Anthropic released a research report on September 30, 2026, revealing that while robots are technically capable of performing 74% of physical tasks in the U.S. economy, the high cost of hardware remains a significant barrier to widespread adoption. This technical capability translates to roughly 34% of all working hours currently logged by human employees.
The report, titled “What work can robots do?”, indicates that the gap between what robots can do and what is economically viable to automate is substantial. Based on a historical 3% annual decline in robot prices since the 1990s, Anthropic researchers estimate it would take 40 years for robots to become cost-competitive for just 10% of human work.
Packers and packagers are among the few roles where robots are currently cost-competitive.
The Cost Gap: Packers vs. Taxi Drivers
The economic feasibility of automation varies wildly across industries, depending on the complexity of the task and the current cost of human labor. According to the Anthropic study, packers and packagers represent the largest occupation where robots are already cost-competitive. In these roles, a robot currently costs roughly $2,500 less per year than a human worker.
In contrast, other high-exposure roles remain more expensive to automate. Taxi drivers face a 97% technological exposure rate—meaning robots can technically perform almost all of their duties—yet robotaxis currently cost about $7,000 more per year to operate than employing a human driver. This figure does not include the significant regulatory hurdles that autonomous vehicles still face.
Environmental Complexity and “Safe” Jobs
Anthropic categorized workplaces into a tier system (E1 to E3) based on the level of structure and predictability in the environment. Robots perform best in E1 environments, which are highly structured and repetitive, such as assembly lines.
The report identifies nursing and maintenance or repair work as the least likely to be automated in the near term. These roles often exist in E3 environments—unstructured physical spaces where robots struggle with navigation, unpredictable obstacles, and complex human interactions.
Robots struggle with unstructured environments like nursing and maintenance work.
When physical robotics are combined with Large Language Models (LLMs), the total work exposure across the entire economy jumps to 81%, as the AI handles cognitive tasks while the hardware manages physical output.
While the technical potential for automation is high, the current “physical labor layer” of the economy remains largely insulated by the high cost of manufacturing and maintaining sophisticated hardware. For most industries, the transition to a robotic workforce is a matter of decades rather than years, dictated by the slow decline of hardware prices rather than the rapid advancement of software.
The post Robots Capable of 34% of Work Hours But Price Barriers Limit Adoption appeared first on World Weekly News.


Bengali (Bangladesh) ·
English (United States) ·