American Employment: A Tale of Two Worlds - Carbon-Based and Silicon-Based

Since the beginning of the year, the overall employment situation in the United States has shown signs of stabilization, but a closer look at the structure reveals a much more complicated picture. The creation of new jobs is highly concentrated in a few essential services industries such as healthcare, with a narrow coverage of employment recovery in the entire industry and insufficient broad-based recovery momentum. At the same time, the total number of jobs in the information industry has declined by over 300,000 since its peak in November 2022. This fractured employment environment among industries seems to directly point to the labor force restructuring brought about by AI replacing carbon-based labor structures. Several micro indicators have also been releasing similar warning signals: Overseas leading technology giants are normalizing personnel optimization and layoffs, with AI-related layoffs accounting for over 20% of the total layoffs caused directly by AI technology substitution in the first five months of 2026. So, what role is AI playing in the U.S. labor market currently? How should its short and medium-term impacts on employment be measured? 1. How significant is the negative impact of AI on employment? The wave of layoffs in the current technology industry and the AI diffusion is essentially a result of the combination of cyclical and structural factors: Firstly, the layoffs in the U.S. technology industry in recent years have certain cyclical correction characteristics. It should be noted that the layoffs in the U.S. technology industry since 2023 cannot be entirely attributed to AI skill substitution. Instead, they are largely a result of the "cyclical regression" due to the excessive labor during the pandemic period. From 2020 to 2022, catalyzed by the demand for digitalization due to the super-low interest rate environment and home economy transformation, leading technology giants mistakenly interpreted short-term dividends as long-term sustainable demand and initiated defensive expansion. During this period, both Meta and Amazon saw their employee numbers increase by over 90%, and Alphabet's three-year employee growth rate also reached 60%. By November 2022, the number of non-farm jobs in the U.S. information industry reached a historical peak of 3.12 million. With the Federal Reserve raising interest rates and macro total demand returning to its mean, the excessive labor costs from the previous period became unsustainable, and companies were forced to initiate large-scale layoffs. A comparison with the information industry, which is deeply affected by AI, clearly shows that compared to the overall private sector non-farm employment (which is basically in line with the pre-pandemic trend), the employment trend in the information industry in 2022 has deviated significantly from the long-term trend. As the economic cycle returns, non-farm employment in the information industry has fallen by more than 10% from its peak in November 2022, but when compared to the reasonable position corresponding to the long-term trend line, the previous excessive expansion actually only cleared 6%. In addition to the return of the economic cycle, the rise of AI capital expenditures further influences companies' resource allocation logic. It is worth noting that some of the layoffs by giants are not directly due to the skill substitution of AI for existing labor but rather the "shift of resources to computing power" under the high-stakes AI arms race. Faced with the paradigm shift in AI, tech giants have spared no expense in increasing computing power and AI infrastructure investment. The combined capital expenditures of Google, Amazon, Microsoft, and Meta increased from $142.6 billion in 2023 to $372.6 billion in 2025. In the first quarter of 2026, the capital investment further exceeded $128.8 billion. The expansion of capital expenditures and the reduction in recruitment positions and freeze of vacant positions occur simultaneously. Just as in the historical period of enclosure movements in England, where Thomas More referred to the social phenomenon of the wool industry occupying farmers' land as "sheep eating people," this phenomenon is also evident in the current AI employment impact. The four super-large tech companies of Microsoft, Google, Meta, and Amazon are expected to have a combined capital expenditure of $720 billion in 2026, equivalent to 2.3% of the U.S. GDP. In this ultra-capital-intensive competition, the high depreciation of computing power and R&D investment inevitably compress traditional labor costs. AI catalyzes "unemployment," but its micro mechanism is not solely "machines replacing humans" but rather "capital expenditure squeezing out expense expenditure" on financial statements. Therefore, excluding cyclical factors, the actual replacement effect of AI on the workforce is still in its early stages. Moreover, structurally, this impact is not evenly spread but highly concentrated in high AI exposure, medium-to-high-paying, and young age groups: Industry-wise, according to research from the Federal Reserve Bank of St. Louis, since 2022, there is indeed a certain correlation between the proliferation of AI and the rise in unemployment. However, this impact is more concentrated in cognitive-intensive industries such as computer mathematics, finance, business management, and arts and creativity. These industries have a higher exposure to AI and involve a large amount of repetitive cognitive work such as standardizing information processing, quantitative measurement, drafting standard documents, and modular solution design. At the current stage, generative AI can efficiently handle such process-oriented tasks, making it the first area where labor force displacement is evident; in contrast, positions highly dependent on non-standardized on-the-spot communication, offline production, and personalized decision-making are less affected by short-term AI replacement. In addition, the vulnerability of highly educated young people in the wave of AI iterations is significantly increasing. The final impact of AI is highly concentrated at the entry level of the industry and is creating a "new talent gap" through its "substitution effect." Using high-frequency salary and employment data from ADP, Stanford University has empirically calculated that since the end of 2022, the employment scale of young people aged 22-25 in high-exposure AI positions has dropped by 16%, while employment among senior professionals in the same field has remained stable. The "new talent gap" created by AI is not just a temporary employment issue but may represent a deeper structural change. Dario Amodei, CEO of Anthropic, issued a warning in May 2025, predicting that AI could eliminate about half of entry-level white-collar jobs in five years. The decrease in entry-level white-collar jobs will disrupt the current career advancement balance, making it harder for businesses to balance talent development and external recruitment. Therefore, even though AI has objective substitution effects on the labor market, its impact is structurally unique. Currently, the main focus of employment impact is on suppressing the supply of new jobs and mass elimination of existing employees. Data from the New York Federal Reserve corroborates this feature: companies are currently using AI to reduce new hiring and opt for employee retraining, while being relatively cautious in the process of layoffs. At the same time, the employment transformation brought about by AI is not entirely negative, as it continues to nurture new job opportunities while eliminating some positions. Harvard Business School has quantitatively analyzed the structural disparities in the job market: employment demand for highly standardized, labor-intensive automated jobs has dropped by 24%, while demand for AI-enhanced positions that rely on specialized skills and value-based decision-making has increased by 15%. In other words, while AI is reducing the demand for traditional job skills in automated environments, it is simultaneously creating new job opportunities that require specialized skills, driving the iterative restructuring of labor skills. While the new demands may not match the pace of the replacement effects, they do serve as a counterbalance. Data on recruitment postings from job platforms further confirm the structural differentiation in the labor market. According to data disclosed by Indeed platform, while overall recruitment volumes in various industries have shrunk, the demand for AI-related positions has increased against the trend. Since 2023, the number of recruitment positions related to AI has increased by over 200%, reflecting the current industrial situation in which AI reshapes employment structure by causing shifts in demand for new and old positions. 2. Short-term employment in the United States: Cyclical factors outweigh structural impacts Although AI has caused structural disturbances in specific sectors, its overall impact on the total employment landscape in the United States is relatively limited due to the early stage of industrialization in AI: On one hand, the employment base in the industries affected by AI is limited, making it difficult to generate widespread job substitutions. Information technology and finance, as the two major areas most affected by AI substitution, account for a relatively low percentage (about 8%) of total private non-farm employment in the United States. The scale of position reductions and layoffs in these areas is unlikely to significantly impact the overall employment landscape in the U.S. Other physical industries have a low penetration rate of AI industrialization, and the job replacement effects of technology on a larger scale are relatively mild. On the other hand, the effects of AI-induced layoffs and new job creations have limited disturbances on the overall pattern of the current job market, which is characterized by a lack of layoffs and recruitment. Looking at the layoffs, according to data from the Challenger Association, although the average monthly layoffs caused by AI have increased by 283.9% year-on-year in the first five months of this year, the total labor force layoffs have decreased by 21%, with the increase in the proportion of AI-related layoffs attributed to the passive increase in cyclical factors (reduced DOGE layoffs, economic stabilization, etc.); Furthermore, the recruitment side exhibits clear structural differentiation, with the increased demand for AI-related positions on the Indeed platform seemingly unable to counterbalance the widespread recruitment contraction across the entire industry. Therefore, in the short term, cyclical factors play a more important role in the U.S. employment market. The recent overall improvement in non-farm employment in the U.S. reflects economic recovery under previous preventive rate cuts. The significant rebound in job vacancies in April also signals a positive trend, reflecting a marginal increase in business demand for labor, which may indicate that the unemployment rate may not rise rapidly in the short term and the downward pressure on wages may also be limited. Combined with the high risk of input inflation, the likelihood of a sharp decline in the short term of the Federal Reserve interest rates has significantly decreased. Looking ahead, however, doubts remain about the breadth and sustainability of this round of employment recovery. On one hand, the negative impact of increased energy costs due to the Middle East geopolitical conflict has yet to fully transmit to the employment side and may restrain the subsequent non-farm recovery pace; on the other hand, the current employment recovery shows significant structural differentiation, with insufficient coverage of industry-wide revival. Therefore, the predominant features of the U.S. employment market in the second half of the year will continue to be the lack of layoffs, lack of recruitment, and weak wage levels. This implies that there is a low probability of the Federal Reserve raising interest rates. 3. In the long run, the real turning point has yet to arrive Current research results are subject to some lag effects. The conclusions drawn based on generative AI research may not fully apply to the intelligent body AI that has been rapidly implemented since 2026. Regardless of how deeply generative AI intervenes in the value chain and workflow, humans remain the "owners" of these processes or stages, with demands initiated by humans and judgments and decisions made by humans. Intelligent body AI breaks the existing rule of "replacing tasks rather than positions," being able to complete end-to-end workflow tasks that involve multiple steps of reasoning, tool usage decisions, and autonomous decision-making in the absence of human employees, and making judgments and iterative corrections based on results. For the labor market, this means that AI may soon see a surge in its substitutive effects on positions. Structurally, the impact will spread from entry-level positions to mid-level white-collar workers, and blue-collar service industries. Once these groups experience simultaneous pressure on employment and wage growth, the multiplier effect on the consumer side will result in macro impacts beyond the direct manifestation in employment numbers. If the macro realization cycle of productivity gains continues to shorten, the accompanying restructuring of employment structures will also come at a faster pace. From a macro market perspective, this acceleration of technology and institutional lags is generating structural domestic demand suppression risks. The combination of "profit growth, weak employment, stable wages" in the "no employment prosperity" scenario is likely to erode the foundations of domestic demand once intelligent body AI further penetrates, and it could be difficult to offset through monetary policies alone. In conclusion, in the medium to long term, the impact of AI on employment is evolving from "task substitution" to "position substitution," from creating a "new talent gap" to penetrating mid-level white-collar workers, and accumulating structural pressure from localized effects. However, this process is gradual and will be accompanied by the emergence of new positions and the reallocation of labor market rather than an immediate cliff-like replacement. Risk Warning: Risk of macro information distortion; Risk of non-linear increase in AI substitutability; Risk of lack of labor market buffers.