01 · The gap
Workers make more. They don't get more.
Every wave of technology has made American workers more productive. Since 1979, the gains from that productivity have mostly gone somewhere else. AI arrives on top of a gap that is already 46 years wide.
Growth since 1979
Economic Policy Institute, 1979 to 2025. Productivity is net productivity of the total economy; pay is for production and nonsupervisory workers.
03 · The build-out
The biggest bet in corporate history.
Four companies plan to spend more than $700 billion this year on data centers, chips and the power to run them, nearly double last year. That is about $2,100 for every person in the United States.
2026 capital spending plans
Company guidance from mid-2026 earnings, via Sherwood News. Per-person figure divides the roughly $710 billion midpoint by about 340 million people.
Data centers' share of U.S. electricity
Lawrence Berkeley National Laboratory for the Department of Energy, December 2024. Scale runs to 12%.
Who pays the power bill?
Data centers used 176 billion kilowatt-hours in 2023, triple the level of a decade earlier. Utilities build new plants and lines to serve them, and unless regulators say otherwise, households share the cost.
04 · The jobs
The first rung is disappearing.
AI has not ended work. It is already changing who gets hired, and the first people squeezed out are the young workers trying to get started.
Challenger, Gray & Christmas, September 2026 (announced cuts, not confirmed layoffs) · Brynjolfsson, Chandar and Chen, Stanford Digital Economy Lab, August 2025 · Federal Reserve Bank of New York
Of every 100 jobs, how many are exposed to AI?
International Monetary Fund, January 2024. Exposure means AI can do part of the job; it does not mean the job is lost.
Exposure is not destiny. Whether AI makes a job better or makes it disappear depends on who decides how it is used: the workers who know the job, or managers told to cut costs.
The Stanford researchers found that older workers in the same exposed jobs held steady or gained. The cuts landed on hiring, on the entry-level positions where people learn a trade and build a career.
2025 for comparison
Employers cited AI for 54,836 announced cuts in all of 2025. In the first nine months of 2026 the count more than doubled.
Challenger, Gray & Christmas, 2025 year-end report
05 · The robot boss
When the algorithm is your manager.
Software already screens résumés, sets schedules, tracks keystrokes and flags workers for discipline. Most workers have no right to know when a machine made the call, or to appeal to a person.
U.S. workers on AI in the workplace
Pew Research Center, survey of 5,273 employed U.S. adults, October 2024, published February 2025
Vetoed, then won
California's first No Robo Bosses Act (SB 7) was vetoed by Governor Gavin Newsom in October 2025 as overly broad. Lawmakers came back with a narrower bill, SB 947, and Newsom signed it on September 30, 2026. Starting July 1, 2027, employers may not rely solely on an automated system to discipline or fire a worker, and a human must review the decision.
Fisher Phillips, October 2025 · SHRM, October 2026
Without rules, the default is simple: the company that buys the software decides how it watches, ranks and replaces the people who work there.
06 · The lobby
The industry wants to write its own rules.
As states began passing protections, the AI industry flooded Washington with lobbyists and its political allies moved to block state laws altogether.
OpenSecrets data via TechCrunch, July 2024 · NBC News, 2025 · Bloomberg Government, January 2026 (FEC filings)
The turn
The future isn't automatic. Someone decides who it's for.
AI is trained on what people wrote, drew, coded and said. It runs on a grid the public pays for. The people whose work built it should have a say in how it is used, and a share of what it earns.
07 · The fix
Make the machine work for workers.
- A seat at the table. A right to bargain over how AI is brought into a workplace, before it arrives, with protection for workers who organize.
- No robot bosses. Notice when AI is used in hiring, scheduling, pay or discipline, and a human decision with a right to appeal before anyone is fired.
- Limits on workplace surveillance. No tracking off the clock, no selling worker data, and no monitoring that a worker cannot see.
- Share the gains. Tax policy that stops rewarding companies for replacing workers, and a public share in AI built with public research, data and power.
- A real safety net. Modern unemployment insurance, wage insurance and free training for workers whose jobs change or disappear.
- Data centers pay their own way. No shifting grid costs onto household bills, and local approval for new projects.
- Keep states free to act. No federal preemption of state protections unless Congress passes stronger ones.
Proof it works
After a five-month strike in 2023, the Writers Guild of America won some of the first AI protections in any union contract: AI cannot write or rewrite scripts, AI output does not count as source material, studios must disclose AI-generated material, and no writer can be forced to use it.
VentureBeat and NBC Los Angeles, September 2023
08 · The mandate
Americans want rules, even if it means slower.
What should the government prioritize?
Gallup and the Special Competitive Studies Project, 2025. Support for rules was 88% among Democrats and 79% among Republicans and independents.
The public is not split on this. The money is. Lawmakers who take nothing from the AI industry are free to side with the people they represent.