Cosmetically Stable
SGGI | MAY 2026
The economy's vital signs are being taken with a broken thermometer.
The headline unemployment rate held at 4.3% in April 2026. What the headline does not say: total employment fell by 226,000 that same month. The labor force shrank by 92,000. The participation rate dropped to 61.8%, its lowest since October 2021. Six million people who want a job are not counted because they stopped looking. And 4.9 million Americans are employed part-time not by choice but because their hours were cut or they cannot find full-time work, a number that jumped by 445,000 in April alone. The unemployment rate stays cosmetically stable because the denominator keeps shrinking. Every month in 2025 saw negative payroll revisions. The annual benchmark stripped 898,000 jobs from the prior 12 months. The last three months averaged 48,000 jobs added per month. In 2025 as a whole, the monthly average was 15,000.
This is not a strong labor market with rough edges. It is a deteriorating labor market with a well-maintained facade.
The Entry-Level Economy Is Closed
The share of unemployed Americans who are new workforce entrants hit a 37-year high in 2025, peaking at 13.3% in July. That is higher than at any point during the Great Recession. Finance and information services, the industries that once absorbed the majority of college graduates, shed an average of 9,000 jobs per month since 2023. Before the pandemic, those same industries added 44,000 per month. The swing is 53,000 jobs per month, gone, in the exact sectors that constituted the entry-level white-collar economy.
For six months in 2025, workers with a skilled trades associate's degree posted better employment outcomes than college graduates. The first time that has happened since the federal government began tracking the data in the 1990s. Fifty-two percent of the class of 2023 were underemployed one year after graduation. Millennials and Gen Z, the marginal consumer cohort, carry the highest debt-to-income ratios and the worst housing cost burden in modern history, locked out of the income base that was supposed to service that debt.
What the Debt Is Doing
Total household debt stands at $18.79 trillion. Delinquency rates across mortgages, credit cards, and auto loans hit 4.8% of all outstanding debt in Q4 2025, the highest level since 2017, driven by defaults among low-income and young borrowers. Student loan 90-plus day delinquency reached 10.3%. Subprime auto loan delinquency sits at approximately 6.8%, an all-time high.
Klarna, the buy-now-pay-later platform, went public at $40 per share and now trades roughly 70% below that price. The reason is instructive. Interest-bearing consumer loans grew 193% year-over-year. That is not a growth story. That is consumers financing groceries and everyday purchases at interest because they cannot absorb the cost of goods upfront. Credit provisions rose 59% year-over-year. BNPL credit is now being reported to major credit bureaus, meaning the hidden leverage load carried by subprime and near-prime borrowers is about to surface in official debt tallies for the first time. Klarna is not a fintech story. It is a consumer stress indicator wearing a fintech hat.
University of Michigan Consumer Sentiment stood at 49.8 in April 2026. At the peak of the 2008 financial crisis, it was 55.3. The consumer is telling you what they expect to do with their wallet.
The Energy Shock That Has Not Fully Landed
On February 28, 2026, the United States launched military operations against Iran. By March 4, the Strait of Hormuz was closed. The IEA characterized what followed as the largest supply disruption in the history of the global oil market. Brent crude surged from $72 per barrel to $126 at peak. The largest monthly oil price increase ever recorded occurred in March 2026.
The strait remains functionally closed today, with a fragile and repeatedly violated ceasefire in place. Gulf oil producers curtailed output by at least 10 million barrels per day. Qatar's Ras Laffan LNG complex was struck on March 18, reducing production capacity by 17%, with damage requiring an estimated three to five years to fully repair. LNG spot prices in Asia rose over 140%.
Q1 2026 was only partially impacted. Q2 is the first fully impacted quarter and those numbers have not yet reported. The Federal Reserve, facing 3.3% headline CPI and 0.5% Q4 2025 GDP growth simultaneously, is in strategic paralysis. It cannot cut without blowing out inflation. It cannot hike without collapsing an economy already running on fumes. Rates stay higher longer. The cost of the debt-financed capital expenditure driving the entire AI buildout rises with them.
The Land Is Drying Up While the Servers Come Online
Sixty-one percent of the continental United States is in some stage of drought as of May 2026. The Southeast has recorded 96.83% drought coverage, a regional record since monitoring began in 2000. Georgia, North Carolina, and South Carolina have experienced conditions drier than any recorded since 1895. Texas, Oklahoma, and Kansas are in their sixth consecutive year of drought. Winter wheat conditions collapsed from 48% good-to-excellent last year to 35% now. Ninety-five percent of U.S. cotton is in drought-affected areas.
The Hormuz closure compounded the damage directly. The Gulf accounts for 30 to 35% of globally traded urea exports. With fertilizer supply disrupted, energy costs elevated, and a spring planting season failing in real time, food price inflation has a structural agricultural underpinning that extends well into 2027.
What is not being discussed in the same breath: the $725 billion AI infrastructure buildout is landing in the same geography. A single hyperscale data center consumes between one and five million gallons of water per day for cooling. Microsoft, Google, Amazon, and Meta are standing up facilities across the Southeast, Texas, and the broader Sun Belt, the same regions recording historic drought conditions. Farmers in the sixth year of drought are watching aquifer levels collapse while the construction crews breaking ground next door are planning to draw from the same depleted water table for the next 20 years. The infrastructure boom and the agricultural crisis are not parallel stories. They are competing for the same resource in the same counties at the same time.
What the Numbers Are Actually Pricing
Hyperscalers will likely hit their 2026 capex targets. Microsoft, Alphabet, Amazon, and Meta have collectively guided to $650 to $725 billion in capital expenditure this year, up 77% from 2025's record. Cloud revenue is accelerating. Google Cloud grew 63% in Q1. The backlogs are real and contracted.
Capex-to-revenue ratios are at levels analysts describe as seemingly untenable: 86% for Oracle, 54% for Meta, 47% for Microsoft, 46% for Alphabet. Amazon is projected to go free cash flow negative in 2026. Microsoft's free cash flow is estimated to slide 28% this year. Analyst consensus puts 2027 hyperscaler capex at $1 trillion or above. That consensus was formed before a consumer base running at sub-2008 sentiment levels, before a food and energy shock with a multi-year tail, before an entry-level labor market at a 37-year low, and before a Federal Reserve with no room to maneuver were fully visible in corporate revenue. The question is not whether AI infrastructure investment is real. The question is whether the customers paying for that infrastructure can sustain the growth rates that justify the multiple.
October and November
Q3 earnings season runs October and November. It is the first reporting window where two full quarters of simultaneous oil shock, agricultural failure, consumer debt deterioration, and labor market weakness will be visible in actual corporate revenue. Formal 2027 guidance gets attached to those prints for the first time, converting analyst projections into on-record numbers.
The Nasdaq collapse of 2000 did not begin when the technology failed. It began when the multiple being paid for 2005 earnings got repriced to what 2001 earnings actually looked like. The 2022 correction did not begin when cloud revenue decelerated. It began when the market stopped paying 2028 prices for 2025 earnings as rates shifted. Genuine underlying growth, legitimate transformation narrative, real revenue, and then an overnight consensus shift where the forward multiple becomes indefensible.
Every one of those conditions is present. The only variable left is when the market decides to look down.
The people who built this economy are being asked to keep spending their way through a crisis they did not create, with wages that have not kept pace, in a labor market that no longer has a door open for the next generation. That is not a soft landing. That is a controlled demolition.