China’s Dark Factories: So Automated, They Don’t Need Lights.

Walk into a traditional automotive factory and you expect noise, bright lights, hundreds of workers, forklifts moving material, supervisors walking the floor, and production lines stretching across enormous buildings. Now imagine walking into parts of a factory where the lights are dim, hundreds of robots are moving almost continuously, and only a handful of people are visible.

That is the idea behind China’s so-called dark factories, sometimes called lights-out manufacturing. The name sounds futuristic, but the concept is surprisingly practical. When robots perform most of the work in an area, the factory no longer has to be designed entirely around human needs. Machines do not need bright lighting to weld a vehicle body, take a coffee break between shifts, or go home at the end of the day.

The Wall Street Journal video on this page visits Chinese electric-vehicle manufacturer Zeekr to show what this transformation looks like in the real world. The factory is not completely empty of people, and the entire facility does not literally operate in darkness. Instead, highly automated areas can run with extremely limited human involvement—so little, in fact, that some areas could theoretically operate with the lights turned off.

That distinction matters because the real story is not about turning off the lights. It is about how far automation can change the economics, speed, quality, and competitive power of manufacturing.

What Is a Dark Factory?

A dark factory is a manufacturing operation, or often a highly automated section of one, where machines perform enough of the production process that very few workers need to be physically present. Robots may weld, move components, position parts, inspect products, transfer material, and coordinate production while sensors, cameras, and software continuously monitor what is happening.

The phrase “dark factory” comes from the simple idea that machines do not require the same working environment humans do. A robotic welding cell does not need bright overhead lighting so the robot can see. It relies on programming, machine vision, sensors, and control systems instead of human eyesight. The better way to think about dark manufacturing is therefore not as “a factory with no people,” but as a factory increasingly designed around machines first, with people involved where human judgment, flexibility, maintenance, engineering, or problem-solving still adds the most value.

That represents a much bigger transformation than simply replacing workers with robots. The real shift is toward factories that are designed from the ground up to operate with higher levels of automation, better information flow, and tighter coordination between machines, software, and people.

Inside Zeekr: Hundreds of Robots Working Together

The video takes viewers inside Zeekr, a Chinese electric-vehicle manufacturer that demonstrates just how automated a modern automotive factory can become. Hundreds of robots work throughout the plant, performing repetitive and highly precise tasks such as welding and material handling, while the surrounding production system coordinates the flow of parts and information.

What makes the operation particularly interesting is not simply the number of robots. It is the way automation has been incorporated into the production system itself. In an older factory, automation is often added to processes that were originally designed for people. In newer facilities, the factory can be designed from the beginning around automated production, which changes how machines are positioned, how material moves, and how production is scheduled.

This affects more than labor requirements. Production information can move directly between machines and software systems, material movement can become increasingly automated, quality data can be collected continuously, and production decisions can be based on real-time information rather than waiting for someone to identify and report a problem. The factory begins to behave less like a collection of separate operations and more like one connected manufacturing system.

Why China Is Automating So Aggressively

China built much of its manufacturing strength over the past several decades with a combination of enormous industrial scale, growing infrastructure, strong supplier networks, and relatively low labor costs. But the economics of Chinese manufacturing have been changing as wages have risen, products have become more complex, and manufacturers have faced increasing pressure to improve productivity, quality, and speed.

Automation provides one way to continue expanding production capability without increasing labor at the same rate. It can also help companies improve consistency, reduce dependence on repetitive manual work, and create more predictable production environments. That makes robotics especially attractive in industries such as automotive manufacturing, where thousands of repetitive processes must be performed with high precision.

China has also made advanced manufacturing and robotics an important part of its broader industrial strategy. Investments in electric vehicles, batteries, robotics, automation, electronics, and manufacturing technology are helping create an industrial ecosystem where one capability strengthens another. Dark factories should therefore not be viewed as isolated experiments; they are part of a much larger shift toward highly automated manufacturing.

The competitive advantage does not come simply from installing a robotic arm. It comes from building manufacturing capability at scale and connecting automation with engineering, suppliers, software, data, logistics, and production planning.

The Real Advantage Is Not Turning Off the Lights

The name “dark factory” attracts attention, but electricity savings from lower lighting requirements are hardly the main benefit. The real economic opportunity comes from changing how production operates and reducing the number of processes that depend on direct human participation.

Traditional manufacturing is organized heavily around people. Workers arrive for shifts, take breaks, hand work from one team to another, and leave at the end of the day. Staffing levels change, employees need training, and human performance naturally varies. Machines operate differently because a properly designed automated system can perform the same task repeatedly with very little variation while collecting data continuously.

That does not mean robots run forever without problems. Equipment breaks, sensors fail, tooling wears out, software has errors, and maintenance is still required. However, automation reduces the number of processes that have to stop simply because a human shift has ended. That can increase equipment utilization, throughput, and consistency while potentially lowering the labor required for each unit produced.

Quality May Matter More Than Labor Savings

It would be easy to look at China’s dark factories and assume that their main purpose is eliminating labor cost, but that misses one of the biggest advantages. Automation can also reduce variation, which is especially important in industries where thousands of operations have to happen repeatedly and accurately.

Automotive welding is a good example. If hundreds or thousands of welds must appear in precise locations on every vehicle, robots can perform those movements with remarkable consistency. That level of repeatability can reduce defects, lower rework, improve quality, and make production performance more predictable.

This means the automation business case may include far more than the wages of the workers being replaced. Improved consistency can reduce inspection, scrap, delays, warranty claims, and customer problems. The better question is not simply, “How much labor does this robot eliminate?” but rather, “How does this technology improve the total performance of the process?”

The best automation investments often improve several things at once, including quality, throughput, safety, consistency, data visibility, and cost. That is where automation creates much more value than a simple headcount reduction.

Dark Does Not Mean Human-Free

One of the most important lessons from the video is that highly automated manufacturing does not mean people have disappeared. Some tasks remain extremely difficult to automate, especially work involving flexible materials, unusual shapes, variable conditions, and complicated assemblies.

Automobiles contain wiring, cables, trim, connectors, and countless components that do not always behave in perfectly predictable ways. Humans remain remarkably good at manipulating unfamiliar objects, adapting to changing conditions, identifying unusual problems, and making judgment calls when something does not fit the expected pattern.

People are also required to keep the automation itself operating. Someone must maintain robots, repair equipment, investigate quality problems, program systems, improve processes, analyze production data, and respond when reality does not match what the automated system expected.

The transformation therefore may not be as simple as humans being replaced by robots. It may be a shift from humans performing large amounts of repetitive production work toward humans designing, maintaining, improving, and supervising increasingly automated production systems. That changes what manufacturing work looks like and changes the skills manufacturers need.

Human Work Moves Up the Value Chain

Imagine two factories producing similar products. In the first factory, hundreds of workers manually perform repetitive tasks. In the second, robots handle many of those processes while a smaller group of technicians, engineers, maintenance specialists, programmers, production planners, and quality experts keep the system operating.

The second factory may employ fewer people directly on the production line, but the people who remain may need significantly different skills. They must understand equipment, troubleshoot problems, work with software and data, and understand how automated processes connect to one another.

This creates one of the biggest workforce challenges associated with advanced manufacturing. Companies may be able to purchase robots faster than they can develop enough people who understand how to support, maintain, and improve them. The future manufacturing worker may spend less time physically making the product and more time ensuring the system making the product performs correctly, which makes training and workforce development part of the automation strategy.

The Factory Is Becoming a Connected System

The biggest opportunity in manufacturing automation is not simply putting robots everywhere. It is connecting the robots to everything else so production, quality, maintenance, and material movement can respond to one another in real time.

Imagine a production line where a machine-vision system detects a quality issue. Instead of allowing hundreds of defective parts to continue down the line, the system identifies the abnormality quickly, determines which station may be responsible, isolates potentially affected products, alerts maintenance, and updates production information.

That is much more powerful than automating one task because it creates a connected response. Traditional factories often contain delays between when a problem occurs and when someone realizes what happened. An operator sees something unusual, a supervisor investigates, quality becomes involved, maintenance is called, and eventually the production schedule is adjusted.

In a highly connected factory, some of that information can move almost instantly. The real breakthrough therefore may not simply be automation of physical work, but the combination of automation and information moving together.

AI Could Make the Dark Factory Even Smarter

Artificial intelligence adds another layer to this transformation. Traditional automation generally follows predefined rules, which works extremely well for repetitive and predictable processes. AI can potentially help automated systems deal with more complexity by identifying patterns, detecting anomalies, and analyzing information that would be difficult for people to process quickly.

Machine-learning systems may identify unusual patterns in equipment behavior before a breakdown occurs. Computer vision can inspect products at high speed, while AI can analyze production variables and help engineers understand why a process is drifting.

Over time, manufacturing systems could become increasingly capable of sensing what is happening, identifying risk, recommending adjustments, and eventually making some changes automatically. Automation performs the task, sensors show what is happening, data creates visibility, and AI helps determine what should happen next.

When those capabilities begin working together, the factory becomes much more than a group of robots. It becomes a system capable of continuously responding to changing conditions.

A Faster Factory Requires a Faster Supply Chain

There is an overlooked challenge with extremely automated factories: you still have to feed them. An automotive plant capable of producing hundreds of thousands of vehicles requires an enormous and reliable flow of batteries, electronics, tires, glass, steel, seats, motors, wiring, fasteners, packaging, and thousands of other components.

If the factory becomes faster but the supply chain does not, automation simply allows the factory to reach the shortage sooner. That means highly automated manufacturing increases the importance of supplier reliability, inventory accuracy, transportation performance, planning, and visibility.

Suppliers need to deliver the right material at the right time, quality problems must be caught early, transportation must be synchronized with production, and inventory records need to be accurate because automated systems rely heavily on the information they are given.

This creates a fundamental manufacturing truth: you cannot build a world-class automated factory on top of an unreliable supply chain. As manufacturing becomes faster, the connections surrounding manufacturing become even more important.

Automation Exposes Weak Processes

There is another major mistake companies can make when pursuing automation: they automate a bad process. Suppose a component moves through six unnecessary steps before reaching final assembly. A company could install sophisticated automation to move that component through all six steps faster, but the result would still be an expensive automated version of a poorly designed process.

That is why Lean manufacturing and first-principles thinking still matter in a highly automated world. Before installing technology, leaders should ask why the process exists, whether the product needs to move through every step, why an inspection is required, whether the design can be simplified, and whether unnecessary variation can be removed.

Technology works best when it strengthens a good operating system. It should not become a substitute for thinking, because automation can make a strong process better, but it can also make a bad process fail faster and at greater scale.

China’s Advantage Is Bigger Than Cheap Labor

For years, companies often thought about Chinese manufacturing primarily through the lens of labor cost. That view is becoming increasingly outdated because modern Chinese manufacturing capability includes extensive supplier ecosystems, electric-vehicle production, battery manufacturing, industrial robotics, electronics, automation, engineering talent, infrastructure, and enormous manufacturing scale.

Dark factories add another element to that system. Companies competing with highly automated Chinese manufacturers may not be competing against one isolated cost advantage. They may be competing against a network of advantages that reinforce one another, including faster product development, strong supplier networks, advanced automation, engineering capability, battery technology, software, and large-scale production.

This changes the strategic question. It is no longer simply, “Where can we find the cheapest labor?” The better question becomes, “Where can we build the strongest overall manufacturing system?”

That is a much more sophisticated decision because it requires leaders to think about the entire network rather than one cost variable.

Producing More Is Not the Same as Winning

There is another side to the dark-factory story. Automation dramatically increases the ability to produce, but it does not automatically increase the number of customers.

A factory could become one of the most efficient manufacturing operations in the world and still produce too much of something the market does not need. This matters especially in industries such as electric vehicles, where manufacturers have invested enormous amounts of capital into capacity while competition continues intensifying.

An automated factory may produce faster, but demand planning still matters. Sales and operations planning still matters, inventory still matters, and market strategy still matters because the robot does not know whether the customer actually wants the product unless the broader business system provides that information.

This is one of the most important systems lessons in supply chain. Optimizing one part of the system does not automatically optimize the whole system, and maximum factory output is not the same thing as maximum business performance.

Automation Changes the Economics of Manufacturing

Heavy automation also changes a company’s cost structure. A labor-intensive factory carries significant labor expense but may have more ability to adjust staffing as demand changes. An automated facility requires substantial upfront investment in equipment, robotics, integration, software, and engineering.

That pushes more cost into the fixed-cost structure. When volume is high, the economics can be excellent because those fixed investments are spread across large production quantities. When utilization falls, the same investment can become a burden.

This means highly automated manufacturers need to understand demand and capacity extremely well. The more productive the equipment becomes, the more important it is to make sure the equipment is producing something the market actually needs.

Automation therefore does not eliminate capacity planning. In many ways, it makes capacity planning even more important.

What Supply Chain and Manufacturing Leaders Should Learn

The obvious lesson from China’s dark factories is that robotics and automation will continue changing manufacturing. The more useful lesson is that companies should not chase automation simply because competitors are installing robots.

Leaders should begin by understanding what is limiting performance. Is the problem labor availability, quality, throughput, safety, capacity, lead time, cost, or reliability? Once the problem is clear, it becomes much easier to determine whether automation is the right solution.

A few principles are especially important:

Simplify before automating. Removing unnecessary work is usually better than automating unnecessary work.
Look beyond labor savings. Quality, safety, throughput, uptime, flexibility, and customer performance also matter.
Strengthen the supply chain around the factory. Faster production demands better suppliers, logistics, planning, and data.
Invest in people as well as machines. Automated factories still need engineers, technicians, maintenance experts, planners, and problem-solvers.
Keep capacity connected to demand. Producing more creates value only when customers need what is being produced.

The objective should not be to build a factory with the fewest possible people. It should be to create a manufacturing system that delivers the best combination of cost, quality, speed, flexibility, resilience, and customer value.

The Factory of the Future May Never Really Stop

The image of an enormous factory quietly operating in darkness is powerful, but the deeper idea is continuous operation. Machines can communicate with machines, production systems can collect data around the clock, AI can analyze maintenance and quality signals, autonomous material-handling systems can move inventory, and robots can perform highly repetitive production tasks continuously.

Eventually, the factory may behave less like a building filled with individual machines and more like one connected system that senses what is happening and responds. Humans do not disappear from that system; their role changes.

People define objectives, design processes, improve the factory, maintain equipment, manage exceptions, make strategic decisions, and decide what should be automated in the first place. The machines execute more of the work, while humans remain responsible for deciding whether the work makes sense and whether the system is producing the right result.

Final Thought: The Important Part Isn’t That the Lights Are Off

China’s dark factories make great headlines because the image is so memorable: enormous manufacturing facilities filled with robots operating in areas where the lights could theoretically be switched off. But focusing on the darkness misses the real transformation.

The important change is that factories are becoming more automated, more connected, more data-driven, and less dependent on humans performing every production step themselves. The Zeekr factory shown in the video gives us a glimpse of that transition, with robots performing enormous amounts of repetitive production work while people remain involved where flexibility, maintenance, judgment, and problem-solving are still required.

For manufacturers elsewhere, the lesson should not simply be, “We need more robots.” The better question is whether the entire operating system is ready for automation. Are the processes simple enough? Are suppliers reliable enough? Is the data accurate enough? Are employees trained? Is demand strong enough? Can maintenance support the technology? Does the automation actually improve the customer outcome?

Those questions matter far more than whether the factory lights are on or off. The factory of the future will not win because it has the most robots. It will win because people, automation, AI, suppliers, data, manufacturing, and logistics work together as one connected system that creates a better business result.

Want to stay ahead in the supply chain game? Subscribe to our newsletter for the latest trends, insights, and strategies to optimize your supply chain operations.

Comments about Dark Factories

“The dark factory doesn’t dream of electric sheep—it builds them, one every 76 seconds, without ever turning on the lights.”
“In a lights-out plant, the only heartbeat you hear is the rhythmic thud of robotic arms; humans became optional somewhere around 2024.”
“Dark factories are the ultimate introvert’s workplace: zero small talk, zero coffee breaks, zero daylight.”
“We used to say ‘the factory never sleeps.’ Now we say ‘the factory never wakes up,’ because no one inside has eyes.”
“Energy savings from switching off the lights are nice, but the real revolution is switching off the payroll.”
“The dark factory is honest capitalism: it finally admits that most manufacturing jobs were only kept for the warmth of human bodies, not the skill of human minds.”
“When the last worker leaves and the lights go out forever, that’s not the end of industry—it’s the beginning of its purest form.”
“In the dark factory, quality control is done by machines that never blink, never get bored, and never call in sick on the day of the audit.”
“A dark factory running at 3 a.m. looks exactly the same as one running at 3 p.m.—and that sameness is the whole point.”
“We feared robots would take our jobs. Turns out the scarier future is one where they don’t even need us to flip the switch.”