‘AI is now the leading reason companies give for cutting jobs,’ says new report—what that means for workers

Source: CNBC

Published: 2026-06-06

Entity Analyzed: Tech Capital Reallocation

URL SCAN

U.S. employers announced just over 97,000 job cuts in May 2026, the highest May total since 2020. Employers cited AI as the primary reason for almost 40% of May’s announced job cuts — up from 7% in January. The total AI-attributed cuts in the first five months of 2026 reached 87,714, already surpassing the 54,836 recorded for all of 2025.

The Triage

This is not a doomsday story. It is something more insidious: a doomsday narrative that experts are actively debunking while the data keeps getting worse. The CNBC piece is notable because it gives voice to the skepticism that most coverage suppresses. Daniel Zhao of Glassdoor says companies may be ‘scapegoating’ AI. Fabian Stephany of Oxford calls it ‘projection into AI’ — using the technology as ‘a good excuse.’ Columbia’s Daniel Keum says the labor market is ‘humming along just fine’ with payrolls up 172,000.

The triage diagnosis: the experts are right, and that is the problem. The layoffs are not driven by AI capability. They are driven by AI narrative. When attribution rises from 7% to 40% in five months — faster than any technology could plausibly deploy — the causality is backwards. The cuts are not happening because AI works. They are happening because saying ‘AI’ works.

The critical observation is what the skeptics inadvertently reveal. Zhao says companies are changing resource allocation. Keum says the impact is ‘very concentrated’ in tech. Stephany says efficiency gains are not the real driver. Together they describe a system where capital is being reallocated not by technological necessity but by narrative permission — and the experts are so focused on debunking the technology that they miss the mechanism of the narrative.

The Autopsy (with DT-LAG)

Mechanical Collapse Point

The 7% → 10% → 25% → 26% → 40% trajectory of AI-attributed layoffs is not a technology adoption curve. It is a narrative contagion curve. The mechanical reality: companies are cutting workers and citing AI because investors, analysts, and the business press have created a social license to do so. The 87,714 AI-attributed cuts year-to-date already exceed the full-year 2025 total of 54,836. The technology did not improve enough in five months to justify this acceleration. The attribution did.

The article’s most important data point is hidden in plain sight: employers announced 80,742 planned hires in May, described as ‘historically low by prepandemic standards.’ This is not a hot labor market. It is a cold labor market with warm payroll numbers. The 172,000 payroll growth is real but it is not replacement. Thomas Thompson captures the mechanical reality: ‘the jobs that are open aren’t replacing the jobs that are lost.’ An engineer replaced by AI in biopharmacy does not become a warehouse worker. The skills are not transferable because the problem is not a skills gap. It is a role elimination.

Lag-Weighted Social Timeline

Phase 1 (Now – Q3 2026): The ‘don’t panic’ narrative dominates. Experts correctly debunk AI’s doomsday potential while the layoffs accelerate. The 40% attribution becomes normalized. Workers believe the experts and do not prepare for displacement.
Phase 2 (Q4 2026 – Q2 2027): The skepticism becomes irrelevant. Whether AI caused the layoffs or merely justified them, the workers are still gone. The 80,742 planned hires — already historically low — contract further. The ‘humming along’ macro narrative collides with the micro reality of structurally unemployed knowledge workers.
Phase 3 (2027-2028): Political and institutional recognition arrives. By then the narrative has served its purpose: the capital has been reallocated, the infrastructure is depreciating, and the workforce has been restructured. The experts who said ‘don’t panic’ were correct about the technology and wrong about the consequences.

Lag Factors

Expert Credibility Lag: The article’s experts — Keum, Zhao, Stephany — are quoted as reassurance. Their skepticism is framed as ‘don’t panic.’ But the skepticism actually confirms the problem: if companies are scapegoating AI, the layoffs are real and the justification is false. The reassurance is the lag. Workers who believe the experts will not prepare for a transition that is happening regardless of whether the technology caused it.
Macro Data Masking: The 172,000 payroll growth is genuine but misleading. It measures aggregate employment, not sectoral displacement. The tech sector’s 38,242 May cuts are a drop in the 172,000 bucket. But the bucket is filling with different water. The jobs being created are not the jobs being destroyed. The macro number hides the structural mismatch.
Hiring Theater: 80,742 planned hires in May, ‘historically low by prepandemic standards.’ This is the real labor market signal, not the payroll number. The pipeline is dry. Companies are not replacing the workers they cut. The narrative that ‘AI is not eliminating jobs, just changing them’ requires a hiring market that absorbs the displaced. That market does not exist.
Cognitive Dissonance: The article’s framing asks ‘what that means for workers’ and then quotes experts saying ‘don’t panic.’ The cognitive dissonance is the lag. The reader is told the data is alarming (40% attribution, 87,714 YTD) and then told the interpretation is benign. The gap between data and interpretation is where the displacement happens unnoticed.

Defensive Moats

Regulatory Armor: None discussed in the article. The absence is notable.
Trust Shield: The ‘human touch’ in customer-facing roles. But the article focuses on back-office and technical layoffs, not frontline. The shield is irrelevant to the displaced.
Physical Chains: The article notes that displaced biopharmacy engineers are not going to warehouse jobs. This is not a moat. It is the absence of one. The specialized skills that once provided job security now prevent mobility.
Adaptability: Zhao’s advice to ‘diversify their approach’ and ‘explore other fields’ is the only moat offered, and it is not a moat. It is a hope. The skills are ‘applicable across many different fields’ only if those fields are hiring. The article establishes they are not.

Future-Proofing Scorecard

| Timeline | Score | Commentary |
|———-|——-|————|
| 1 year | 2/10 | The 40% attribution is the baseline. The ‘don’t panic’ narrative delays preparation. The 80,742 planned hires — already historically low — will drop further. The jobs open will not replace the jobs lost. |
| 2 years | 1/10 | The skepticism becomes irrelevant. Whether AI caused the cuts or merely justified them, the workers are gone. The macro ‘humming along’ narrative collides with structural unemployment in knowledge work. Skeleton crews for edge cases and regulatory theater. |
| 5 years | 0/10 | The narrative has served its purpose. The capital is reallocated. The infrastructure is depreciating. The workforce has been restructured around the absence of the displaced. The concept of ‘tech worker’ bifurcates: elite architects versus gig maintenance. |
| 10 years | 0/10 | The employment contract that built the knowledge economy — specialized skills, career progression, industry expertise — exists only in regulatory residuals. The infrastructure remains. The humans do not. The experts who said ‘don’t panic’ were correct about the technology and catastrophically wrong about the system. |

The Verdict

The CNBC article is the most dangerous kind of coverage: accurate data, expert skepticism, and a framing that neutralizes both. The article documents the 40% AI attribution, the 87,714 YTD cuts, the ‘historically low’ hiring — and then quotes experts saying the labor market is ‘humming along just fine’ and workers should not panic about ‘doomsday potential.’ The verdict is that the doomsday is not coming. It is here. It is just not the doomsday the experts are debunking.

The experts are correct that AI is not yet capable of causing this level of displacement. Stephany is right: ‘I’m really skeptical whether the layoffs that we see currently are really due to true efficiency gains.’ Zhao is right: ‘A company can say AI is why we’re doing layoffs, but that doesn’t necessarily mean that’s actually why.’ Keum is right: the macro labor market is fine.

But the layoffs are still happening. The 87,714 workers are still gone. The 80,742 planned hires are still historically low. The ‘jobs that are open aren’t replacing the jobs that are lost.’ The experts debunk the technology while the narrative devours the workforce.

The verdict: this is not technological displacement. It is narrative displacement. AI does not need to work for the layoffs to happen. It only needs to be believable. The 7% → 40% attribution curve proves that belief has been achieved. The capital reallocation is proceeding on schedule. The workers who believed the experts will not panic. They will not prepare. They will not adapt. And they will not be replaced.

The employment contract is being voided not by a superior technology, but by a superior story. The AI era does not need fewer workers. It needs a different narrative, and the humans currently being exited are the characters being written out of the script.