News & Commentary

This is the most granular confirmation yet of the bifurcation we’ve been tracking across multiple data sources — Dallas Fed wage research, Fortune’s 37-year entry-level high, the Anthropic Economic Index learning curves, and now millions of actual job postings. The pattern is identical everywhere you look: AI augments experienced workers and displaces inexperienced ones. What makes the FT analysis especially valuable is that it captures a Jevons paradox in action — AI makes software cheaper to produce, so demand for software goes up, which increases demand for the senior humans still needed in the loop. Total job postings rise. But entry-level postings don’t, because the work juniors used to do is now handled by the AI itself. The headline number looks healthy; the age distribution underneath it is hollowing out. Ellich’s observation about delegation is the mechanism: the job is shifting from writing code to directing AI that writes code, and that requires judgment and context that come from experience, not textbooks. The pipeline problem she flags is real and underappreciated. If companies stop hiring juniors because AI handles junior work, where do the next generation of seniors come from? The industry’s implicit answer right now is: someone else’s problem. For investors, this is the post-labor economy in miniature. The sector most exposed to AI is seeing total demand rise and entry-level demand collapse simultaneously. That’s not what a simple automation story predicts. It’s what a restructuring story predicts — and it’s exactly the dynamic our thesis is built on.

The public narrative from CEOs is that AI creates more jobs than it destroys. The private data from CFOs tells a different story: nearly half of U.S. firms are planning AI-driven cuts this year. The 502,000 figure sounds manageable against 125 million jobs — Graham calls it “not doomsday” — but context matters. Last year, 55,000 AI-attributed layoffs were enough to dominate headlines and move markets. Nine times that is not a rounding error. It’s a structural acceleration. And these are the cuts CFOs are willing to admit to in a survey. The real number may be higher, because many companies will frame AI cuts as “restructuring” or “efficiency gains” without flagging AI as the cause. The ADP turnover data is equally telling: people who have jobs aren’t leaving them, because they sense what’s coming. When voluntary quits hit record lows and AI-planned cuts hit record highs simultaneously, you’re looking at a labor market that’s freezing from the bottom while eroding from the top. That’s not a cycle. That’s a transition.

This is the displacement thesis with receipts. Anthropic isn’t speculating about what AI might do to the labor market — it’s measuring what its own product is already doing. Three findings matter most for investors. First, the migration of tasks from Claude.ai (human-in-the-loop) to the API (automated, directive, no human required) is the canary. When customer service, sales outreach, and trading operations move to automated pipelines, those aren’t productivity gains — they’re headcount reductions waiting to be announced. Second, the learning curve finding is the entry-level problem in data form. Experienced workers get better at using AI; they tackle harder tasks and succeed more often. Entry-level workers, who lack the tacit knowledge to direct AI effectively, become the layer that gets squeezed out. The Dallas Fed research and the Fortune 37-year-high data we’ve been tracking land in exactly the same place. Third, the adoption curve is moving down the wage ladder. Early adopters were coders and analysts. Now it’s spreading to product comparisons, home maintenance, and sports questions — everyday tasks. When AI becomes ambient, the displacement isn’t concentrated in tech anymore. It’s everywhere.

This is the consumer demand problem in embryo. The post-labor thesis isn’t just about workers losing jobs — it’s about an entire generation never getting them in the first place. Entry-level roles are where people become earners, spenders, borrowers, and homebuyers. When those roles disappear, the pipeline that creates future consumer demand narrows. The Dallas Fed research we track on our Signals page confirms the mechanism: AI automates codifiable knowledge (the kind entry-level workers have) while augmenting tacit knowledge (the kind experienced workers have). The result is a labor market that rewards the top and abandons the bottom. Finance and information services flipping from +44,000 jobs per month to −9,000 is not a cycle — it’s a structural reversal. And the 37-year-high in new entrant unemployment didn’t happen during a recession. It happened while GDP was growing and corporate earnings were hitting records. That’s the post-labor signature: the economy doesn’t need to contract to leave people behind. It just needs to stop needing them.

This is what the post-labor trap looks like from inside the Fed. Powell can see the labor market breaking — he said so plainly — but he can’t cut because inflation is running hot and oil is above $100. In a normal recession, weak jobs give the Fed room to ease. But this isn’t normal. Companies aren’t laying off workers because demand collapsed — they’re laying off workers because AI made them unnecessary, and redirecting the savings into more AI. That kind of displacement doesn’t reverse when rates come down. The Fed’s framework assumes unemployment is cyclical: cut rates, stimulate demand, jobs come back. What if the jobs are structurally gone? Powell doesn’t have a tool for that. Nobody does. Meanwhile, Miran’s fifth consecutive dissent tells you the internal tension is real — part of the committee sees a labor market that needs help now, while the majority sees inflation they can’t ignore. The post-labor economy puts those two mandates in direct conflict, and yesterday’s meeting was the clearest illustration yet.

This is the post-labor feedback loop in one corporate decision. Meta is not cutting workers because revenue is falling — 2025 revenue topped $200 billion. It’s cutting workers to fund the technology that replaces workers. The layoffs are a financing mechanism for AI: fire humans, redirect payroll savings into GPU clusters and AI talent, then use the AI to justify keeping the headcount permanently lower. Zuckerberg’s quote about “big teams” being replaced by “a single very talented person” is the thesis in nine words. And notice the market’s response: the stock went up. Wall Street is explicitly rewarding the substitution of capital for labor. Jefferies called it “a broader shift.” That’s an understatement. When the world’s seventh-largest company by market cap announces it can function with 20% fewer people and investors cheer, the market is telling you which side of the capital-vs-labor divide it’s on. For GAPL investors, Meta sits in the Enablers category — a company building the infrastructure that makes displacement possible across every industry it touches.

The GDP-jobs decoupling we’ve been tracking just got more complicated. The story was: GDP strong, jobs weak — the post-labor signature. Now GDP is weak too. Q4 growth of 0.7% — before the Iran war hit — means the economy entered this crisis with less momentum than anyone thought. Oil above $100 is more likely to deepen the slowdown than to spark lasting inflation; it’s a tax on consumption, not a wage-price spiral. A skeptic would ask: isn’t this just a conventional recession? Maybe. But conventional recessions don’t feature companies reporting 22% revenue growth while cutting workers and citing AI as the reason. The cyclical weakness and the structural displacement are happening simultaneously — and that combination is what makes this different.

Atlassian isn't Oracle. Oracle is building AI infrastructure and cutting workers to fund it — an Enabler. Atlassian is facing something more existential: AI may be eliminating the need for the humans who use Atlassian's products. If AI agents write and manage code, you need fewer developers. If you need fewer developers, you need fewer Jira licenses. Analysts are calling it the "SaaSpocalypse" — the possibility that per-seat software subscriptions lose value when AI shrinks the number of seats. The stock's 84% decline from peak suggests the market is pricing this in. This is what disruption looks like from the other side.

This is the post-labor economy in one earnings report. Revenue growing 22%. Workforce shrinking. AI explicitly cited as the reason. And the entire buildout funded by massive debt issuance. Oracle is simultaneously an Enabler (building AI infrastructure) and a case study in displacement (cutting its own workers to fund the machines).

This is the number the post-labor framework has been pointing toward. For months, we've tracked the divergence between GDP growth and job creation — an economy producing more output with fewer workers. That divergence just turned into outright job losses. The consensus expected +50,000 and got -92,000. That's not a miss. That's a different economy than the one Wall Street thought it was looking at. The timing is hard to ignore. This report lands in the same week that: the New York Times ran a 2,000-word essay asking whether AI is about to gut the white-collar middle class; Goldman Sachs documented a 30% productivity gain in the two sectors most exposed to AI; Oracle announced thousands of layoffs to fund its AI data center buildout; and India's $300 billion outsourcing industry showed signs of structural collapse. One month doesn't make a trend, and the Kaiser strike is a real factor. But strip that out and you're still negative. The question Claudia Sahm raised in the CNBC preview — whether the ultra-low hiring rate makes the economy "vulnerable" — just got answered. The Sahm Rule recession indicator, which uses changes in the unemployment rate to flag downturns, is now within range. We're watching.

This is the most important chart in the AI-and-jobs debate, and the company that made Claude is the one publishing it. Anthropic's index shows a huge gap between what AI can theoretically automate and what it's actually automating today. That gap is the investment signal. Right now, only 33% of automatable tasks in computer and math occupations are actually being done by AI. When that number moves to 50%, to 70%, the displacement that's currently "ambiguous" becomes unmistakable. The researchers explicitly compare this to the China shock — where millions of manufacturing jobs were lost but it took economists years to agree on what was happening. That's the playbook: by the time the data is definitive, the displacement has already occurred. What makes this unusual is who's publishing it. Anthropic's CEO, Dario Amodei, is one of the most vocal voices warning that his own technology could displace half of entry-level white-collar jobs within one to five years. Building the measurement tool before the disruption arrives is either an act of unusual corporate responsibility or an acknowledgment that the disruption is close enough to measure. Either way, the framework tells you exactly what to watch: the red area growing to fill the blue.

Oracle illustrates a mechanism we haven't seen clearly until now: companies cutting workers to fund AI, not just because of AI. The capital demands of building AI infrastructure are so enormous — $50 billion in a single year, cash flow negative until 2030 — that the workers themselves become a funding source. You are being laid off so the company can afford the machines that will also replace you. That's the post-labor flywheel in its purest form: invest in AI infrastructure, cut workers to pay for it, use AI to replace the work those workers did, then reinvest the savings into more AI. Oracle isn't unique here. The hyperscalers collectively plan $667 billion in capex for 2026 (per Goldman). That capital has to come from somewhere. When the spending is this large and the returns are years away, payroll is the most liquid line item on the balance sheet. This story also connects to the India outsourcing disruption: Oracle's clients include the same global enterprises that once offshored work to India. Now they're building the AI infrastructure to eliminate the need for that labor entirely — whether it's in Bangalore, Austin, or Redwood City.

It's like the New York Times has been reading postlaborinvesting.com. This piece pulls together everything we've been tracking into a single narrative — and the most important thing about it isn't the data, it's the venue. When the NYT runs a 2,000-word essay asking whether AI is about to gut the educated middle class, and sources it to Harvard economists and the FT's Martin Wolf, this is no longer a fringe conversation. The Steinberger essay does something subtle that most coverage misses: it connects the GDP-jobs divergence (strong growth, weak hiring) to sector-specific evidence. Finance, insurance, consulting, and tech aren't just slowing — they're posting solid results with flat or shrinking headcounts. That's the productivity-without-people pattern the post-labor framework predicts. Wolf's quote is the one to remember. Deindustrialization hollowed out the working class and reshaped politics for a generation. If AI does the same to the college-educated professional class — the people who, as Wolf puts it, "run our societies in almost every possible way" — the political consequences will be larger, faster, and harder to contain. The essay also captures the policy vacuum: a Warner-Hawley bill to track AI job losses but nothing to actually help displaced workers. We're measuring the flood but not building the levee.

This is the most useful data point in the AI-and-jobs debate right now — because it quantifies the gap between talk and action, and then shows you what happens when the gap closes. Seventy percent of CEOs talk about AI. Ten percent can point to a use case. One percent can show it in earnings. That's the adoption funnel, and it tells you we're early. But look at what happens at the narrow end: 30% productivity gains in software and customer service. Those aren't projections — they're reported results from companies that actually implemented. The investment implication is the lag between the 70% talking and the 1% showing up in earnings. That gap is closing, and every percentage point it closes sends two signals simultaneously: higher margins for companies that execute, and fewer jobs for the roles being automated. Goldman's own number — 11 million jobs eventually displaced — sits alongside $667 billion in capex. That's the trade: massive capital spending to build the infrastructure that eliminates the labor. Companies discussing AI in their workforce context are already cutting job openings 50% faster than the broader market. The displacement isn't theoretical. It's in the hiring data.

This is the post-labor transition at national scale. India's entire outsourcing model was the original labor cost arbitrage: Western companies shipped work to lower-cost workers overseas. AI eliminates the need for the arbitrage entirely — not by making Indian workers cheaper, but by making the work nearly free. When the cost of an AI agent approaches the cost of electricity, it doesn't matter whether you're paying $150,000 in San Francisco or $15,000 in Bangalore. Both lose to $0.02. The fact that this is hitting India's IT sector — the most successful knowledge-work export economy in history — should end the debate about whether AI displacement is limited to a few tech companies in Silicon Valley. It's not. It's global, it's structural, and it's moving up the value chain. The engineering college placement numbers are the leading indicator: the pipeline of new workers entering the knowledge economy is narrowing everywhere, simultaneously.

Fortune ran the same story on Friday. The Wall Street Journal ran it on Saturday. When both papers of record for the investment class publish the same thesis in the same weekend — and the headline uses the word "real" — the mainstream narrative has shifted. What matters for investors isn't the human-interest angle (though it's devastating). It's the market structure the Journal describes: companies that cut get rewarded, companies that don't get punished. That's not a one-quarter trend. That's a repricing mechanism that feeds on itself. The most telling detail is the Block employees who built with AI and got laid off anyway. The narrative that "learning AI will save your job" is breaking down in real time. The tools don't protect the worker — they replace the need for the worker. Capital doesn't share productivity gains. It captures them.

When Fortune runs a 3,000-word lead story connecting the same dots we've been tracking — Citrini's Ghost GDP, the Block layoffs, the divergence between economic data and lived reality — the thesis has left the fringe. What's notable isn't that Fortune agrees with the doomsday scenario. It's that the pushback is getting weaker. Citadel's rebuttal cites rising software engineer demand — but that's demand to build the tools that replace everyone else. Morgan Stanley's answer is "new job titles will emerge" — Chief AI Officer, computational geneticist — which is a hope, not a labor market. And BofA compares the selloff to a bank run triggered by rumors, missing the point that Block's 40% cut isn't a rumor. The most honest quote comes from an insurance CEO: "It's slow, and then it's sudden." We're somewhere in the transition between those two speeds.

This is one of the cleanest pieces of Fed research on the two-speed labor market AI is creating. Fewer jobs, higher pay for survivors — that's not a contradiction, it's the pattern. What makes the Dallas Fed framework especially useful is the codified-vs-tacit knowledge distinction. AI doesn't just replace "low-skill" workers — it replaces anyone whose value comes from knowing things that can be looked up. A junior lawyer doing document review is more exposed than an electrician. A new financial analyst running models is more exposed than a veteran dealmaker reading a room. The paper's most important implication is buried at the end: the entire white-collar career ladder — start entry-level, learn on the job, become senior — breaks down when companies can't justify hiring the entry-level rung. That's not a labor market problem. That's a civilizational design problem.

A profitable company fires 40% of its workforce, explicitly credits AI, and the market rewards it with a 24% stock pop in a single session. The signal isn't that Block is struggling — it's the opposite. Dorsey cut from strength, and Wall Street cheered because fewer humans means wider margins. Every CEO in America watched that stock chart yesterday. Dorsey's prediction — that most companies will reach the same conclusion within a year — isn't a warning, it's a roadmap. The pattern we've been tracking (companies rationally cutting headcount while collectively destroying the consumer base) just got its most explicit corporate endorsement yet. When a CEO publicly says AI replaces half his workforce and shareholders add $3 billion in market value in response, the incentive structure is locked in. The question is no longer whether companies will cut — it's who moves first and gets rewarded, and who moves last and gets punished.

These comments from a Fed governor pretty much sum up the post-labor economy. Cook's specific concern — that rising unemployment in a productivity boom won't signal "slack" in the usual way — is exactly the dynamic post-labor investors need to understand. The Fed's own tools may be poorly suited to the transition ahead. That's not a bearish signal for the market. It's a bullish one for companies that are on the right side of it.

This is what AI displacement looks like in real time — not a labor statistic, but a $31 billion repricing of a business model in a single afternoon. IBM's COBOL consulting moat wasn't built on technology; it was built on scarcity of human expertise. The people who understand COBOL are retiring and universities stopped teaching it. IBM monetized that scarcity for decades. Then a blog post from an AI startup flipped the equation: if the machine can read what humans can't, the scarcity premium evaporates. This is the pattern investors should internalize. AI doesn't need to replace the worker — it just needs to replace the skill that made the worker expensive. Every industry has its COBOL: specialized knowledge that commands premium billing rates because few people possess it. Legal research. Actuarial analysis. Radiology reads. When AI can perform the cognitive task that justified the premium, the entire fee structure reprices. IBM is the case study. It won't be the last.

Citrini's thesis — that AI bullishness is actually bearish because productivity gains destroy the consumer demand that sustains the market — is the argument we laid out in our Capital Markets analysis last week. The "negative feedback loop" where companies rationally cut headcount to boost margins but collectively destroy their own customer base is exactly what we called Phase 2: the Recognition Phase. The market is now pricing this scenario in real time. Note what's happening beneath the headline: the selloff isn't hitting AI infrastructure (Nvidia was flat). It's hitting the companies whose business models depend on human workers as customers, licensees, and intermediaries — software seats, consulting hours, brokerage fees. The market is sorting capital-side winners from labor-side losers, which is the core of our investment framework. Citrini explicitly disclaims it as "a scenario, not a prediction." Fair enough. But when a thought experiment moves the Dow 800 points, it's no longer theoretical — it's a risk the market is actively pricing.

Both sides of this debate are missing the labor variable. Warsh sees AI productivity as non-inflationary growth that justifies rate cuts. Current Fed officials see AI infrastructure demand as inflationary, justifying holding rates. Neither is fully accounting for what happens when AI productivity gains flow to capital while labor income erodes — which is disinflationary on both the demand side (fewer wages) and the supply side (lower production costs). The 2.7% productivity jump with zero job creation is the early data point: output is decoupling from employment. If that continues, the Fed will eventually be cutting rates not because productivity enabled it, but because labor market deterioration demands it.

When the largest retailer in America — built on serving lower-income households — reports that its growth engine is now affluent consumers trading down for value while its core customers are paycheck-to-paycheck, you're looking at the K-shaped economy in a single earnings call. This is the capital-vs-labor divergence expressed through consumer spending data: asset owners are flush, wage earners are squeezed, and the gap is widening.

The gap between narrative and data is itself the story: companies want to attribute cuts to AI because it signals forward-thinking efficiency to investors, while the actual displacement hasn't yet hit at scale. But Altman's own caveat — "I would expect the real impact of AI doing jobs in the next few years will begin to be palpable" — is the more important line. We're in the window where AI washing inflates the perceived pace of displacement while masking the real structural shift still building underneath.

Fortune's framing is exactly right — this isn't the fringe scenario in Barr's speech, it's the one that matters for investors. "Concentrated among a small group of capital holders and AI superstars" is the post-labor investment thesis stated by a Federal Reserve Governor. His admission that government has a poor track record managing these transitions makes the case for private capital positioning even stronger. If you're waiting for policymakers to protect labor income, Barr himself is telling you that history says they won't do it in time. The investment implication: own the capital side of this equation.

This is a sitting Federal Reserve Governor — from the podium at the New York Association for Business Economics — stating that job creation has been near zero while the economy continues to grow. That's the post-labor thesis delivered in Fedspeak. His warning that the labor market is "especially vulnerable to negative shocks" is the quiet part out loud: the traditional employment safety net is fraying. When the Fed starts dedicating entire speeches to AI labor displacement, the thesis has moved from the fringe to the center of monetary policy discussion.

The January jobs data is striking: strip out healthcare — an industry dedicated to managing an aging population — and the economy created almost no jobs. The 55,000 AI-attributed layoffs are still small relative to the 1.5 million monthly churn, but the trajectory matters more than the level. The historical parallel to deindustrialization is apt: that displacement created a generation of economic dislocation. The difference this time is speed and scope — white-collar workers are the new factory workers, and AI capabilities are compounding monthly, not over decades.

When Mohamed El-Erian — one of the most respected macroeconomists alive — describes exactly the phenomenon at the center of our thesis, it's worth paying attention. GDP growing while employment stagnates is the post-labor economy in one sentence. His framing is precise: AI capex drives output but doesn't have the same employment multiplier as traditional investment. This isn't a fringe idea anymore. It's becoming the mainstream macro view. The investment implication is clear: own the capital, not the labor.

When the CEO of the company selling the tools says the jobs are going away in 18 months, listen — not to the timeline, but to the incentive structure. Suleyman is simultaneously building the replacement technology and warning you about it. That's not altruism; it's a sales pitch dressed as a prophecy. The timeline is almost certainly aggressive — adoption friction, regulatory drag, and institutional inertia will slow deployment. But the direction is undeniable. What matters for investors: this isn't a fringe prediction anymore. The people building these systems are now openly stating that white-collar labor is the target, not the customer. Every company buying Microsoft Copilot licenses is implicitly agreeing. The capital-vs-labor rebalancing we track isn't theoretical — it's being priced into enterprise software contracts right now.

This is the "jobless expansion" entering the lexicon of mainstream economics. GDP up, jobs flat. The white-collar employment peak coinciding exactly with ChatGPT's launch is a data point that will be studied for decades. Oxford Economics warns this leaves the economy "vulnerable to shocks" because the labor market is the main firewall against recession. Translation: the economy is structurally shifting to favor capital over labor, and most people's financial security still depends on the old model.

Shumer's essay struck a nerve because it articulated what millions of knowledge workers are quietly feeling. The skeptics — Gary Marcus, Oxford Economics, Fortune's own analysis — make fair points about timeline and hype. But the directional argument is hard to dismiss: AI is improving at everything simultaneously, and unlike previous automation waves, there's no obvious adjacent sector for displaced workers to move into. Whether the timeline is 2 years or 10, the investment implications are the same. Read it, then read the rebuttals. Form your own view. That's what this page is for.

This is the single most important labor market data point of the year. The economy we thought we had in 2025 didn't exist. Stock markets hit all-time highs while the job market was essentially at a standstill. That disconnect — booming corporate profits, stagnant employment — is the post-labor economy in miniature. When companies can grow earnings without growing headcount, the thesis isn't theoretical anymore.

Companies are betting their balance sheets on AI — and they're not building labor-intensive factories. They're building data centers that run on chips, not people. When the four largest companies in the world spend $700B on automation infrastructure while the economy creates 181,000 jobs, the capital-over-labor reallocation isn't subtle. Follow the capex.

Some skeptics argue companies are using AI as a pretext for normal layoffs. That may be partly true — but it doesn't matter for the thesis. Whether companies are actually replacing workers with AI today or merely signaling that they intend to, both lead to the same investment conclusion: capital is flowing toward automation and away from labor. The direction of the bet is clear even if the timing is debated.

This is the post-labor thesis expressed in government data. Output up 5.4%, hours up 0.5%. Companies are producing dramatically more with barely any additional labor input. Falling unit labor costs mean capital's share of the value is expanding. The comparison to the late 1990s productivity boom is apt — but this time the gains are concentrated in sectors where AI is displacing cognitive work, not just automating manufacturing. For investors, this is the clearest signal yet: the returns are accruing to the owners of the machines, not the operators.

This distinction — potential vs. performance — is crucial. It means the labor displacement hasn't fully happened yet, which is why the thesis is still an investment opportunity rather than yesterday's news. The market is pricing in future AI productivity, but the workforce disruption is still in early innings. The companies making these bets are telling you their plans. Listen.

A credible counterpoint, and we include it because the thesis should be tested, not protected. Oxford's argument: it's cyclical, not structural. They may be right — for now. But their own data shows entry-level hiring is collapsing regardless of the cause. Whether the mechanism is AI displacement or AI-justified cost cutting, the outcome for workers is the same: fewer jobs, lower bargaining power, and a labor market being restructured. We watch for the productivity spike that would confirm true substitution.

When VCs volunteer labor displacement predictions without being asked, that tells you something about where private capital sees the world going. These aren't labor economists theorizing — they're investors deploying billions into companies built to replace human work. The smart money isn't debating whether AI will displace labor. They're funding it.

The convergence of graduate and non-graduate unemployment rates may be the most underreported economic story of the decade. The entire premise of the knowledge economy was that education = employment security. That premise is breaking. If a college degree no longer guarantees a job advantage, the social contract around work, education, and economic mobility is being rewritten. This is where the post-labor thesis meets real people's lives.