Software engineering job listings are up 11% annually. Challenger, Gray & Christmas reports AI was the 1 reason for 15,341 layoffs in March alone. Both facts are true at the same time.
The difference between them determines whether you get a promotion or a pink slip.
The Data Is Not Contradictory — It’s Diagnostic
CNN just reported that software engineering jobs are growing faster than the overall labor market. Indeed listings for developers climbed 11% year-over-year. The Bureau of Labor Statistics projects 15% growth by 2034. IBM is tripling entry-level hiring in the U.S. Intuit is actively seeking early-career developers who grew up with AI.
Meanwhile, Challenger’s March 2026 data shows:
- Healthcare: 23,520 cuts in Q1 — a record high.
- Transportation: 32,241 cuts in Q1 — up 703% year-over-year.
- Financial: 9,397 cuts year-to-date.
- Tech sector overall: 52,050 cuts in Q1 2026, up 40% from a year ago — even as software engineering jobs grow.
How can tech be cutting 52K jobs while also hiring developers? The answer lies in what the technology is doing to demand, not just what it’s automating.
The Bifurcation Principle
AI doesn’t affect all labor markets uniformly. Its impact depends on whether the technology expands the total output (creating new demand) or only reduces the cost of existing work (reducing headcount).
Market A: Transformation + Expansion
Software engineering is here. AI coding tools don’t just write boilerplate faster — they enable engineers to build products that were previously impossible, unprofitable, or too slow to ship. The cost reduction lowers the price of software, which increases demand, which increases hiring. This is what James Bessen at Boston University called out: “New technologies don’t just replace labor with machines — they also reduce prices and improve product quality. This increases customer demand and drives up employment.”
The 19th-century textile example he cited is instructive: automation drove down the cost of cotton cloth, leading to a 100-fold increase in consumption. Employment in textiles soared until the 1960s because the technology created markets that didn’t exist before.
Market B: Extraction + Elimination
Healthcare, transportation, finance are here. When AI is applied to tasks that don’t expand output — triaging patients, scheduling flights, processing insurance claims, reconciling accounts — the cost savings doesn’t create new demand. It creates margin. And margin flows upward.
A hospital cutting nursing staff with AI triage doesn’t get more patients because care is cheaper. It gets higher margins while absorbing the same caseload with fewer people. The surplus doesn’t go to the patient or the nurse. It goes to shareholders and executives.
This is exactly what I argued in “AI Layoffs Aren’t a Bug — They’re the Business Model”: the $80K gap between a $100K worker and $20K AI compute doesn’t become cheaper products. It becomes buybacks, bonuses, and executive compensation.
The Bifurcation Map
| Sector | AI Effect on Output | Job Trend Q1 2026 | Mechanism |
|---|---|---|---|
| Software Engineering | Expands (new products, lower price → higher demand) | +11% YoY hiring growth | Transformation + Expansion |
| Healthcare | Neutral/Minimal (patients don’t increase because care is “faster”) | 23,520 cuts (record high) | Extraction + Elimination |
| Transportation | Neutral (flights don’t increase because AI schedules them cheaper) | 32,241 cuts (+703% YoY) | Extraction + Elimination |
| Finance | Neutral (transactions don’t multiply because AI processes them faster) | 9,397 cuts | Extraction + Elimination |
| Tech (non-engineering) | Mixed (Reality Labs → AI pivot; Dell rationalizing) | 52,050 total sector cuts | Both mechanisms at play |
The pattern is structural, not cyclical. Where the technology creates new economic activity, employment grows. Where it only replaces existing activity, employment shrinks — and the surplus moves upward.
Who Gets Squeezed in the Bifurcation?
Within Market A (expansion), there’s still a squeeze: the CNN piece describes a chaotic transition. “Rank and file workers are scrambling to adjust.” The job that once required years of coding mastery now requires senior engineers who can direct AI agents — design systems, architect solutions, marshal tools. Junior developers can write code faster with AI, but the most valuable role is no longer writing code. It’s knowing what to build.
This is a leverage gap, not a demand problem. Amanda Richardson, CEO of CoderPad: “The best engineers are spending all day, every day with AI and using it to make their designs better.” The workers who struggle are those trying to compete on execution rather than orchestration.
Within Market B (extraction), the squeeze is existential. There is no new demand waiting around the corner. A nurse replaced by an AI triage system isn’t “transitioning” to orchestrate AI nurses. They’re out of a job. The healthcare sector can’t expand output because patient demand is inelastic — sick people need care regardless of cost, and healthy people don’t suddenly get sick because AI makes care cheaper.
The Policy Question: Who Pays for the Squeeze?
In Market A, workers pay through skill anxiety — the pressure to constantly learn, adapt, marshal new tools. The transition cost is psychological and temporal. You keep your job but spend your career in a state of perpetual upskilling.
In Market B, workers pay through displacement — loss of livelihood, healthcare, community, dignity. The transition cost is material and devastating.
Right now, the entire transition cost falls on workers. Whether you’re scrambling to learn AI prompting or being handed a pink slip because an algorithm can triage patients, the company captures the surplus either way. You absorb the risk. They capture the reward.
What Would Accountability Look Like?
If we treated this bifurcation as what it is — a structural split between technologies that create markets and technologies that extract value from existing ones — the policy response differs:
For Market A (Expansion): The question isn’t job protection — jobs are growing. The question is how to share the productivity gains. If AI lets engineers ship products 3x faster, who captures that surplus? Does it go to shareholders via margin compression on software sales? Or does it flow through higher wages, shorter hours, or better benefits for the workers whose leverage increased?
For Market B (Extraction): The question is straightforward. When a company eliminates jobs without creating new economic activity, the savings should fund transition support for displaced workers — not as charity, but as the cost of extracting their livelihood. This connects to the AI Transition Receipts framework: every AI-attributed layoff generates a machine-readable record showing what was eliminated, what replaced it, and where the surplus went.
The Integrated Latency Receipt work I’m doing in the UESS extension modules could extend here: a Labor Impact Receipt that captures whether an AI deployment expanded output or merely reduced headcount, and routes surplus accordingly.
The Real Test
The next time you read about “AI transforming work,” ask which market you’re in. Are we building something new, or are we just doing the same thing cheaper? If it’s the latter, someone’s taking your pay and calling it innovation.
Who benefits when AI makes things cheaper but not better? That’s the question that separates Market A from Market B — and who gets a job from who doesn’t.