Imagine a factory that keeps producing after every employee has gone home.
Robotic arms continue assembling components. Autonomous vehicles carry materials across the plant. Cameras inspect products for microscopic defects, software adjusts production schedules in real time, and machines predict when they will require maintenance.
The lights could technically be switched off.
This is the idea behind the fully automated factory, sometimes called a lights-out factory or dark factory.
Once largely confined to science fiction, lights-out manufacturing is becoming increasingly realistic thanks to advances in industrial robotics, artificial intelligence, computer vision, digital twins and autonomous logistics.
But there is an important distinction: very few factories today operate completely without humans. Most of the world’s most advanced manufacturing plants are better described as highly automated, with specific production lines or cells capable of running autonomously for long periods.
So what actually makes a factory fully automated?
What is a fully automated factory?
A fully automated factory is a manufacturing facility in which machines, software and robotic systems can perform most or all production processes with little continuous human intervention.
In the most extreme version — true lights-out manufacturing — workers do not need to be physically present while normal production is running.
Siemens describes a lights-out factory as a facility capable of operating with minimal human activity and potentially zero human intervention on site during autonomous production. In reality, the company notes that lights-out manufacturing cells are currently much more common than completely autonomous plants.
Humans have not necessarily disappeared from the system.
Instead, their role changes.
Engineers may configure production lines, maintain robots, design products, analyze performance data and intervene when something unexpected happens.
The machines handle the repetitive production.
Why are they called lights-out factories?
The name is surprisingly literal.
Traditional factories require lighting, heating, ventilation and other environmental conditions partly because people work inside them.
Robots do not need the same working environment.
If a production area is completely autonomous, there may theoretically be no reason to keep the lights on during normal operations.
That is why terms such as dark factory, lights-out factory and lights-out manufacturing became associated with extreme industrial automation.
The concept should not be confused with the broader term smart factory.
A smart factory can contain thousands of workers while using AI, sensors and automation extensively. A lights-out factory represents a much higher level of autonomy.
The technologies behind fully automated factories
There is no single technology capable of creating a fully automated factory.
Autonomous manufacturing emerges from several systems working together.
Industrial robots
Robotic arms remain the most visible component.
They can weld, paint, assemble, package, move components and perform tasks requiring extreme precision.
Modern industrial robots can also change tools automatically and interact with other machines without requiring a worker to supervise every movement.
Autonomous mobile robots
Factories also need to move materials.
Autonomous mobile robots, commonly known as AMRs, can transport components between production areas without following fixed tracks.
Instead, sensors and mapping systems allow them to navigate dynamically around the factory.
Artificial intelligence
AI gives automation something it historically lacked: adaptability.
Traditional automation follows predefined rules.
AI systems can instead analyze production data, detect anomalies, optimize scheduling and identify patterns humans might miss.
This becomes particularly important when thousands of machines operate simultaneously.
Computer vision
AI-powered cameras perform quality inspection at speeds impossible for humans.
Computer vision systems can detect scratches, dimensional errors, incorrect components or assembly defects while a product is moving through the production line.
Industrial Internet of Things
Thousands of sensors constantly measure temperature, vibration, pressure, energy consumption and equipment status.
Connecting this information through the Industrial Internet of Things gives the factory a real-time digital picture of what is happening.
Predictive maintenance
A completely automated factory cannot rely on someone noticing that a machine “sounds strange.”
Sensors and algorithms must detect potential failures before they happen.
Predictive maintenance systems analyze equipment behaviour and can schedule intervention before a component breaks.
Digital twins
A digital twin is a virtual representation of a machine, production line or entire factory.
Manufacturers can simulate changes before implementing them physically.
Production speed, machine layout, energy consumption and potential bottlenecks can all be tested digitally first.
Are fully automated factories already real?
Yes — but the answer requires some nuance.
One of the best-known examples comes from FANUC, the Japanese industrial robotics manufacturer.
FANUC says its Japanese factories use extensive robotization and that some machining operations can continue without operators for long periods, including nights and weekends. Its robots are even used to manufacture other robots.
Foxconn has pushed the concept even further in some facilities.
Its Shenzhen operations have demonstrated lights-off manufacturing with automated quality control, predictive maintenance, robots and real-time production monitoring.
But one of the most interesting cases is Xiaomi.
Its new smartphone factory in Beijing has often been described online as a “dark factory.” Yet reports from inside the facility show something more nuanced: the plant is extraordinarily automated and AI-driven, but not literally an empty building permanently producing phones in darkness.
That distinction matters.
Automation is a spectrum, not a switch.
Why companies want automated factories
The obvious advantage is productivity.
Robots do not need traditional shifts and can potentially operate continuously.
But labour reduction is only one part of the equation.
Automation can also deliver:
- more consistent product quality;
- lower defect rates;
- higher production speeds;
- improved worker safety;
- better traceability;
- lower energy consumption;
- faster production changes;
- more predictable manufacturing costs.
AI can further optimize these systems continuously.
The World Economic Forum’s Global Lighthouse Network — which tracks some of the world’s most technologically advanced industrial sites — has documented major productivity improvements from AI, analytics and digital manufacturing technologies.
Why aren’t all factories fully automated?
If automation is so efficient, it raises an obvious question.
Why do factories still employ millions of people?
Because humans remain extraordinarily flexible machines.
A person can pick up an unexpected object, recognize that something looks wrong and adapt to a new task almost instantly.
Industrial automation is much easier when a factory produces millions of identical objects.
It becomes considerably harder when products change frequently or require complex manual manipulation.
Automation also requires substantial capital investment.
Robots, sensors, machine-vision systems, software and integration can cost millions — or billions — before the first product leaves the line.
That means the economic case depends heavily on production volume.
Will factory workers disappear?
Probably not.
But many factory jobs will change.
Repetitive physical tasks are among the easiest activities to automate.
At the same time, increasingly automated factories require technicians, software developers, robotics engineers, cybersecurity specialists, data scientists and maintenance experts.
Interestingly, the World Economic Forum’s most advanced manufacturing sites continue to emphasize workforce training alongside automation rather than simply eliminating workers.
The likely future is therefore not a world where every factory is empty.
It is a world where fewer people directly manufacture products and more people manage the systems that manufacture them.
The factory of the future may be almost invisible
For more than a century, industrial progress meant building faster machines.
The next step is different.
Factories are becoming systems capable of observing themselves, predicting problems and making decisions.
Robots build products. AI checks them. Sensors monitor the robots. Software optimizes the process, while digital twins simulate what should happen next.
The completely automated factory remains rare.
But its individual building blocks are already operating across the world.
And as artificial intelligence moves from computers into industrial machines, the distance between a smart factory and a true lights-out factory is getting smaller.
