2026 delivered no recession: GDP grew 2.5% and 2.2% in the first two quarters and unemployment stayed in a 4.1%–4.3% band. The absence of a downturn is not the absence of risk. Three exposures matter heading into 2027: sticky inflation, AI returns that must eventually justify the capital, and a fiscal position with little room to respond.
This is the fourth and last of four letters published together in October 2026. The first three covered the rate regime, the multi-speed expansion and the energy shock. This one pulls them together into a year-end preview: what 2026 actually delivered, why the recession many expected has not arrived, and which three risks matter most going into 2027. The central point is that the absence of a recession is not the absence of risk — it changes where the risk sits.
Real GDP grew 2.5% in the first quarter and 2.2% in the second. Unemployment has stayed in a narrow 4.1%–4.3% band since March, reaching 4.2% in September, and as this site's October monthly insight documented, jobless claims, manufacturing and credit markets stayed calm. The Fed reversed course: after a cutting cycle that began in September 2024, it raised rates on September 16 to 3.75%–4.00%. An energy shock lifted headline inflation to 3.4% while core stayed at 2.4%. Mortgage rates rose to 7.28%. And in the last data point of the year so far, September payrolls added only 29,000 jobs.
| 2026 so far | Data point | Read-through |
|---|---|---|
| Growth | Real GDP +2.5% (Q1), +2.2% (Q2) | No recession |
| Labour | Unemployment 4.2%; payrolls avg. ~45,000/mo over 12 months | Softening |
| Policy | Fed hiked to 3.75%–4.00% on Sept. 16 | Cuts replaced by hikes |
| Inflation | Headline CPI 3.4%, core 2.4%; core PCE rising 12 straight months | Sticky |
| Rates | 10-yr ~5.28%; 30-yr mortgage 7.28% | Highest in years |
| Energy | Brent ~$101; gasoline CPI +27.4% | Shock persisting |
Three things have held the expansion up: AI capital spending, which Carson Group puts at about 45% of growth over the last five quarters; high-income households, who account for nearly half of spending; and a labour market that, until September, absorbed higher rates without breaking. None of those is an obvious shock absorber against what is coming. Each is a source of strength that depends on financial conditions staying benign. That is the sense in which the missing recession is not good news so much as deferred information.
Core PCE inflation has risen month over month for twelve consecutive months. Year-ahead inflation expectations are 4.6% in the Michigan survey, up from 3.4% before the Iran conflict. The reassuring evidence is the Dallas Fed's finding that effects on 5–10 year inflation expectations are tiny, between 0 and 0.07 percentage points. Together they suggest anchored long-run expectations but fragile short-run ones. If energy stays near $100, headline inflation stays elevated and the Fed's room to ease stays narrow; if energy falls, the shock fades and the 2.4% core number becomes the story.
Hyperscaler capital spending for 2026 is now estimated at $688 billion to $740 billion, up from roughly $470 billion early in the year. Spending is demand today; the return has to arrive as revenue later. There is evidence of capability — in one reported test, frontier models clustered near the top of CPA performance on month-end close tasks, though on simplified, well-specified work — but capability is not the same as realised profit. Two things raise the bar. First, financing: at a 10-year yield near 5.3%, hurdle rates are higher than when the build-out began. Second, social license: residential electricity averages about 18 cents per kWh, roughly 37% above 2020, and data centers are a contested part of why. If returns arrive late, the 45% of growth they supply becomes a drag in reverse.
The federal deficit reached about $2 trillion in the first eleven months of fiscal 2026, according to the CBO. Net interest costs rose $111 billion, or 12%, and interest payments now exceed defence spending, with national debt above $40 trillion. Corporate tax receipts fell 25% after the 2025 tax changes. The November 3 midterms will shape what is politically possible, and this letter takes no view on the outcome. The analytical point is that however control is distributed, the arithmetic leaves little room for a large fiscal response to a downturn: divided control tends to make big measures harder, while unified control makes them easier but does not shrink the interest bill.
This is a setup, not a forecast. Several things that will decide how 2026 ends have not happened yet: September CPI on October 14, the Fed's October 27–28 meeting, the midterms on November 3 and the October jobs report in early November. Any of them could change the balance of these three risks. The grouping into inflation, AI returns and fiscal constraint is also this publication's selection; other risks (credit events, a policy mistake, a geopolitical escalation) are not ranked here.
A recession that does not arrive is information about the economy's shock absorbers, not a guarantee about the next shock.
The risk to this letter's view is that 2027 turns out simpler than it looks: energy settles, inflation falls toward 2.4%, AI productivity shows up in earnings, and a manageable fiscal path emerges. In that world, a defensive stance leaves returns behind. The opposite risk is that two or more of these pressures hit together — sticky inflation, a financing squeeze on AI spending and a political stalemate over the response — in which case the missing recession of 2026 becomes the delayed one of 2027.
2026 delivered growth without a recession and without relief: inflation did not fall, rates did not come down, and the expansion leaned on a narrow set of engines. The risks that matter for 2027 are not hidden. They are inflation persistence, the return on a historic AI investment, and a fiscal position with little slack. Position for the regime you can observe, keep flexibility for the three signposts above, and treat the next set of data releases as the real year-end review.