🏭 Still Pretending Manufacturing = Jobs
We actually manufacture a lot in the U.S. - it just doesn't nee a lot of workers...
To be clear: I am not anti-manufacturing. Far from it — I think building stuff is cool, and I think we should build more (and consume less). In another life I would be one of those reckless photographers who explores massive abandoned steel mills.
But the discourse around manufacturing feels genuinely insane (sometimes). The U.S. manufactures a ton of stuff: about $2.4 trillion worth in 2017 dollars, and up about $600 billion from 2005 (Canada’s economy was about $2.3 trillion in 2023, for reference). The U.S. also employs about 12.7M people in manufacturing, and has lost about 1M manufacturing jobs since 2005. So, production is up but jobs are down.
Manufacturing is largely a story of efficiency, and efficiency is a sometimes a double-edged sword. A more efficient factory produces more but has fewer workers (it also pollutes less!). In the 1950s, it took over 50 jobs to produce $1 million in manufactured goods. Fast forward to the late 1970s and that number declined to around 26 per $1M. Today it takes only 5 employees to produce $1M in manufactured goods.
All this is to say: manufacturing is a good thing, but it’s also highly labor-efficient —especially in advanced economies like the U.S., Germany, and Japan. For U.S. manufacturing to grow significantly, it would likely need to become even more efficient. By comparison, China’s vast manufacturing sector employs nearly 10 times as many workers as the U.S. does. Yet, while the average American manufacturing worker produces about $150,000 in output per year, the average Chinese worker produces only around $16,000. Looking ahead, the next major leap in efficiency will almost certainly come from advanced artificial intelligence (AI) combined with increasingly sophisticated robotics.
1920s–30s: electrification → less physical strain → flexible factory layout → faster throughput
1940s–50s: basic automation → fewer manual assemblers, less waste → more output
1960s–80s: computer numerical controls (CNC) + industrial robots → less need for skilled machining → precision, speed, safety
1990s–00s: lean + digital tools → fewer line workers → minimized waste, fast turnaround
2010s–20s: smart factories, machine learning → continuous optimization → near autonomy
2020s-???: AI → full autonomy