management, and bridge inspection, plus paying the employees who act on the findings.

A dedicated public AI infrastructure fund can be created and financed by a small, transparent tax on large data-center power or compute capacity, matched by federal and philanthropic capital.

When voters go to the polls Tuesday, they won’t vote on GDP. They’ll vote on whether things at home got easier or harder since their last ballot was dropped in the box. With AI, the question isn’t whether we can point this technology at the problems that decide whether people can afford to live and create a solution. We can do that. The real question is whether anyone in charge is willing to, or if they just want to keep eliminating the future of work for the next generations.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.