The Economist estimates that AI has created about 1 million new jobs in the U.S. so far
The Economist estimates that AI has created about 1 million new jobs in the U.S. so far, surpassing approximately 200,000 AI-related layoffs since mid-2023, resulting in a net increase in employment effects.
The U.S. Bureau of Labor Statistics reported on September 4 that non-farm employment increased by 162,000 in August, exceeding expectations, with an unemployment rate of 4.1%, which is lower than most months in nearly half a century. The unemployment rate for those aged 20 to 24 is nearly the narrowest gap compared to the overall unemployment rate in decades. The article's title suggests that the doomsday narrative for employment has been postponed, and the AI employment wave has arrived.
Jobs directly serving AI have increased by about 730,000 since 2022, covering engineers, software developers, and mathematics and data science. The Burning Glass Institute estimates that about 1% of professional jobs can be classified as AI jobs, with computer and life sciences reaching 4% to 5%. Annual spending on data center construction has risen by about 60%, exceeding $75 billion, with approximately 320,000 jobs added in five related industries since 2023, including electricians, HVAC, power grid, and commercial building demand rising, with Indeed showing that data center installation and maintenance salaries are about 40% higher than similar positions.
Customer service jobs have decreased by about 10% since January 2023, and administrative assistants have decreased by about 15%. Microsoft and Meta have downsized while restructuring around AI, and Block and Intuit have replaced some positions with automation. Since 2026, companies have announced an average of about 16,000 AI-related layoffs per month; U.S. employers typically lay off about 1.7 million people in a normal month. Noah Smith states that it is still very difficult to accurately identify jobs eliminated by AI in labor data; Pew shows that 50% of adults are more worried than excited about it, up from 37% in 2021.
Construction jobs fluctuate with project cycles, and staffing will decrease once data centers are completed unless new projects continue to start. The Bureau of Labor Statistics data itself also faces revisions and credibility discussions, and the article limits its conclusion to "so far."
In market mechanisms, this is capital expenditure transforming into payroll: large-scale cloud vendors convert bonds and capital expenditures into rebar, copper wire, and cooling systems, with high-skilled jobs consuming model and data work, while blue-collar jobs consume civil construction and electricity. Beneficiaries include electricians, HVAC, power grid contractors, and development positions close to models; those under pressure are standardized customer service and administrative processes. Funds flow from training and cabinet orders to construction contracts and premium hourly wages, rather than being evenly distributed among all white-collar workers.
The net increase is predicated on data centers still being excavated and companies still adding positions for AI jobs. Once the construction peak passes or white-collar replacements accelerate, the same set of data may change signs.
Source: Public information
ABAB AI Insight
The narrative of employment doomsday frames replacement as immediate layoffs, while labor statistics first see data centers and power grids. The 730,000 professional jobs exceeding trends correspond to models, assessments, and data pipelines; the 320,000 construction jobs exceeding trends correspond to cement, copper cables, and cooling. Summing both sides yields a million. Customer service and administrative roles are already declining, but the speed of decline is drowned out by the normal flow of 1.7 million monthly layoffs, leading the media to see a net increase while those affected see their jobs disappearing.
The capital path is very short: cloud vendors issue bonds, order chips, and dig foundations, with wages flowing to contractors and STEM positions. The 40% premium for data center installation and maintenance indicates that the scarcity lies with certified electricians rather than prompts. Microsoft and Meta are laying off while hiring AI positions, representing a replacement of roles within the same balance sheet, not a contraction of the overall economy.
This can be compared to the early days of computer proliferation when secretarial jobs decreased while programmer jobs increased, as well as the surge in electricians and pipefitters driven by shale oil and gas. The current phase is infrastructure expansion: replacement occurs in standardized text roles, while new jobs arise in non-standardized physical installation and modeling roles.
This reflects a restructuring of the industrial chain: wage premiums shift from office processes to electricity and cooling. The mechanism is that computing power must translate into construction, which must hire certified workers, thus when fear surveys rise, the unemployment gap actually narrows.
ABAB News · Cognitive Law
- The narrative of replacement arrives first, while the payroll follows with rebar and electricians.
- Monthly layoffs in the millions can drown out thousands of thematic layoffs.
- Once data centers are completed, the net increase in blue-collar jobs will change signs.