Survey says 99% of executives are ‘prepared’ for AI layoffs in next two years

Source: Mashable / Mercer

Published: 2026-05-25

Entity Analyzed: Global Executive Workforce Sentiment

URL SCAN

Mashable reports on Mercer’s Global Talent Trends 2026 survey of 12,000 executives, HR leaders, and employees: more than 99% of executives expect AI to lead to at least some headcount reduction in the next two years, 98% are planning organizational design changes, and 65% expect between 11% and 30% of their workforce to be redeployed or reskilled because of AI. So far in 2026, Amazon, Atlassian, Block, Fiverr, Pinterest, and Snap have announced AI-linked layoffs. An estimated 50,000 AI layoffs occurred in 2025. Goldman Sachs CEO David Solomon argues the AI job apocalypse is overblown, while a Harvard study found generative AI increasing short-term demand for augmentation-prone roles. Only one-third of executives believe human and machine capabilities can be effectively combined, and more than a third of employees would consider leaving if they feel disadvantaged by AI.

The Triage

This is not a survey about AI. This is a survey about executive intent, and the intent is unanimous. 99% is not a majority — it is a consensus so complete that the 1% who dissented are statistical noise. When 12,000 respondents across 16 industries and every management tier agree that headcount reduction is coming, the question is no longer whether AI will displace workers. The question is whether any worker in any industry can still pretend they are safe.

The triage reveals three fractures. First: the executive class has already decided. The 98% planning organizational design changes are not waiting for AI to mature. They are restructuring now, before the technology is fully reliable, because the competitive pressure to show AI-driven efficiency gains to investors is greater than the operational risk of premature automation. Second: the gap between executive belief and operational reality is a chasm. Only 32% believe their workforce can combine human and machine capabilities effectively, yet 99% expect layoffs. This means the cuts are not contingent on successful human-machine collaboration. The cuts are contingent on something else entirely: the belief that labor is a cost to be eliminated, not a capability to be evolved. Third: the employees know. More than a third would leave if they feel disadvantaged by AI — which means they already do. The Mercer data captures not a future disruption but a present estrangement, where the workforce and the C-suite are operating on fundamentally different timelines.

The triage verdict: this survey is the end of the AI jobs debate. The debate was always a public relations exercise. The numbers make it official.

The Autopsy (with DT-LAG)

Mechanical Collapse Point

The mechanical collapse is not that AI has replaced 50,000 workers in 2025, or that Amazon and Snap are cutting jobs in 2026. The mechanical collapse is that 99% of executives have crossed a psychological threshold where labor reduction is no longer a risk to be managed but a plan to be executed. The Mercer data reveals a decision architecture, not a prediction. When 65% of executives expect 11-30% of their workforce to be redeployed or reskilled, the operative word is not redeployed. It is or. The or is doing the work. Redeployment implies the worker moves to a new role. Reskilling implies the worker learns a new skill. Both assume the worker remains employed. But the 99% headcount reduction figure sits beside the 65% redeployment/reskilling figure like a shadow: the 35% gap is the workers who will not be redeployed, will not be reskilled, and will not be retained.

The mechanical reality is that organizational design changes at 98% of companies means 98% of companies are currently drawing new org charts with fewer boxes. The cuts are structural, not cyclical. Amazon, Atlassian, Block, Fiverr, Pinterest, and Snap are not reacting to quarterly pressure. They are implementing a multi-year workforce compression strategy that Mercer has now quantified. The 50,000 AI layoffs in 2025 were the pilot. The 99% executive consensus means 2026-2028 will be the rollout.

Lag-Weighted Social Timeline

Immediate (0-6 months): The survey becomes the reference point for every HR department justifying cuts. The 99% figure will appear in memos, board decks, and investor presentations as external validation. The immediate social reality is that the executive class now has cover: they are not making a risky bet on AI. They are following industry consensus. The immediate lag is that employees will not see the survey as a warning. They will see it as the moment their employer stopped pretending.
Short-term (6-18 months): The companies named in the survey (Amazon, Atlassian, Block, Fiverr, Pinterest, Snap) become case studies for industries not yet cutting. The short-term mechanical reality is that the 65% redeployment/reskilling expectation will be tested and will fail for a significant portion of workers, because reskilling programs at most corporations are underfunded, misaligned with actual AI capabilities, and timed to coincide with severance negotiations. The short-term social reality is visible panic among early-career workers, who are disproportionately affected, and visible complacency among senior workers, who believe their judgment-based roles are safe. Both are wrong. The senior workers are next.
Medium-term (1-3 years): The 99% consensus produces a self-fulfilling prophecy. Executives who were uncertain about AI layoffs now implement them because the survey tells them everyone else is doing it. The medium-term mechanical reality is a competitive race to the bottom: no company wants to be the one with higher labor costs than its AI-automated competitors. The medium-term social reality is that the 35% of workers not covered by the redeployment/reskilling figure have been absorbed into the gig economy, entrepreneurship, or long-term unemployment, and the labor market has not created equivalent replacement roles at equivalent compensation.
Long-term (3-7 years): The Harvard Business School finding that generative AI increases short-term demand for augmentation-prone roles will have inverted. The long-term mechanical reality is that augmentation-prone roles were a temporary bridge: humans trained the models, refined the outputs, and built the workflows that eventually removed the need for human augmentation. The long-term social reality is that the Goldman Sachs optimism — David Solomon’s argument that the US economy adapts to technology — will be tested against the largest structural labor displacement since industrialization. The lag is that economic adaptation at the macro level does not translate to individual survival at the micro level.

Lag Factors

The ‘Or’ Deception: The 65% figure promises redeployment or reskilling, but the 99% figure promises headcount reduction. The lag is that workers hear the 65% as a safety net and miss the 35% gap. The or is not a promise. It is a statistical distribution, and a significant portion of the workforce will fall into the neither-nor category: neither redeployed nor reskilled, simply removed.
The Solomon Counter-Narrative: Goldman Sachs CEO David Solomon’s argument that the AI job apocalypse is overblown is not wrong at the macro level. The US economy has adapted to technological disruption. But the lag is that macro adaptation takes decades, and micro displacement happens in months. The workers being cut in 2026 will not be reassured by economic theories about 2040. The Solomon argument is a lag factor because it provides executives with a moral framework for cuts they have already decided to make.
The Harvard Short-Term Augmentation Trap: The Harvard study finding that generative AI increases short-term demand for augmentation-prone roles is the most dangerous lag factor. It tells workers that AI is creating jobs, not destroying them. The lag is that augmentation demand peaks early in the adoption curve, when models are unreliable and need human correction. As models improve, augmentation demand collapses. The workers who entered augmentation roles in 2025-2026 will discover in 2028-2029 that the role was transitional, not foundational.
The Pew Encroachment Illusion: Pew Research found that 65% of American workers say AI has not encroached on their jobs. This is the lag of perception. The mechanical reality is that 21% say AI is already partially doing their work, and that percentage is concentrated among younger, early-career workers. The 65% who feel safe are, statistically, older workers in roles that AI has not yet reached. The lag is that AI moves up the skill ladder faster than workers perceive the movement. By the time the 65% feel encroached, the encroachment will be irreversible.

Defensive Moats

The Reskilling Lottery: The 65% of executives planning redeployment or reskilling creates a narrow moat for workers who can position themselves as AI-human interface specialists. The moat is narrow because most corporate reskilling programs are theater — underfunded, poorly designed, and timed to coincide with layoff announcements. The workers who survive will be those who reskill themselves, not those who wait for employer programs.
Regulatory Theater as Delay Mechanism: California’s Gavin Newsom has issued executive orders attempting to protect workers from AI displacement. The EU AI Act contains labor provisions. The moat is that regulatory action moves slowly while executive action moves fast. The 99% consensus will produce cuts faster than any legislature can produce protection.
The Exit Window: More than a third of employees would leave if they feel disadvantaged by AI. The moat is that the most marketable workers — the ones with portable skills and competitive profiles — will exit before the cuts arrive. This creates a brain drain that accelerates organizational decline, but it also means the workers who move early avoid the competition of a flooded labor market. The lag is that most workers will not move until they feel the disadvantage, and by then the exit window is crowded.
Narrative Collapse of Augmentation: The Harvard study’s short-term augmentation finding will collapse as models improve. The moat for workers in augmentation roles is to recognize that augmentation is a phase, not a destination, and to use the augmentation period to build skills that are not augmentable: judgment under uncertainty, ethical reasoning, creative synthesis, and interpersonal trust. The lag is that most augmentation workers are too busy augmenting to build these skills.

Future-Proofing Scorecard

| Timeline | Score | Commentary |
|———-|——-|————|
| 1 year | 2/10 | The 99% consensus produces a cascade of announced layoffs. Companies race to demonstrate AI-driven efficiency to investors. Early-career workers bear the brunt. The 65% redeployment promise begins to fray as reskilling programs prove inadequate. |
| 2 years | 1/10 | The augmentation-prone roles identified by Harvard begin to compress as models mature. The 35% gap in the Mercer data — the workers neither redeployed nor reskilled — becomes visible unemployment. The gig economy absorbs some, but not at equivalent wages. |
| 5 years | 0/10 | The 99% executive consensus has been fully executed. The workforce structure of most Fortune 500 companies is unrecognizable compared to 2025. Middle management has been algorithmically replaced. The concept of corporate employment as a stable career path is historical memory for anyone under 35. |
| 10 years | 0/10 | The Mercer 2026 survey is studied as the turning point document — the moment when executive consensus made AI layoffs inevitable, independent of technological readiness. The 1% who dissented are footnotes. The 99% who agreed built the new economy. The new economy does not need them either. |

The Verdict

This article is not a warning. It is a census of a decision already made. The Mercer Global Talent Trends 2026 report, with its 12,000 respondents and its 99% executive consensus, is the most important labor document of the decade because it removes the last ambiguity from the AI jobs debate. The debate was never about whether AI could replace workers. It was about whether executives would choose to replace workers. The survey answers that question with a number so close to unanimity that dissent is statistically irrelevant.

The verdict is that the AI jobs apocalypse is not coming. It is here, and it is executive-sponsored. The 50,000 AI layoffs in 2025 were not an anomaly. They were a pilot program. The 99% consensus means that every company surveyed — and by extension, every company competing with those surveyed — is now structurally committed to headcount reduction. The 98% planning organizational design changes are not waiting for better AI. They are drawing new org charts now. The 65% expecting redeployment or reskilling are not planning a soft landing. They are planning a statistical distribution where a significant portion of the workforce will receive neither redeployment nor reskilling, because the 99% headcount reduction figure does not reconcile with the 65% preservation figure. The math is simple: at least 34% of the affected workforce is scheduled for removal.

The verdict on the counter-arguments is equally clear. David Solomon’s macro optimism is correct and irrelevant. Economies adapt over decades. Workers need to eat next month. The Harvard Business School finding that generative AI creates short-term augmentation demand is correct and dangerous, because augmentation is a phase, not a destination, and the workers who enter augmentation roles believing they are safe will be the most surprised when those roles evaporate. The Pew Research finding that 65% of workers feel unthreatened is correct and tragic, because the feeling of safety is the lag factor that prevents preparation.

The discontinuity is not technological. It is decisional. The technology did not force 99% of executives to plan layoffs. The executives chose to plan layoffs, and the technology provided the justification. The Mercer survey is the document that makes this choice visible. The verdict: the AI jobs crisis is not a side effect of progress. It is the intended outcome of a strategy. The 99% consensus is the strategy’s signature. And the workers who still believe their jobs are safe are the strategy’s target.