A six-step plan built around Shenzhen’s core advantages
A translated article published by TechFlowPost highlights a viral essay by US industrial entrepreneur Zane Hengsperger, who argues that Shenzhen’s manufacturing strength comes from three things: density, speed, and a culture that treats making physical goods as high-status work. His central claim is that none of those advantages are unique to China, and that the US failed to copy them over the past three decades because it chose outsourcing instead.
He writes that Shenzhen was still a fishing village in 1980, but today exports more electronics than the entire United States. In Shenzhen, a team can move from a CAD file to an injection-molded prototype in 48 hours. In the US, he says, the same process takes six weeks and requires coordination across four suppliers. On that basis, he lays out a six-step roadmap and says the sequence matters.
Step one: fix culture first
Hengsperger calls culture the bottleneck beneath every other bottleneck. Factories and machines can be bought, he argues, but a generation that wants to work in manufacturing cannot.
He cites several figures to make that case: only 6% of US high school students see manufacturing as a career path; for every five skilled tradespeople who retire, only two new workers enter the field; and the average welder is 55 years old. He contrasts that with Shenzhen, where a 26-year-old running a CNC shop can be the person others in the same social circle want to become.
He also says the US spent decades telling smart students to leave the shop floor for the office. The result, in his telling, is 570,000 unfilled manufacturing jobs and $1 trillion in output that could go unrealized by 2030. The fix, he says, is status rather than subsidies: make machinists visible again, connect trade schools to real projects and jobs before graduation, and restore practical shop programs in high schools.
Step two: build far more factories
The article describes Shenzhen’s superpower as density. Within a 50-mile radius, PCB plants, injection molding shops, CNC facilities, and assembly lines sit close together, letting iteration happen in hours rather than weeks.
The US, by contrast, has an industrial base that is scattered, aging, and shrinking, according to the piece. Hengsperger says four of the country’s six primary aluminum smelters are partially or fully idled. He puts US steel output at 79 million tons a year, versus 1.005 billion tons in China, and says base metals often come with lead times of eight to 30 weeks.
His answer is not a handful of flagship factories. It is volume. He calls for hundreds of new steel mills, foundries, machine shops, sheet metal shops, and assembly plants, intentionally clustered in places where industrial infrastructure already exists, including Detroit, Houston, Phoenix, and the Carolinas. The goal is to keep the trip from one production step to the next within a day.
Step three: turn energy into a weapon
One of the article’s main arguments is that AI-driven power demand is changing industrial economics. US electricity demand grew only 0.1% a year on average from 2005 to 2019, he writes. That has changed. By 2027, US data center demand is projected to reach 66 GW, up from 31 GW in 2025. Data centers are expected to account for 8.5% of summer peak demand by 2027, versus 4% two years earlier.
He adds that data centers already consume more than a quarter of Virginia’s electricity, while Texas grid load is rising 10% a year. In that environment, steel mills and aluminum smelters are bidding for the same electricity as hyperscale cloud companies with trillion-dollar balance sheets. Under existing power market structures, he argues, heavy industry loses. He points to a smelter closure in Missouri while a server campus was being built a few kilometers away.
His proposal is to make heavy industry part of the grid itself. Mills should have on-site generation, using natural gas now and shifting later to next-generation nuclear and geothermal if those options become available. They should also act as flexible loads, running hard when prices are low and releasing as much as 100 MW back to the grid when data center demand spikes. He also wants steel mills and data centers co-located around shared generation and shared interconnection points because grid connection queues now stretch into years.
Step four: build a domestic supply chain that can deliver everything
Hengsperger argues that Shenzhen’s real product is not cheap labor. It is proximity. If a team needs a gearbox, a coating, a mold, or a custom bracket at 4 p.m., someone nearby can make it and deliver by the next morning. That is what makes the 48-hour iteration loop possible.
His US version does not require one-hour delivery ranges, but it does require keeping procurement inside national borders. Factories should be able to source what they need domestically within days, from ore and alloys to castings, coatings, fasteners, and finished assemblies.
He says the US is far from that standard today. The article states that China refines about 90% of the world’s rare earths, while each F-35 contains 920 pounds of rare earth materials. The US also depends on imports for machine tools and aluminum, leaving critical choke points outside its control.
His remedy is to map every critical material and manufacturing layer, then close the gaps company by company, including rare earth refining, forgings, castings, specialty alloys, and machine tools. On top of that, he wants a software layer that lets a shop in Ohio order “6061 aluminum plate, anodized, 500 units, Thursday” and get it from 300 miles away rather than from a container ship.
Step five: use software and AI to strengthen capacity, not replace it
The article says the US still holds a real advantage in software, AI, and robotics talent, but that advantage only matters if capacity comes first. China, in Hengsperger’s framing, built capacity first and then layered on technology. The US tried to skip straight to the technology layer and lost that round.
He notes that China’s robot density overtook the US in 2023 and that China now operates about five times as many industrial robots. Even so, he says the US has the stronger software stack and has simply failed to aim it at factory floors.
He envisions a 2026 American factory as AI-native: quoting in hours rather than days, live scheduling instead of whiteboards, predictive maintenance instead of unplanned downtime that can cost $2 million an hour, and cobot cells handling repetitive work so certified welders can focus on tasks robots still cannot do. He also argues that every repair, parameter change, and manual override should be captured as data, especially when veteran TIG welders carrying decades of tacit knowledge may retire within four years.
Step six: government should focus on speed, demand, and capital
Hengsperger says the US government already has some bipartisan alignment on industrial policy, but still needs to play a large role. He points to Shenzhen’s designation as a special economic zone in 1980 and says the state cleared the path while the market did the rest.
His criticism of Washington is narrower and concrete: permitting can take years, environmental review can outlast product cycles, and Small Business Administration loan caps sit at $5 million even though a machine purchase may cost $8 million. In his view, the financing system is designed for real estate, not CNC cells.
He says Washington should do only three things. First, speed up industrial permits to 90 days instead of five years. Second, provide demand anchors through buyers such as the Pentagon, using large multi-year procurement. He points to recent ammunition expansion as proof this can work, and says the 2027 deadline to remove Chinese rare earths from US weapons systems is the right kind of forcing function. Third, expand loan guarantees for industrial equipment and facilities, including what he calls “Made in America loan guarantees.”
He rejects industrial planning that tries to pick winners or design factories from above. The state, he argues, should clear obstacles, anchor demand, back capital, and step aside.
His conclusion: Shenzhen is not magic
Hengsperger ends by saying Shenzhen took 45 years to build, but the US does not need 45 years and is not starting from a fishing village. It is starting with deep capital markets, strong software talent, and powerful demand signals tied to defense and reshoring.
His summary is simple: repair the culture around making things, build more factories, secure energy, fill supply chain gaps, embed software into plants, and keep government focused on speed, demand, and capital. In his view, Shenzhen’s edge comes down to density, speed, and status. All three, he says, can be copied.