America’s next labor shock is demographic. Census projects the 65+ population to jump from 57.8 million in 2022 to 71.2 million in 2030, up 23%. The population ages 18 to 64 barely moves: 203.0 million to 204.8 million, up 0.9%. By 2030 seniors outnumber children. For every 100 working-age adults, there will be 68 people outside working age. By 2080, 81. The same country debating how many jobs AI destroys is aging into a labor-supply problem. How much automation can an aging economy absorb before labor scarcity turns into unemployment?

US electricity demand was flat for 17 years. From 2007 to 2024 the grid added 211 billion kWh of annual sales, roughly 0.3% growth a year. EIA now forecasts it adds 236 billion kWh in just 3 years, 2024 to 2027, mostly data centers. More new demand in 3 years than the previous 17 combined. Texas already paused connecting new data centers to its grid, and EIA still expects that region to lead all sales growth anyway. Everything about this system was planned around flat demand: generation, transmission, rate structures. The buildout gets recovered through rates, which means households help underwrite the compute boom long before it pays off. What happens to your power bill when the grid has to compress 17 years of growth into 3?

The New York Fed just published Q2 2026 household debt data. Student loans: $1.65 trillion outstanding, with 10.6% of balances at least 90 days past due. That is roughly $175 billion in serious delinquency, up from 10.3% last quarter. For all household debt combined, 4.7% sits in any stage of delinquency. Student debt is the outlier by a wide margin. Borrowers financed degrees on the promise of entry-level white-collar work. AI is absorbing exactly that work. If the first rung of the career ladder keeps disappearing, who repays the $1.65 trillion?

BLS just published its 2025-35 jobs outlook. Office and administrative support: 18.8 million jobs in 2025, 18.0 million in 2035. Down 752,100 jobs. Customer service reps lose 141,800. Bookkeepers lose 85,600. Data entry loses 33,600, one quarter of the occupation. The US still adds 5.9 million jobs overall. The growth is somewhere else. Every one of these workflows sits directly in AI's path. Where do 752,100 displaced workers go when the white-collar entry ramp disappears?

US workers now produce 68.5% more per hour than they did in 2000. Their inflation-adjusted compensation per hour rose 23.9%. Both figures come from BLS nonfarm-business data through Q2 2026. Productivity grew nearly 3 times as fast as real compensation. Since ChatGPT launched, the gap is still widening: productivity +9.1%, real compensation +4.6%. AI is supposed to unlock another productivity boom. The last 26 years already did. Who captures the next one: workers, shareholders, or customers?

The Fed just updated its wealth distribution data. Top 1% share of US net worth, Q2 2026: 32.5%. The highest reading in the 37 years the Fed has tracked it. The bottom 50% of households holds 2.3%. Do the math and one percent of households owns 14 times the combined wealth of half the nation. In 1989 the top 1% held 22.8% and the bottom half held 3.5%. I think the AI buildout accelerates this. Asset owners capture the productivity gains, and the assets doing the work never ask for a raise. If this line reaches 40%, what does the country look like?

The labor market quietly froze. BLS JOLTS, July 2026: hires rate 3.2%. In 25 years of data, the only months lower were the Great Recession (late 2008 through 2010) and the April 2020 lockdown. Quits are down to 1.9%. People who have jobs are not letting go of them. Job openings still sit at 4.4%, a level that used to signal a hot market. Companies keep postings up. They just stopped filling them. I think this is the labor market AI walks into. Every resignation is now a chance to automate the seat instead of refilling it. No layoff headline required. How many of those posted-but-never-filled seats quietly become software?

AI is splitting software work into winners and losers. BLS's new 2025-35 outlook projects computer programmer employment down 7%. The agency explicitly says companies will use AI to automate repetitive programming tasks. Meanwhile software developers, QA analysts, and testers are projected to grow 10%. Median pay for programmers is $100,390. For software developers: $135,980. The $35,590 premium sits on owning a broader share of the problem. If AI writes more of the code, does "programmer" disappear while "developer" gets more valuable? Or does the whole stack eventually compress?

The Census Bureau just released 2025 income data. Real median household income: $87,460. In 2019: $85,320. That's a 2.5% gain in six years. Over the same span, real GDP per person rose from $62,677 to $69,749, up 11.3%, per BEA data. The economy produced 11% more per person. Real median household income moved 2.5%. AI is supposed to make the productivity curve steeper. If ownership stays concentrated, output can surge while the typical household barely moves. Who gets the next productivity boom: workers, consumers, or asset owners?

The sector building and carrying the digital economy has lost almost 1 in 8 jobs since ChatGPT launched. US information employment fell from 3.115 million in November 2022 to 2.745 million in August 2026, per BLS. That's 370,000 jobs gone, down 11.9%. The sector includes software publishers, telecom, data processing and hosting, web search, media, and information services. Total US payrolls added 4.8 million jobs over the same period. The industry's unemployment rate is now 5.7%, versus 4.1% nationally. Is this AI replacing information work first, old media and telecom shrinking, or one sector hiding several different stories?

Deaths now outnumber births in 17 U.S. states. In 2011, it was two: Maine and West Virginia. In the Census Bureau's latest state estimates, natural decrease reaches one-third of states. The U.S. still grew by 1.78 million people in 2025. Births minus deaths contributed 519,000. Net international migration added 1.26 million, 71% of total growth. This is the demographic model now: more states shrinking naturally, national growth carried by migration. Is that a temporary post-pandemic distortion, or America's permanent population math?

8.8 million Americans now work more than one job. That's up 1.16 million since November 2022, a 15.2% increase, per BLS data through August. Total employment grew 2.7% over the same period. Multiple jobholding is growing nearly six times faster than employment. The share of workers holding more than one job rose from 4.8% to 5.4%, about 1 in 19. The AI economy arrived alongside a sharp expansion of the second-shift workforce. What explains the surge: flexibility, rising costs, unstable work, or one paycheck no longer covering the math?

US software publishers produced 30% more real output in 2025 than in 2022. Employment grew 2%. The industry's labor productivity rose 25% over those three years. In 2025 alone it jumped 12.9%, the fastest increase since 2004, per BLS Industry Productivity data. Software is the cleanest early test of whether AI lets companies scale output without scaling teams. Does demand eventually pull hiring up, or is 2% headcount growth the new shape of a 30% output boom?

The federal minimum wage turned 17 this summer. It has been $7.25 since July 24, 2009. Prices are up 55.55% since then. That $7.25 now buys what $4.66 bought when the rate was set. Keeping the same purchasing power would require $11.28. Average private-sector hourly pay rose 70.20% over the same span. The legal floor rose 0%. AI is arriving while the national wage floor is worth 35.71% less. Is the minimum wage still a labor-market rule, or has inflation quietly repealed it? Source: U.S. Department of Labor and BLS.

Here is the white-collar AI scorecard since ChatGPT launched: Legal services: +65,500 jobs Management consulting: +75,800 Architecture and engineering: +140,300 Accounting, bookkeeping and payroll: -9,400 Net across the four sectors: +272,200 jobs, per BLS data through August. A broad headcount wipeout has not reached these professional-service payrolls. Productivity may be rising. Work may be changing. Headcount is still here. Are the layoffs delayed, or is AI expanding demand faster than it removes labor?

From Q4 2022, when ChatGPT launched, through Q1 2026, US household wealth grew $38.5 trillion. The top 10% captured $28.3 trillion, 74% of the entire gain. The top 1% alone gained $14.6 trillion. The bottom 90% gained $10.2 trillion. The top 1% gained $4.4 trillion more than the other 90% of households combined. Straight from the Federal Reserve's Distributional Financial Accounts. This is the ownership structure the AI economy is landing on. If AI creates a historic new pool of value, does this structure distribute it or concentrate it?

From Q4 2022 through Q2 2026, US nonfarm business output rose 10.0%. Total hours worked rose 0.9%. Output per hour rose 9.1%. Real compensation per hour rose 4.6%, per BLS data on FRED. The economy produced 10% more with less than 1% more human time. Workers' hourly purchasing power captured about half the productivity gain. This is the promise and the fault line of AI in one chart. More output. Almost no added labor. A smaller wage gain. If machines do more of the work, who owns the productivity dividend?

Auto insurance costs are up 48% since January 2020, even after falling 5% over the past year. Overall prices rose 30% over the same stretch. Average hourly earnings rose 33%. Insurance outran both, per BLS data through August. A worker whose paycheck kept pace with inflation still lost 15 percentage points to the cost of insuring a car. For households that need a car to work, mobility is a bill with no practical substitute. The price comes from risk models consumers cannot inspect or negotiate. What's driving the gap: repair costs, crash severity, climate risk, or pricing power? And how would a consumer know?

Since ChatGPT launched, America added 4.8 million payroll jobs. Computer systems design and related services lost 120,000. The sector peaked at 2.484 million workers in March 2023. It is down 4.9% to 2.363 million now, per BLS data through August. This is the workforce that installs software, integrates systems, and turns new technology into business operations. AI arrived. Its payroll shrank. A technology boom can create enormous output without creating broad employment. That may be the entire point. Is this productivity finally arriving, work moving offshore, or AI deleting the implementation layer?

America's loyalty tax is back. In August, job switchers saw 5.0% median wage growth. Workers who stayed got 3.6%. The 1.4-point gap is the widest since early 2023, per the Atlanta Fed Wage Growth Tracker updated this week. Companies say talent is their most valuable asset, then reserve the market-rate budget for people they haven't hired yet. The worker who knows the systems gets less than the stranger replacing them. AI makes internal knowledge cheaper to reproduce and external hiring easier to test. That could widen the tax. Why does the labor market still pay people to leave and punish people who stay?