Welcome to The Kool-Aid Diaries, Volume One, Issue Five, May twenty-twenty-six — Artificial: Intelligence and Jobs Not Included. The Bureau of Labor Statistics released the April twenty-twenty-six nonfarm payrolls report last week. The headline: plus one hundred fifteen thousand new jobs, decisively beating a bleak Wall Street consensus of negative sixty-five thousand. Markets rallied. Commentators exhaled. The administration declared vindication. We read past the first paragraph. What we found is not a labor market that beat expectations — it is a statistical architecture that has come apart at the seams. The establishment survey says plus one fifteen K. The Birth-Death model inside that same survey imputed plus three-oh-six thousand in jobs that were never directly counted. And the B-L-S's own household survey — a separate measure of the same labor market, from the same report, released the same morning — says the U.S. economy lost two hundred forty-one thousand jobs in April. One report. Two surveys. A three hundred fifty-six thousand job gap. This issue, we follow the receipts.

The monthly jobs report contains two distinct surveys of the labor market conducted simultaneously and independently. The Establishment Survey — also called the payroll survey — contacts roughly one hundred nineteen thousand businesses and asks how many workers are on payroll. This is the number that generates the headline figure. April: plus one hundred fifteen thousand. The Household Survey — also called the Current Population Survey — contacts approximately sixty thousand households and asks individuals directly whether they are employed. This is also the source of the headline unemployment rate. But in April, it said the economy lost two hundred forty-one thousand employed persons. The spread between the two measures is three hundred fifty-six thousand jobs. That is not noise. Divergences of that magnitude occur at turning points — when one survey is capturing a dynamic the other is lagging. Historically, when the household survey leads to the downside, the establishment survey has eventually revised toward it — not the other way around.

Buried inside the same household survey is a breakdown that the headline employment number conceals entirely. In April twenty-twenty-six, full-time employment fell by four hundred fifty thousand. At the same time, part-time employment rose by one hundred twenty-two thousand. The net swing between the two categories was five hundred seventy-two thousand positions — a shift from full-time to part-time work of historic magnitude for a single month.

This is the specific mechanism that corrupts the headline unemployment rate — and it connects directly to the measurement problem we'll cover in the methodology section. The B-L-S counts a person as "employed" if they worked as little as one hour during the survey reference week. A worker who lost a ninety-five-thousand-dollar full-time position and replaced it with three part-time jobs averaging eighteen thousand dollars combined annually is counted, in the official statistics, as employed. The loss of income, the loss of benefits, the loss of economic security — none of that appears in the headline.

This is the specific mechanism that makes a falling participation rate and a stable headline unemployment rate simultaneously possible. When full-time workers are displaced to part-time work, they do not become "unemployed" — they become statistically invisible to the measure most widely cited. When they eventually give up the search entirely, they exit the labor force and disappear from the denominator. At each step, the official number looks better than the underlying reality warrants.

The Birth-Death Model is a monthly statistical imputation that adds estimated jobs to the establishment survey based on the B-L-S's projection of net business formation. In April twenty-twenty-six, that contribution was plus three hundred six thousand jobs — the largest single-month imputation in recent history. Without the Birth-Death adjustment, the establishment survey would have printed negative one hundred ninety-one thousand. The entire positive headline, and then some, was generated by a model — not by directly counting employed workers.

This model serves a legitimate statistical purpose during periods of economic stability, where historical business formation patterns are a reasonable basis for extrapolation. The problem is that this model systematically overestimates job creation in slowing or contracting economies — precisely because it cannot distinguish between a business that quietly opened and one that quietly closed.

The track record is documented and unambiguous. In November twenty-twenty-four, the B-L-S released its annual benchmark revision and erased approximately eight hundred eighteen thousand jobs that the Birth-Death Model had previously contributed to the count. Those jobs circulated in financial headlines, moved markets, and informed Federal Reserve policy decisions for months — before being quietly deleted. There was no press conference.

Even setting aside both the Birth-Death question and the household survey, the labor force participation rate tells a story the headline unemployment figure cannot. In April twenty-twenty-six, participation fell to sixty-one-point-nine percent — down from sixty-two-point-seven percent year over year. The U-3 unemployment rate — four-point-one percent, the number in every headline — does not count people who have stopped looking for work. When participation falls, the denominator of the unemployment calculation shrinks, making the numerator look better than the underlying labor market warrants. The current participation rate implies approximately four-point-two million fewer Americans are in the labor force than pre-pandemic trend would suggest. Those are not all retired workers and students. That is the shadow unemployment pool: people who were working, stopped, and are no longer being counted.

The U-6 broad unemployment measure — capturing part-time workers who want full-time work and marginally attached workers — currently sits at seven-point-eight percent, nearly double U-3. The spread between U-3 and U-6 is widening, not narrowing, indicating the composition of employment is degrading even as the headline holds. Critically, the household survey's negative two-forty-one print for April is directionally consistent with the falling participation rate. The two household-based data points are confirming each other. The establishment survey is the outlier.

This newsletter does not traffic in conspiracy theories. But intellectual honesty requires acknowledging a documented fact: the methodology for calculating unemployment has been revised multiple times since the nineteen-eighties, and those revisions have — in aggregate — produced more favorable-looking headline numbers on the same underlying economic reality. The most structurally significant change came in nineteen-ninety-four, when the B-L-S redefined the criteria for labor force participation and narrowed the definition of "discouraged worker." People who had stopped looking for work for more than a year were reclassified out of the labor force entirely — removing them from the denominator and lowering the measured unemployment rate by an estimated one-and-a-half to two percentage points relative to the prior methodology applied to the same population.

Independent researchers applying the pre-nineteen-ninety-four methodology consistently forward estimate that true unemployment — measured the way the government measured it during the nineteen-eighty recession — currently sits at approximately twenty-two to twenty-five percent. Even discounting those estimates significantly, the directional point is sound: the ruler has been shortened, and historical comparisons must account for that.

The practical significance of a misleadingly strong payrolls number is not academic. It directly shapes Federal Reserve policy. A headline of plus one-fifteen K gives the Fed a convenient justification to hold rates in the face of still-elevated inflation at three-point-two percent. The stagflation policy trap described in Issue Four closes a little tighter.

Meanwhile, the underlying consumer stress has not abated. Credit card delinquencies at ninety-plus days have ticked up to two-point-six-three percent, approaching the three percent threshold identified as a structural warning signal. The tax refund buffer is largely exhausted. Major retailers are reporting sequential deterioration in discretionary spending. The labor market the official data describes and the one millions of households are navigating are increasingly two separate realities.

The base case — stagflation or prolonged malaise — is unchanged. The April payrolls report does not alter the base case; it deepens the measurement problem surrounding it. The Fed holds rates. Inflation stays sticky at two-point-eight to three-point-four percent. The household survey continues signaling labor market deterioration while the establishment survey posts misleadingly benign headlines. Credit stress broadens as the refund buffer expires. The Birth-Death model continues generating phantom jobs until the next benchmark revision erases them.

The elevated downside risk is what we are calling a benchmark revision shock. The November twenty-twenty-four revision erased eight-eighteen thousand jobs. If Birth-Death contributions continue at elevated rates through mid-twenty-twenty-six, the next annual revision — typically released February twenty-twenty-seven — could be materially larger. A major revision arriving into an already-stressed consumer environment and a banking system still carrying roughly four hundred eighty billion dollars in unrealized losses could function as a catalytic confidence shock to both markets and Fed credibility simultaneously.

The soft counterpoint — the resilience case — cannot be dismissed. It is possible the elevated Birth-Death contribution reflects genuine small-business formation driven by A-I-enabled micro-entrepreneurship — a structural dynamic the model's historical calibration doesn't recognize. If so, the labor market is more resilient than the household survey suggests, and the soft landing materializes. But this scenario requires more simultaneous favorable developments than any other in the range.

First: the household survey versus establishment survey divergence. A second consecutive month of household deterioration while establishment prints positive is the pattern that preceded the twenty-twenty-four revision. Three consecutive months above a hundred-fifty K divergence elevates the benchmark revision tail risk significantly.

Second: credit card delinquencies versus the three percent threshold. April's two-point-six-three is closing in. With the refund buffer exhausted, any acceleration signals that consumer credit stress is broadening beyond low-income cohorts into middle-income households.

Third: the Birth-Death monthly contribution versus two hundred thousand. A second month above two hundred K in an environment of rising delinquencies and falling participation would be historically anomalous. Treat any reading above that threshold in the current backdrop as a signal to discount the establishment headline accordingly.

Fourth: the ten-year Treasury yield versus the four-and-a-half percent threshold. Currently at four-point-four-two — the closest approach since the stress peaks of twenty-twenty-five. Stanford research identifies four-point-five as the threshold above which bank unrealized losses become acutely destabilizing. Each approach is a live stress test on a system that has not been tested.

Fifth: corporate earnings versus C-A-P-E expectations. At a CAPE of approximately forty-two, the market is pricing a very optimistic earnings trajectory into a labor market that the household survey says is contracting. Any meaningful miss from mega-cap tech names carrying index weight has outsized downside at current valuations.

THE AI ACCOUNTABILITY RECKONING

For three years, the market operated on a single assumption: AI spending was sacrosanct. Every dollar burned on G-P-Us and data centers was treated as a vote of confidence in a generational technology shift. Asking whether the returns justified the spend was treated as a failure of imagination. That era is ending. Wall Street is now asking a question it avoided for years: where is the money?

The evidence that AI investment is failing to convert into measurable returns is no longer anecdotal. A twenty-twenty-five M-I-T study examined between thirty-five and forty billion dollars invested in corporate A-I initiatives and found that ninety-five percent of companies reported no measurable return on investment — zero impact on profits. Only five percent reported any demonstrable value, and those were companies that identified a single focused operational pain point and executed narrowly against it.

Deloitte's twenty-twenty-five survey of nearly nineteen hundred executives found that while eighty-five percent of organizations increased AI investment, the typical payback period was two to four years — compared to the seven to twelve months expected for standard technology investments. Only six percent saw payback in under a year. Apollo's chief economist Torsten Slok put it plainly: "A-I is everywhere except in the incoming macroeconomic data. Today, you don't see A-I in the employment data, productivity data, or inflation data." He added that outside the Magnificent Seven, there are "no signs of A-I in profit margins or earnings expectations."

No single company has become more emblematic of the AI accountability reckoning than Oracle. Oracle's stock is down more than twenty-five percent year to date in twenty-twenty-six — the worst performer among large-cap technology names — having shed roughly forty percent from its September twenty-twenty-five peak. The company carries one hundred twenty-four-point-seven billion dollars in long-term debt, up forty percent year over year, with net debt exceeding ninety-five billion dollars. Cash outflows climbed from two-point-seven billion to ten billion dollars in a single year. Oracle is now one of the most heavily shorted large-cap stocks in North America.

Its five-year credit default swap spread — the market's price for insuring against Oracle defaulting on its debt — has surged to one hundred ninety-eight basis points, the highest on record. Bond markets are not being subtle here. The risk concentration at the center of Oracle's story is particularly acute: five hundred fifty-three billion dollars in remaining performance obligations, up three-twenty-five percent year over year — but three hundred billion of that is a single cloud deal with OpenAI. In late April twenty-twenty-six, the Wall Street Journal reported that OpenAI has recently missed its own internal revenue growth projections, and that its finance chief has warned colleagues the company could face difficulty funding future compute agreements. Oracle dropped four percent on that single news report. When your entire A-I growth thesis is a single counterparty that is itself missing revenue targets, the risk topology becomes a very specific shape.

Now here is where the story demands a sharper kind of attention. The three-trillion-dollar A-I data center buildout — estimated as high as five to seven trillion when all infrastructure is counted — is too large for even the largest technology companies to fund from operating cash flow. According to J.P. Morgan, hyperscalers are already diverting roughly five hundred billion of their seven hundred billion in annual net operating income toward capital expenditures. The remaining gap is being filled by complex structured debt products with limited transparency, tranched into risk layers, and distributed to institutional investors. J.P. Morgan projects annual data center securitization issuance — primarily through commercial mortgage-backed securities and asset-backed securities — could reach thirty to forty billion dollars annually in both twenty-twenty-six and twenty-twenty-seven. U-B-S forecasts nine hundred billion in new technology sector debt globally in twenty-twenty-six alone. Morgan Stanley and J.P. Morgan project the sector may need to issue one-point-five trillion in new debt over the next few years.

The products being created to distribute this risk have a familiar structure. Loans are pooled together — data center leases, G-P-U-backed debt, infrastructure project finance — and sliced into tranches with graduated risk-return profiles. Senior tranches with priority claims and lower yields. Mezzanine tranches with moderate risk. Equity tranches at the bottom absorbing first losses. This is, structurally and functionally, the collateralized debt obligation architecture that populated bank balance sheets ahead of two-thousand-and-eight — with A-I infrastructure replacing subprime mortgages as the underlying asset.

Rajat Rana, a partner who worked on structured finance litigation after twenty-oh-eight, described the current dynamic to C-N-B-C in April twenty-twenty-six as follows: "We're talking about trillions of dollars, and almost going back to the same cycle where there's almost no transparency about the financing structures." He called the A-I data center buildout the "largest peacetime investment project in human history, which is financed largely off balance sheet." The parallel is not that A-I infrastructure is worthless. It may not be. The parallel is the structural mechanism: debt packaged, tranched, rated, and distributed in ways that obscure the underlying risk and the dependency chain between an asset's value and the revenue assumptions underwriting it. In two-thousand-and-six, those revenue assumptions were home prices. In twenty-twenty-six, they are A-I demand projections from companies that, on current evidence, are not generating the returns their capital expenditures require.

Now consider the full picture assembled across this issue. The labor market is weaker than the headline suggests. Consumer credit stress is rising. Bank unrealized losses remain near crisis-level territory. The CAPE ratio is at forty-two against a historical median of sixteen. Into this environment, Wall Street is constructing a one-point-five trillion dollar structured debt edifice underwritten by A-I demand projections — at the precise moment that enterprise-level evidence for those returns is, at best, mixed and, at worst, decisively negative. Ninety-five percent of companies are seeing no return. Forty-two percent scrapped their A-I projects in twenty-twenty-five. A-I is, in Apollo's words, "nowhere" in the macroeconomic data. And yet the capital expenditure continues, funded by debt being packaged and distributed with, in the words of someone who lived through twenty-oh-eight, "almost no transparency." The whispers on Wall Street are getting louder. Follow the receipts.