Bank of America estimates that since 2026, the US non-residential construction industry has added approximately 95,000 jobs, while AI and data center-related manufacturing has added approximately 32,000, together contributing about a quarter of the private sector job growth this year.
The latest BLS data also supports the first part of this trend: as of July, non-residential construction employment was still up about 3% year-on-year, while residential construction employment was down 1.5% year-on-year, showing a very clear divergence in employment.
This indicates that AI capital expenditure has already been transmitted to the labor market through "physical construction": data centers require construction workers, electricians, and personnel manufacturing server, power, and cooling equipment. Therefore, AI is currently potentially compressing demand for white-collar jobs in information technology and finance while simultaneously creating jobs in traditional blue-collar and industrial sectors. Bank of America found that labor demand in the information technology and financial insurance industries declined by about 1.9% and 1.1% respectively in the first few months of this year, but a general relationship has not yet emerged where higher AI exposure leads to a faster overall job decline.
Note: Bank of America's estimates are based on a recombination of the "AI and data center" related manufacturing industries; non-residential construction growth also includes non-AI projects such as factories and energy. Therefore, strictly speaking, it's not that "AI created these 127,000 jobs," but rather that investment in AI infrastructure is becoming a significant source of new employment in the US goods-producing sector, partially offsetting the cooling of white-collar employment. This also explains why, despite the surge in AI spending, the US has not experienced the total "job doomsday" that the market had previously feared.
