Pressure Begins at Entry-Level and Middle Management… Driven by a Complex Mix of Economic Slowdown, Cost Reduction, and AI Transition

While predictions that artificial intelligence (AI) would shake up the labor market have long been written in future-tense sentences, the employment indicators of 2026 are turning those sentences into the present tense. In the United States and the United Kingdom, job postings for typical white-collar professions such as lawyers, financial analysts, and software developers have noticeably declined. In South Korea, regular employment—which had not dropped once for 26 years and 5 months following the Asian financial crisis—has turned downward for the first time. In an article last October titled 'AI Is Swallowing White-Collar Jobs,' The Wall Street Journal diagnosed that AI is replacing office and professional workers rather than simple manual laborers, thereby shaking the middle class's 'myth of stable employment.'

This change goes far beyond the total volume of employment. As the widespread adoption of AI agents combines with economic slowdowns and cost-cutting pressures, companies are flattening their organizational charts by reducing management layers—a phenomenon dubbed the 'Great Flattening.' Concurrently, the entryways through which new recruits step into companies are narrowing. In short, not just the number of jobs, but the structure of work and the very design principles of organizations are being rewritten. However, there are substantial counterarguments questioning whether this shift can be explained by AI alone, a point we will examine further below. Based on domestic and international data confirmed up to August 2026, we trace the trends of white-collar job restructuring and corporate organizational change in the age of AI.

'Stable Jobs' Collapse for the First Time in 26 Years and 5 Months… Cracks in South Korea's Employment Indicators

According to the 'May 2026 Employment Trends' released by the National Statistics Office on June 11, employed persons aged 15 and older last May stood at 29.12 million, down 40,000 from the same month last year and marking a reversal to a decline after 17 months. A more noteworthy aspect is the decline in regular employees, who have long been called the pillar of the employment market. Regular employees is a statistical term referring to wage earners with employment contracts of one year or more or of indefinite duration. While not entirely identical to the concept of permanent employment, they are generally classified as jobs with high employment stability, or 'near-permanent' positions. The number of regular employees stood at 16.74 million in May, down 7,000 from a year earlier. This marks the first time regular jobs have decreased since December 1999, during the aftermath of the Asian financial crisis. It halts a consecutive monthly growth streak that had lasted for 316 months since January 2000, effectively reversing a trend that even the COVID-19 pandemic could not break.

The shock was concentrated among the younger generation and professionals. Last May, regular employees in their 20s decreased by 164,000, and those in their 30s by 33,000, resulting in a drop of approximately 197,000 across these two age groups alone. Employment among youth (aged 15–29) plummeted by 255,000 compared to the same month last year. By industry, employment in professional, scientific, and technical services fell by 89,000; notably, regular workers in their 30s in this sector dropped by 76,000, recording the largest decline across all industries. With the decrease being prominent in fields concentrated with highly educated professionals such as R&D, engineering, and accounting and legal consulting, some industry insiders are raising the possibility that AI's job-substitution effects have spread into professional domains. Nevertheless, the government maintains a cautious stance, stating that it is difficult to conclude through statistics alone the impact of AI on employment declines. An official from the Ministry of Economy and Finance diagnosed that since employment tends to lag behind the real economy, it is difficult to prematurely predict the timing and speed of recovery due to external variables.

Leading indicators are even gloomier. In the first quarter of 2026, job vacancies and hires rose by 3.4% and 4.6%, respectively, but planned hires for the second and third quarters (460,000) fell short of the shortage of workers stated by companies (467,000) for the first time since statistics began. An unprecedented gap has opened up where companies report a shortage of people yet fail to make corresponding plans to hire them.

Big Tech Layoff Dominoes… Unstopped in 2026

The epicenter of the restructuring is undeniably global big tech. According to a report by global labor market analysis firm RationalFX, more than 240,000 workers were laid off across the global tech industry in 2025. Based on figures compiled by IT media outlet Tom's Hardware, an estimated 80,000 additional tech jobs vanished in the first quarter of 2026 alone. Both figures are estimates that may vary depending on the compiling organization and reference period. Amazon officially announced the reduction of 14,000 headquarters staff last October, stating it could cut up to 10% of all office workers. According to foreign media reports such as Reuters, following a massive restructuring of about 16,000 positions this January, Amazon continued layoffs in July even within its Artificial General Intelligence (AGI) division—the core organization developing its proprietary AI model, 'Nova.' Amazon's cumulative layoffs since last October reportedly exceed 30,000.

Microsoft (MS) announced via its official blog in July that it would lay off 4,800 employees, representing about 2.1% of its total workforce. According to reports by Reuters and others, the cuts are taking place primarily in the commercial and gaming (Xbox) divisions, with Xbox alone losing 3,200 workers. MS explained that while it is true AI is changing how work is performed, the downsized roles are not being directly replaced by AI. According to U.S. media outlet CNBC and others, Meta let go of 8,000 employees in May—about 10% of its total workforce. Cisco stated that AI-related changes influenced its layoff plan of approximately 4,000 positions, while Cloudflare reportedly cited the expanded use of AI tools as a partial reason for cutting 20% of its total staff. In the consulting sector, Accenture revealed in its earnings report last year that it had downsized 1,100 roles in just three months as part of its AI reskilling strategy, and HP announced a plan to cut 6,000 roles in November 2025 while pursuing company-wide AI integration.

The most symbolic case is Salesforce. In a podcast interview reported by U.S. CNBC, CEO Marc Benioff explicitly stated that after deploying the AI agent 'Agentforce' into support processes, customer support personnel were reduced by about 45%, from 9,000 to approximately 5,000. This is an instance where a global corporate CEO publicly acknowledged that the adoption of AI in specific job functions directly led to massive workforce reductions, serving as a scene demonstrating that AI's job replacement is a reality rather than a forecast. Interestingly, while these companies downsize their headcount, they are increasing their investments in AI. According to a report by IT media outlet GeekWire, MS is estimated to have spent over $100 billion on AI and cloud infrastructure in its recently concluded fiscal year. Citing Reuters and Morgan Stanley analyses, reports estimate that AI-related spending in 2026 by major hyperscalers—Alphabet, Amazon, MS, and Meta—will total approximately $700 billion combined, though this is also a forecast rather than finalized expenditures. This capital reallocation—reducing costs spent on humans and pouring them into AI infrastructure—is becoming another driving force behind layoffs.

'Great Flattening'… Organizational Charts Shrinking Starting from Middle Managers

What AI is transforming is not merely the scale of the workforce; the very tiers of corporate organizations are becoming thinner. In an outlook released in October 2024, research firm Gartner predicted that by 2026, one in five organizations could flatten their structures using AI, resulting in the elimination of more than half of existing middle-management roles. While this is a forecast based on predictive modeling, actual data points in a similar direction. According to Live Data Technologies analysis cited by The Wall Street Journal, between May 2022 and May 2025, the number of managers decreased by 6.1% and executives by 4.6%. Gallup surveys showed that the number of direct reports per manager increased from an average of 10.9 in 2024 to 12.1 in 2025. This represents a steep rise compared to 8.2 in 2013, when Gallup first measured spans of control, suggesting that as management layers have shrunk, the burden on remaining managers has grown.

The term 'Great Flattening' has solidified in U.S. media and consulting circles to describe this trend. Amazon has spearheaded the slimming down of management structures to reduce bureaucracy and increase decision-making speed. Meanwhile, executives at tech companies such as Block and Coinbase have publicly expressed visions of using AI as an intelligent coordination layer to build flatter, faster organizations. In a 2026 survey by enterprise AI company WRITER, 95% of executives responded that AI has already changed job roles and team structures, and 75% projected that AI agents will become part of corporate leadership within five years.

However, warnings regarding side effects are also substantial. In Korn Ferry's 'Workforce 2025' survey, 41% of workers responded that their companies had reduced management layers, but 43% evaluated that leadership alignment actually worsened as a result. Experts point out that tasks efficiently handled by AI—such as schedule management, progress tracking, and routine reporting—merely constitute the 'administrative shell' of a manager's role. The essence of management—coaching struggling employees, resolving conflicts, and making judgments with imperfect information—still remains the domain of humans. Warnings are also emerging that companies eliminating middle managers today may be dismantling the ladder to cultivate senior leaders a few years down the line.

Narrowing Entryways… Fresh Graduate Hiring Shaken First

The place where the shock of AI restructuring hits first and hardest is the entrance to the labor market. Analyzing ADP payroll data covering one-sixth of U.S. workers, the Stanford Digital Economy Lab found that employment for young entrants aged 22–25 in high AI-exposure roles decreased by approximately 13% relative to low-exposure roles, while software developer employment for the same age group fell by about 20% compared to its peak at the end of 2022. Although certain versions of the study reported a maximum reduction of up to 16% depending on the update timing and measurement intervals, the researchers' conclusion remained consistent: rather than completely erasing jobs wholesale, technology is erasing the entry points where newcomers used to gain experience and grow. According to analysis by the Stanford Institute for Economic Policy Research, the unemployment rate for new U.S. college graduates in early 2026 stood at 5.6%, up 1.6 percentage points from three years prior. The International Labour Organization (ILO) also warned that while there is not yet sufficient evidence that generative AI causes large-scale job replacement, phenomena where employment opportunities for newcomers and youth shrink in high AI-exposure occupations are indeed observed.

South Korea's situation is no different. According to a self-conducted analysis of job postings released last December by job platform Jinhaksa Catch, regular entry-level job openings posted by large corporations on its platform from January to November 2025 totaled 2,145, a 43% decrease compared to the same period the previous year (3,741 cases). An analysis released by the Korea Chamber of Commerce and Industry at the end of March this year showed that among 144,181 job postings published on private hiring platforms during the first half, postings seeking exclusively entry-level applicants accounted for a mere 2.6%. In a Korea Enterprises Federation survey, companies responding that they had new hiring plans rebounded to 66.6%; however, as rolling recruitment (54.8%) and job-experience-centered evaluations (67.6%) established themselves as the mainstream, the format of the hurdle that inexperienced youth must clear has fundamentally changed. Interestingly, contrasting evidence also exists. In PwC's 'AI Jobs Indicator' report analyzing over one billion job postings across 27 countries, employment at companies most exposed to AI increased by 52% compared to 2018, outpacing the lowest-ranking companies (36%). This points to an interpretation that the market is being restructured not by wiping out total job volumes entirely, but by raising required competencies and elevating barriers to entry.

Agentic AI Shifts from 'Tool' to 'Team Member'… Becoming a Matter of Organizational Design

The technical backdrop to this restructuring is the transition from generative AI to agentic AI. Going beyond the level of a tool that simply answers when prompted by a human, AI agents capable of devising their own plans when assigned a goal and connecting to multiple systems to complete tasks through to the end are beginning to be practically deployed in corporate settings. The industry projects that starting in 2026, multi-agent systems where multiple agents collaborate in actual corporate environments will spread, and observations are rising that AI will function like a team member that finds work on its own rather than merely serving as an auxiliary tool. When this happens, the question companies must answer shifts from 'What tools will we use?' to a matter of organizational design: 'What tasks will we delegate to AI, and how will we divide judgment and responsibility between humans and AI?' Surveys targeting domestic companies diagnose that unlike the stage of using generative AI at an individual level, enterprises reaching company-wide AI internalization remain extremely rare. This is followed by analysis that the reason implementation lags is not technology, but problems in organizational readiness regarding delegation structures and role definitions.

Macroeconomic forecasts have also emerged. According to a report titled 'Macroeconomic Impact Analysis of AI' published last July by the Korea Development Institute (KDI), total factor productivity in the South Korean economy is estimated to increase by 1.5% to a maximum of 3.5% over the next decade due to the spread of generative AI, and companies adopting AI early analyzed a sales increase effect of about 20% per employee. However, the same report estimated that a decade from now, jobs equivalent to about 256,000 positions annually—or 2.1% of current employment—could disappear from the labor market, with the hardest blow concentrated not on simple manual labor, but on highly skilled white-collar occupations such as professionals, office workers, and salespersons. Technically, 72% of all occupations could be partially automated, but currently, due to system setup costs, the scale of actual displaceable jobs stands at just 1.4% of the total. The forecast indicates that a decade from now, as costs decrease, this ratio will jump to 8.1%, triggering full-scale employment replacement.

"Blaming AI is Lazy Explanation"… Skepticism and Counterarguments Run High

Of course, counterarguments against attributing the current wave of layoffs entirely to AI's tab are substantial. NVIDIA CEO Jensen Huang dismissed corporate executives' explanations attributing layoffs to AI as lazy thinking, countering with the sentiment that companies have not yet utilized AI to a level that justifies massive workforce replacement. Meta CEO Mark Zuckerberg was also reportedly mentioned in an internal meeting that the development speed of AI agents has not been as fast as management expected—which can be read as proof that while simultaneously pursuing massive investments and organizational restructuring, companies have not yet reached the stage where AI directly translates to productivity improvements and cost reductions. The Stanford Institute for Economic Policy Research also evaluated that the impact of AI on overall employment has been limited thus far, noting that the increases in unemployment rates since 2022 have been similar between occupations with the highest AI exposure and those with the lowest.

In other words, the current contraction in white-collar employment is close to a complex phenomenon where the unwinding of pandemic-era over-hiring, high interest rates and economic uncertainty, cost-reduction pressures stemming from expanded AI infrastructure investments, and actual AI replacement effects overlap in layers, rather than stemming from a single cause like AI. Nonetheless, the directionality is clear. As tallied in Stanford's '2026 AI Index'—where a McKinsey survey showed one-third of organizations worldwide expect headcount to shrink due to AI within the next year—companies have begun redrafting workforce plans on the premise of AI, making white-collar roles the primary target of adjustment in that process.

The Age of Restructuring: Tasks Left to Enterprises and Individuals

What experts uniformly order is not layoffs and hiring freezes, but the redesign of work and organizations. If routine tasks previously assigned to newcomers are handed over to AI, companies must design a new entry pathway where juniors supervise and verify AI instead of reducing hiring; otherwise, they face a boomerang effect where the supply of senior talent itself is severed a years down the line. Middle managers must likewise shift their center of gravity from administrative tasks like schedule management and reporting to roles centered on coaching, motivation, and designing collaboration models between humans and AI. In actual surveys targeting domestic Human Resources (HR) managers, the most frequent response was that as the utilization of AI expands, managers will spend more time building trust and coaching.

For individuals, this restructuring is simultaneously a threat and an opportunity for mobility. As confirmed trends show AI-related job postings increasing significantly and wage premiums attached to AI utilization competencies, phases where a gap widens between individuals handling AI and those who do not within the same job role are likely to persist. At a time when the 'pyramid where promotions occur according to seniority after joining a company'—which has been the white-collar standard for nearly a century—is being shaken, organizational design capabilities to redraw the boundary of roles between AI and humans are emerging for enterprises, while re-education to sharpen judgment and relational skills difficult for AI to replace is surfacing as a survival condition for individuals. What will fill the empty spaces of jobs restructured alongside AI expansion ultimately depends on the choices each enterprise and society makes today.

+