Industrial Automation
The Factory That Runs in the Dark
Some of the most advanced manufacturing facilities on Earth have no lights — because nobody is there to need them.
The Idea
When engineers talk about 'lights-out manufacturing,' they mean it literally: factories where automated systems run continuously in complete darkness, with no human workers present on the floor. The concept sounds futuristic, but it has been quietly spreading through precision manufacturing, electronics assembly, and automotive production for years. What makes it possible isn't any single breakthrough — it's the convergence of several mature technologies arriving at the same moment: robotic arms with sufficient dexterity, machine vision that outperforms human inspection, sensors that detect microscopic defects, and AI systems that can adjust production parameters in real time without a human in the loop. The deeper idea here isn't efficiency — it's a fundamental shift in what a factory *is*. Traditional industrial automation replaced human muscle: machines that stamped, welded, or pressed faster than any person could. The current wave replaces human judgment. Modern systems don't just execute a fixed sequence of movements; they perceive, evaluate, and adapt. A robotic cell can detect that a component is slightly out of tolerance and compensate mid-assembly. This is a qualitatively different kind of automation, and it's why the economic logic is so disruptive. The marginal cost of running a lights-out factory overnight approaches zero — no shift allowances, no fatigue, no errors from the 3am slump. The question industrial economies now face isn't whether this spreads, but how fast.
In the World
FANUC, the Japanese robotics company, runs one of the most cited examples: a factory in Oshino, at the foot of Mount Fuji, where robots manufacture other robots — largely unsupervised, for around thirty days at a stretch. The facility produces roughly fifty robotic arms per twenty-four-hour cycle. When something goes wrong, the system logs the fault and pauses; a human reviews it later. The robots don't need climate control calibrated for human comfort, breaks, or safety lighting. The building sits quietly in the dark, humming. FANUC isn't alone. In Dongguan, China, a factory run by Changying Precision Technology replaced 90 percent of its human workforce with automated systems and reported a defect rate that dropped by over 80 percent while output roughly tripled. In the United States, BMW's Spartanburg plant in South Carolina uses thousands of robotic systems that handle everything from body welding to quality inspection — with human workers now concentrated in roles that require flexibility and judgment rather than repetitive precision. What's striking about these cases isn't the robot count — it's the economic signal they send to every other manufacturer. Once a competitor runs lights-out, the pressure to follow becomes intense. You can't compete on labor cost against a facility that has none. This dynamic is why industrial automation tends to spread in waves: one early adopter forces the hand of everyone else in the same market.
Why It Matters
For most of us, a lights-out factory feels remote — something happening in a supply chain we never see. But the reverberations are anything but distant. The goods these facilities produce are already on your shelves, in your devices, in the components inside your car. And the economic logic they embody is migrating: the same convergence of sensors, vision systems, and adaptive software is beginning to appear in warehouses, logistics hubs, and — more slowly — in service environments. The more useful question to sit with isn't 'will automation take my job?' That framing is too blunt and too personal. The sharper question is about what kinds of human judgment are genuinely hard to automate — and why. The answer tends to cluster around novelty, context, and social intelligence: handling situations the system hasn't seen before, understanding what a customer actually needs rather than what they literally asked for, navigating ambiguity. Knowing where that line sits, and how it's moving, is one of the most practically valuable things anyone can understand right now.
A Question to Ponder
If the cost of producing something approaches zero because human labor is removed from the loop entirely, who actually benefits from that — and what would need to change for the answer to be 'most people'?
Get a new one of these every morning.
Start learning with Thinkable