As embodied AI becomes the next AI frontier, it forces a sobering reality check for employment in the American automotive industry. Reviving the auto industry and bringing back two million manufacturing jobs collide with the reality of stagnant demand for new vehicles, software-defined vehicles that are easier to manufacture, and adaptive robots that can accomplish many of the tasks involved in said manufacturing.
This post was co-authored with Joe White
Introduction
“They’ve got to have the whole thing built in America. That’s what we want,” President Trump said in May 2025, explaining his decision to impose import tariffs of twenty-five percent on imported vehicles and auto parts. The available manufacturing employment data tells a different story. The reasons: stagnant demand for new vehicles, and easier-to-manufacture vehicles due to new architectures. Factory automation, due to the introduction of AI technologies, particularly embodied AI technologies, such as adaptive robots, in the assembly line, will make the situation more dire.
The automotive industry remains stagnant in developed markets like the US and the EU. Since the end of the pandemic, demand in the US and the EU has not even reached the pre-pandemic levels, let alone the peak demand of 2016. High vehicle prices, high interest rates, stable population levels, and increasing comfort with shared mobility, including robotaxis, all point to ongoing suppressed demand.
The manufacturing employment numbers are disheartening. The U.S. lost more than 70,000 manufacturing jobs during 2025. The automotive sector shed 22,000 jobs. Hiring in manufacturing has been in the doldrums. By early 2026, U.S. manufacturing employment had declined by more than 200,000 jobs since 2023, reaching its lowest level since the pandemic ended. The U.S. manufacturing workforce, as a share of all jobs, has declined steadily since the late 1940s and now accounts for just 8% of total U.S. employment.
The new-generation vehicles, e.g., Tesla, Rivian, BMW Neue Klasse, are software-defined. They are based on simpler architectures that use fewer parts. Combined with advances in manufacturing techniques, such as gigacasting, these vehicles are easier to assemble.
Employment in today’s auto factory
In the early 1900s, Ford’s Rouge factory employed more than 100,000 workers. Today, Ford’s Rouge truck assembly plant employs about 3,700 people.
The end of the assembly line, where workers snake wiring through vehicle bodies and install seats and plastic trim, is the final frontier for human labor in auto factories. People still do these tasks more capably than robots. As the dexterity and skills of adaptive robots and the capabilities of other AI technologies increase, and with software-defined vehicles being easier to manufacture, the robots will also take over these tasks.
Employment in the next-generation auto factory
According to a WSJ article that cites research by Pascual Restrepo, in 1980, 66 percent of the value added in factories went to labor as wages and benefits. By the 2000s, that number was down to 45 percent.
Large populations of intelligent adaptive robots will characterize the next-generation factories. Companies like Tesla, Hyundai, BMW, and Mercedes are already testing such robots. China is aggressively working in the same area. An increasingly smaller number of humans will act as supervisors, orchestrators, or mission strategists of such robots.
To understand the automotive industry’s interest in, and even need for, such robots, we must assess the fit across four dimensions: industry growth, task complexity, task risk to humans, and industry worker demographics.
We already mentioned that the U.S. automotive industry’s growth remains stagnant. Unlike growth industries, where humans and robots will form collaborative teams to increase productivity and meet rising demand, mature industries, such as the automotive industry, will use adaptive robots to reduce manufacturing costs. The need to match the cost-efficiency of Chinese competitors will act as an additional forcing function.
New-generation vehicles are software-defined and employ simpler architectures. Combined with advances in manufacturing, such as gigacasting, these vehicles are easier to assemble. As a result, adaptive robots can handle more of the automotive manufacturing tasks. At the limit, we can foresee “dark assembly lines,” production spaces filled with robots and largely devoid of humans.
The U.S. auto industry’s workforce demographics show an aging population. The demographics of the new factories in the U.S. southern and southeastern states are better. However, there is no question that the industry has a demographic problem.
Conclusions
The 21st-century US auto industry bears little resemblance to President Trump’s vision. This is the result of decades of movement toward open trade, competition from foreign manufacturers, the relentless progress of automation, and now AI. The days when factories in Detroit and other Midwestern states relied mostly on manual labor to churn out nearly all of the vehicles sold in the nation are now more than 50 years in the past.
We must move beyond simply acknowledging that jobs will “fade away” and instead proactively prepare employees. The goal of 21st-century leadership must be to
Upskill workers;
Provide a viable transition for those whose roles are impacted by the progress of embodied AI.
Attract people with the appropriate backgrounds to work with the intelligent machines we are developing.
Today, the unfilled openings in the industry are for people with knowledge of advanced manufacturing systems.
Manufacturing is a backbone sector for any economy. To remain globally competitive, the U.S. will need to recapture its manufacturing leadership, including its leadership in automotive manufacturing. But this leadership needs to be built for the 21st century and beyond. And the 21st-century manufacturing is AI-intensive.
