1.    The energy shock that will not stay buried.

Gasoline is the price people see everywhere. Diesel is the price they rarely see but shows up everywhere else.

Diesel powers the trucks that stock grocery shelves and the equipment farmers use at harvest. A diesel shock does not stay at the pump; it courses through freight costs, food, construction and nearly all goods that must travel to reach us.

The psychology of fuel price shocks is more combustible than before the pandemic. Consumers and firms no longer see a jump in fuel prices as an isolated event. Five years of escalating costs have created muscle memory. The reflex is to brace for the next increase before the last one fades.

States are suspending fuel taxes ahead of the midterms. That could temporarily dampen price increases but would face an uphill battle against higher refining costs and create a hole in state coffers.

The larger issue is the damage done to global refining capacity. Repairs and deferred maintenance could buoy diesel prices long after the price of crude oil falls. Nearly half of the 100 oil executives that the Dallas Fed recently surveyed worried that the crisis in diesel fuel will take a full year to unwind.

2. The monster in healthcare benefits.

This bout of inflation is more than an energy story; it is a service sector story. Aging is lifting demand for care, while curbs on immigration are worsening labor shortages in the care economy.

Medicaid cuts leave providers less able to raise wages and fill vacancies, so care is rationed. Eldercare is taking the largest hit: in-home care costs are already rising at a double-digit pace.

Healthcare inflation often outlasts the shocks that ignite it. Medical costs kept rising after the recessions of the 1980s broke the back of broader inflation. Cheaper imports and stronger productivity lowered goods prices in the 1990s, but healthcare costs remained elevated.

San Francisco Fed research explains the hard trade-offs: healthcare prices respond more to insurance industry specific forces than to the business cycle. Those factors insulate them from rate hikes, which means that the Fed must hammer goods prices to offset increases in healthcare costs.

Lowering goods prices was easier when cheap imports were flooding into the country. That is no longer the case. Even imports from China, including data center inputs, are rising in price.

The pressure is slow moving because health costs reset through annual contracts. Insurance premiums are poised to increase 14% for small businesses in 2027, which raises those costs more than 25% in two years.

The fallout extends beyond healthcare. Escalating benefit costs are eating into wage gains, while the ranks of those providing unpaid eldercare are swelling.

3.    The Fed hunts a ghost it can’t catch.

The Fed cannot pump oil or lower healthcare costs. It can only hammer demand to meet a supply-constrained world.

The trade-off is brutal. The Fed must lower goods prices to offset increases in prices farther from its reach. That means dealing a harder blow to home values, autos, business investment and employment through higher rates.

Central banks usually warn of the pain associated with rate hikes at the start of a tightening cycle. Fed Chairman Kevin Warsh took a different tack in September. He emphasized the economy’s remarkable resilience as the Fed raised rates in September.

Forecasts released by Warsh’s colleagues at that meeting reveal that most do not expect inflation to return to the Fed’s 2% target until 2029 - a full year later than projected in June. Warsh did not participate in the forecast, but he has leaned heavily on an AI-driven productivity boom to derail inflation.

That is a hope, not a promise. There is a catch. Data center construction has proven less sensitive to rate hikes, while the costs are arriving before the productivity gains that could offset them. The risk is that the Fed has to raise rates more aggressively to derail inflation, dealing a greater blow to employment.

4.    Bond vigilantes rise from the crypt.    

Something else is stirring. When long-term Treasury yields rise, inflation and expected Fed policy may not be the entire story. Investors demand compensation for fiscal risk, policy uncertainty and the sheer stock of debt they are expected to absorb..

Research repeatedly links larger projected deficits to higher long-term interest rates. The direction is clear: more borrowing lifts yields, raises interest costs and widens future deficits. It is a vicious cycle.

What was once an emerging market concern has spread to developed countries in recent days. The difference between French and German 10-year rates is at its highest level since the European sovereign debt crisis in 2011.

Households eventually meet the bond market through mortgage rates and other long-term borrowing costs. The long end can tighten financial conditions even if the Fed is no longer raising short-term rates.

The Fed can tame the inflation premium, not the flood of debt. Foreign investors still want US assets but are more selective, often favoring equities over Treasuries in search of better returns.

5.    AI arms race adds to our debt woes.

AI investment is substantial. So is the productivity potential. The question is increasingly how we are paying for it.

For years, the largest tech companies could finance investment from vast internal cash flows. The scale of the AI buildout is changing that. Debt markets, private credit and increasingly creative structures are becoming part of the financing mix.

Debt changes the risk. Companies spending their own cash can absorb mistakes; borrowers must deliver the expected returns. An arms race raises the stakes because one company’s spending becomes another’s reason to spend more.

History offers a warning without denying the innovation. Railroads, electricity and the internet transformed the economy, but not every investment earned a return. Transformative technology does not guarantee that every participant will win.

Bubbles are common. They also pop. That matters when growth depends heavily on the AI boom and the wealth it creates. The Magnificent 7 tech behemoths accounted for about 34% of S&P 500 market capitalization in September. A correction could hit affluent spending and investment at once.

6.    AI backlash on the ballot in November.

Finance is only one vulnerability; the buildout needs public consent. Data centers need enormous amounts of electricity and often water. For local communities, the calculation is intensely practical: Who benefits and pays?

A household facing a higher electricity bill does not view a data center as an abstract investment in America’s future. Communities see land use, water and transmission demands, followed by few permanent jobs once construction ends.

Voters will decide 35 data center referendums across eight states on November 3. Most reached the ballot through citizen petitions, showing how quickly the issue became a grassroots fight.

That is a floor, not a ceiling. Even states without a referendum have erected hurdles to data centers.

Regulation trails the technology, but the backlash is gaining speed. November will test the AI boom’s political license to operate. Restrictions need not stop the buildout to slow it; delays and demands that developers absorb more grid and water costs can squeeze margins and stretch timelines.

There is a potential pressure valve. Data centers that cut power use when the grid is strained are different from those that compete with households at peak demand. Flexible demand can unlock capacity and make data centers partners rather than permanent strains on the grid.

Power constraints are accelerating investment in efficient chips, cooling systems and smaller facilities. That can cut delays and improve efficiency, but soaring demand could overwhelm the savings. The buildout needs enough productivity gains to justify its costs.

7.    Productivity: trick or treat?

Productivity is the escape hatch. In theory, it allows wages to rise without stoking inflation. We are not there yet.

The firm-level evidence on generative AI is encouraging. Studies show meaningful productivity gains, especially for novice and lower-skilled workers. AI can help employees learn faster, handle more complex tasks and narrow performance gaps.

The economy-wide evidence is less dramatic. Training, software and organizational change must come before the gains show up in measured output.

AI can transform individual firms well before the gains appear in national productivity statistics; the test is whether those gains become large, durable and broad enough to justify the investment boom.

AI helps companies change prices faster, but not necessarily lower them. Recent research shows pricing models can track competitors and react instantly.

When firms use similar models, they may keep prices higher without explicitly coordinating. AI can make it harder for consumers to do price comparisons and keeps prices elevated. The gain in productivity boosts profits more than paychecks, as we have already seen.

8.    The gains that fail to boost paychecks.

This may best explain why resilience has failed to resonate. If productivity gains flow to profits and owners of capital rather than to workers’ wages, discontent deepens.

Several leaks separate productivity from pay. Weak bargaining power shifts gains toward profits. AI may initially favor capital and highly skilled workers. Rising healthcare costs absorb compensation that might otherwise appear in wages.

The last point is easy to miss. An economist can tell a worker that employer-paid health insurance is compensation. The worker can just as easily respond that a larger deductible does not feel like a raise.

The economy can therefore produce more while many households feel they are receiving less. That gap between measured compensation and lived experience helps explain weak confidence. For younger workers, the problem is more basic: AI may narrow access to the first job itself.

9.    Missing a first rung.

Some of the most consequential labor market changes are invisible because they involve something that never happens. An entry-level job is never posted. A new graduate is not hired. There is no layoff announcement because there was never an employee to lay off.

Stanford research finds hiring has weakened among younger workers in occupations most exposed to AI. The pattern is descriptive, not proof that AI acted alone; tight monetary policy and hiring freezes explain some of it. For a graduate unable to land a first job, the distinction offers little comfort.

Entry-level jobs provide income, industry knowledge, professional judgment and the relationships workers need to advance. AI may make novices more productive once hired while reducing the number of entry-level workers firms need. Fewer entry-level hires today mean fewer seasoned workers tomorrow.

The weakness extends beyond new graduates. Unemployment remains low partly because labor supply and demand are slowing together. Quits rates have plummeted in sectors most exposed to AI; uncertainty about how work will change has made employers more hesitant to hire.

The labor market looks stable from a distance but frozen up close. Long-term unemployment remains elevated despite muted layoffs.

10.    The K-loop and the missing social contract.

Higher income households account for a disproportionate share of consumer spending. They are more likely to own financial assets and benefit from higher yields on savings. Borrowers feel the other side of higher rates.

That produces the K-loop. Affluent households sustain spending. Demand remains stronger. Inflation proves harder to extinguish. Fed policy stays tighter. Higher rates bear down most heavily on rate-sensitive sectors and indebted households.

Meanwhile, the benefits of the AI investment boom do not necessarily accrue to the same households bearing the adjustment costs through higher power bills, insurance costs and borrowing rates.

That is more than a distributional problem. It threatens the political durability of the investment boom. Creative destruction creates conflict. A functioning social contract determines whether that conflict becomes progress or backlash..