Sign in Subscribe
Automation

Dark Factory vs. High Automation: What Lights-Out Manufacturing Verifies

Most plants called dark factories aren't. What lights-out manufacturing verifies at FANUC, Xiaomi, and Siemens, plus the economics.

John D

John D

— 16 min read
Share
Industrial robotic arm in a lab setting
Industrial robotic arm in a lab setting - Photo by Diego Martinez

Dark Factories: A Technical and Economic Briefing on Lights-Out Manufacturing

TL;DR

  • A genuine dark factory is a plant engineered to run production with no humans on the floor for extended, continuous periods, which permits removal of the lighting, comfort HVAC, welfare space, and much of the circulation built solely for human occupancy; verified cases of continuous fully lights-out operation remain rare and confined to narrow, highly standardized processes, while most facilities marketed as "dark" are in fact high automation with a small resident crew or lights-out only for limited unattended windows.
  • The enabling stack (industrial robots, machine vision, autonomous material handling, MES/SCADA orchestration, predictive maintenance, and digital twins) is mature enough to sustain unattended runs in favorable conditions, but the binding constraints are consistent: high-mix/low-volume changeovers, exception handling and error recovery, incoming material variability, and the maintenance and repair labor that must still be physically present.
  • The economics are attractive where volume is high and product geometry is stable (payback on unattended machining cells is commonly cited at roughly 18 to 36 months), but capital intensity, integration risk, and the persistent need for engineering and maintenance staff mean the realistic near-term trajectory is the "dim" factory, not the fully dark one.
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Dark factories briefing summaryA summary panel covering the three-tier classification scale for lights-out manufacturing, seven documented facilities classified against it, the enabling technology stack and its binding constraints, adoption metrics, economics, built-environment effects, and the near-term outlook. Dark factories: what lights-out manufacturing actually verifies Definition, evidence, adoption, economics, and built environment Classification scale Tier 1Continuous lights-outRare, narrow processes Tier 2Unattended windowsNights and weekends Tier 3High automation, crewNot a dark factory Documented facilities, classified FANUC OshinoRobots build robots unattendedTier 1 Foxconn ShenzhenLights-off lines, WEF lighthouseTier 2 Xiaomi Changping81% automated, about 220 staffTier 3 Philips Drachten120+ robots, about 9 QA staffTier 3 Siemens Amberg75% autonomous, human starts lineTier 3 Changying, Dongguan650 to 60 staff, asserted 2015Tier 3 Tesla, 2018Over-automation publicly reversedFailure mode Xiaomi and Changying figures are operator-asserted, not independently verified. Technology and constraints Enabling stack Robots and machine tending Machine vision inspection AMR and AGV material handling MES and SCADA orchestration Predictive maintenance Digital twins Private 5G networking Binding constraints High-mix changeovers Exception handling, error recovery Incoming material variability Maintenance labor on site Integration, not robot capability, is the limiting factor Adoption metrics 542,000installs in 2024 4,664,000operational stock 54%China share of installs 1,220Korea robots per 10,000 $20.4B to $41.7Bmachine vision by 2030 201WEF lighthouse sites Economics and built environment Economics Payback 18 to 36 months on cells Practitioner figures, not audited Robots $25k to $100k+ per unit Integration exceeds hardware cost Labor shifts to maintenance Built environment Sheds lighting, comfort HVAC, welfare and circulation space Sprinklers follow contents, not occupancy, under NFPA 13 Siting freed from labor pools Outlook Outlook: the dim factory, not the dark one Near-term path is the dim factory: extensive automation with lights-out cells and shifts, retaining a small crew for supervision, maintenance, and exceptions. Assumes no step-change in robotic dexterity, continued hardware cost decline, and persistent scarcity of skilled automation labor. DataDeep.Tech - Dark Factories

1. Definition and Scope

A dark factory, or lights-out plant, is a manufacturing facility engineered to operate production without humans present on the floor for extended periods. The defining test is not the level of automation but the absence of resident human occupancy during production: when no one is on the floor, the infrastructure that exists only for human occupancy can be reduced or eliminated. The term derives from the observation that machines do not require light to operate [3].

Press coverage routinely conflates three distinct conditions, and this briefing classifies each facility against that scale rather than repeating the "dark factory" label as marketing shorthand. The first condition is continuous fully lights-out operation, in which a plant runs for days or weeks with no human on the floor. The second is lights-out operation for limited unattended windows, the most common real-world form, in which cells run unmanned overnight or across weekends but are staffed and reset during the day [18]. The third is high automation with a small resident crew, in which humans remain continuously present in reduced numbers for supervision, quality assurance, and maintenance; this is not a dark factory even when automation is extensive [3]. Automated warehousing and dark fulfillment centers, frequently cited alongside these plants, are logistics operations rather than manufacturing and fall outside the scope of this briefing.

The boundary between genuine lights-out operation and conventional high automation therefore sits at continuity of unattended production and the removal of human-occupancy infrastructure. A plant that automates 75 percent of its value chain but retains a human first step, staffed inspection, and a resident maintenance crew is a highly automated conventional factory, not a dark one [5][6].

2. Technology Stack and Binding Constraints

Lights-out operation rests on the integration of several mature technologies. Industrial robots and automated machine tending perform assembly, welding, machining, and part handling. Machine vision and automated metrology close the quality loop by inspecting parts without a human eye. Autonomous mobile robots and automated guided vehicles move material between cells and to and from automated storage. Manufacturing execution systems and SCADA layers orchestrate scheduling, sequencing, and real-time monitoring. Predictive maintenance uses sensor data to anticipate equipment failure before it halts an unattended line, and digital twins allow processes to be simulated and optimized virtually before physical deployment [3]. High-speed industrial networking, such as increasingly private 5G, links these elements with low latency [9].

The individual technologies are not the limiting factor; their reliable integration under unattended conditions is. Four constraints bind in practice. First, high-mix/low-volume production is the hardest to automate to lights-out standard, because frequent changeovers require reprogramming, retooling, and revalidation that today generally need human judgment; fully dark operation is economically justified mainly for highly standardized, high-volume products. Second, exception handling and error recovery remain weak: a dropped part, a jammed feeder, or an out-of-tolerance drift that a human would resolve in seconds can halt an unattended line or, worse, go undetected until significant scrap accumulates. Third, incoming material variability defeats systems tuned for tight tolerances, so lights-out cells demand upstream consistency that is itself costly to guarantee. Fourth, maintenance and repair labor cannot be automated away; skilled technicians must be physically present, at least on call, and their scarcity is a recurring adoption barrier [3]. The consensus in the engineering and integrator literature is that these limitations, rather than robot capability, explain why truly dark factories remain exceptional [3].

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The lights-out technology stack and its binding constraintsA layered diagram of the technologies enabling lights-out manufacturing: robots, machine vision and mobile robots at the execution layer, manufacturing execution and SCADA orchestration above them, predictive maintenance and digital twins for reliability, and high-speed industrial networking linking all of it. Below, the four constraints that bind in practice: high-mix changeovers, exception handling, material variability, and maintenance labor. The lights-out technology stack and what binds it Mature components; integration under unattended conditions is the constraint Enabling stack Industrial robotsAssembly and weldingMachining and handling Machine visionAutomated metrologyCloses the quality loop AMRs and AGVsMaterial between cellsFeeds automated storage Manufacturing execution and SCADA layersScheduling, sequencing, real-time monitoring across cells Predictive maintenanceSensor data anticipates failurebefore it halts a dark line Digital twinsSimulate and optimize virtuallybefore physical deployment High-speed industrial networking, increasingly private 5GLow-latency links binding every element above Four binding constraints High-mix changeoversFrequent retooling needs humanjudgment; dark runs justifiedmainly for standardized volume Exception handlingA dropped part or jammed feederhalts the line, or scrap buildsundetected until discovered Material variabilityCells tuned to tight tolerancesdemand upstream consistencythat is itself costly to hold Maintenance laborSkilled technicians must be onsite or on call; their scarcityis a recurring adoption barrier Consensus in the engineering and integrator literature The limitation is reliable integration under unattended conditions, not robot capability. This is why truly dark factories remain exceptional. DataDeep.Tech - Dark Factories
CNC Manufacturing and Decentralized Industry: Why Distributed Production Still Depends on Concentrated Machine Tool Supply
CNC enables distributed production, but Fanuc, Siemens, and a few OEMs control the tooling layer every decentralization strategy depends on.
DataDeep TechJohn D

3. Documented Operating Examples

The evidence base is uneven, and several widely repeated claims rest on operator assertion rather than independent verification. The examples below are classified against the three-condition scale.

FANUC (TYO:6954), the Japanese CNC and robotics maker, operates the most frequently cited lights-out example at its campus near Oshino-mura, Yamanashi, where robots assemble other robots. The durable headline figures, that the plant runs unsupervised for as long as 30 days at a time and produces robots at about 50 per 24-hour shift, trace to a single 2003 business-magazine article and describe the situation as of 2001; they are historical and should not be read as current specifications [4]. FANUC's own more recent materials indicate its automated factories, which use its robots to make robots, have grown substantially in capacity, and its headquarters factory produces on the order of 25,000 CNCs and robot controllers per month [20]. The company's practice of running machining cells unmanned for long hours, supported by its Zero Down Time predictive maintenance system, is well documented, and the Oshino operation is a credible case of continuous or near-continuous lights-out machining for standardized product [20]. Even here, human staff perform maintenance and oversight, so the plant is best described as extended lights-out operation for core processes rather than a wholly unmanned site.

Philips (AMS:PHIA) operates a high-end electric shaver plant in Drachten, Netherlands, that is frequently described as a dark factory. The verified reality is high automation with a very small resident crew: the line uses more than 120 robots (Adept/Omron SCARA and six-axis units), and the human presence is a small number of quality-assurance staff, commonly reported as around nine, plus engineers who feed raw materials and an on-call team to keep robots running [13][14]. Detailed contemporaneous reporting describes the factory as brightly lit and staffed at the edges, with final acoustic quality inspection still performed by human workers because the human ear remained the best instrument for verifying each shaver [13]. Drachten is therefore a boundary case of high automation, not continuous lights-out operation, though its assembly stations can run with minimal intervention.

Xiaomi's (HKG:1810) Changping smart factory near Beijing, which the company launched in July 2024, is the most prominent recent claim. Xiaomi states the site spans 81,000 square meters, cost about 2.4 billion yuan (roughly 330 million US dollars), runs 11 production lines, and has annual capacity of 10 million flagship smartphones including foldable models [19]. Independent analysis notes that the widely promoted "one phone per second" figure overstates throughput; the realized rate is closer to one phone every 3.15 seconds averaged over a full year [19]. Xiaomi itself reports 81 percent automation across the line and acknowledges that supervision and maintenance staff remain, monitoring from a control room; later reporting describes on the order of 220 workers overseeing the production line, which materially undercuts the fully unmanned framing [19]. Changping is thus a very high automation facility with a small resident crew and lights-out sections, not a verified continuously unmanned plant; the "dark factory" label here is substantially the operator's own framing.

Changying Precision Technology, in Dongguan, China, is the origin of the much-cited claim that a plant cut its workforce from 650 to 60 (a 90 percent reduction), raised per-person output, and lowered defect rates, with management stating headcount could fall to 20. These figures, reported from 2015, come from the company's general manager via Chinese state media and have not been independently audited; they describe high automation with a residual crew monitoring lines and control systems, not a fully dark plant, and should be treated as operator-asserted [10][11].

Hon Hai Precision Industry (TPE:2317), known as Foxconn, operates several facilities it describes as "lights-off," beginning with its Foxconn Industrial Internet plant in Shenzhen, designated a WEF Lighthouse in 2019. Verified specifics include AI-driven optical inspection, automated optimization, and self-maintenance systems, with WEF reporting efficiency gains around 30 percent and a reduced stock cycle at the site [9][8]. Foxconn holds one of the largest clusters of WEF Lighthouse designations [8]. These are genuine high-automation showcases with documented performance gains, but the "lights-off" description applies to specific processes and shifts rather than to whole plants running continuously unmanned; Foxconn continues to employ very large workforces at its major campuses [8].

The Siemens (ETR:SIE) Electronics Works Amberg in Bavaria is the instructive boundary case. It is explicitly not fully dark: about 75 percent of the value chain is handled autonomously by machines, but a human still places the initial bare circuit board on the line, and people remain central to development, planning, and exception handling [5]. Amberg produces roughly 15 million Simatic programmable controller components per year across more than 1,000 product variants with about 350 changeovers per day, at a reported quality rate of 99.99885 percent [5][6]. It demonstrates how far conventional high automation can go while remaining a staffed plant, and it usefully marks the ceiling of what is achievable without going dark.

The Tesla (NASDAQ:TSLA) case documents a failure mode. In 2018, during Model 3 ramp, Tesla's attempt to over-automate final assembly, including what its chief executive called a "crazy, complex network of conveyor belts," failed to hit output targets and was partially reversed [12]. The episode is a well-sourced illustration of the exception-handling and adaptability constraints that make full automation counterproductive in high-complexity, variable assembly.

4. Adoption Metrics and Market Context

The macro data show automation deepening rapidly, but they do not measure genuinely lights-out capacity, which no credible source quantifies directly. According to the International Federation of Robotics, 542,000 industrial robots were installed worldwide in 2024, more than double the count of a decade earlier, and the total operational stock reached 4,664,000 units, a 9 percent annual increase [1]. China accounted for 54 percent of 2024 installations (295,000 units, the highest annual total on record) and holds an operational stock exceeding 2 million, the largest of any country and roughly 4.5 times Japan's [1]. Robot density, the IFR's preferred adoption barometer, continued to climb, with the Republic of Korea highest at 1,220 robots per 10,000 manufacturing employees, ahead of Singapore and Germany [1][2]. These figures capture automation intensity, not the removal of humans from the floor; a high robot density is a necessary but not sufficient condition for lights-out operation.

Market sizing for the enabling layers is consistent with sustained growth. Machine vision, central to unattended quality control, was estimated at USD 20.4 billion in 2024 and is projected to reach USD 41.7 billion by 2030 at a 13.0 percent compound annual growth rate; estimates across research houses vary widely because of differing scope definitions, so this should be read as indicative rather than precise [15]. Industrial robot unit costs have fallen over the past decade, a trend that has widened the set of economically automatable tasks [16].

The vendor landscape that supplies this capability is concentrated. FANUC, ABB (NYSE:ABB), Yaskawa Electric (TYO:6506), and KUKA (owned by Midea Group, SHE:000333) anchor the industrial robot segment; Siemens and Rockwell Automation (NYSE:ROK) supply control, MES, and digital twin software; and Keyence (TYO:6861), alongside others, leads machine vision. The WEF Global Lighthouse Network, co-founded with McKinsey, had grown to 201 designated sites by September 2025, a useful proxy for the leading edge of advanced manufacturing, though Lighthouse status denotes advanced digital integration rather than lights-out operation, and China hosts more than 40 percent of the network [7]. The honest assessment is that verified continuously lights-out manufacturing capacity is a very small fraction of installed automation, concentrated in semiconductors, standardized electronics components, and specific machining and molding cells, and that much of the "dark factory" narrative is vendor and operator framing ahead of verified practice [3].

5. Economics

The economic case for lights-out operation is strongest where capital can run against many productive hours. The core lever is utilization: unattended nights and weekends convert idle capital into output, and two-shift or continuous operation roughly halves payback relative to a single staffed shift by running the same equipment across more hours. Reported payback periods for unattended CNC machining cells commonly fall in the range of 18 to 36 months, with the shorter end associated with higher-volume, higher-utilization operations; these are practitioner and vendor figures and vary with local labor rates, utilization, and uptime rather than independently audited results [18].

Capital intensity is substantial and extends well beyond robot arms. New industrial robots are frequently quoted in a wide band from roughly 25,000 to over 100,000 US dollars per unit, but the binding costs are integration, tooling, safety systems, monitoring instrumentation, and commissioning, which can multiply the hardware figure several times over [16]. Tool and process monitoring, broken-tool detection, and predictive maintenance instrumentation are the necessary mechanisms that prevent an undetected fault from destroying a night's production [18].

Labor is reduced but not eliminated, and its composition shifts. Direct assembly and machine-tending headcount falls sharply, while demand rises for maintenance technicians, controls and automation engineers, and remote supervisors. This substitution is the source of both the cost savings and a principal adoption constraint, because the specialized labor that remains is scarce and expensive [3]. Quality and yield effects are generally favorable where automation is well matched to the process: operators report lower defect and rework rates from the consistency of automated execution, which is a recurring, though largely operator-reported, benefit [10][11]. The counter-case is Tesla's 2018 experience, where forcing automation onto a task poorly suited to it degraded rather than improved output, underlining that yield gains are conditional on process fit [12].

6. Significance for the Built Environment and Planning

Designing for genuine lights-out operation changes the building itself. As no one is on the floor during production, the plant can shed or shrink the systems that exist only for human occupancy: general illumination, comfort heating and cooling, restrooms, break rooms, cafeterias, and large parking and circulation areas. Industry sources identify this reduction in human-centered space and systems as the dark factory's largest architectural advantage, lowering material costs, shortening construction timelines, and reducing long-term operating expense; energy use narrows toward what the equipment and any process-specific climate control require [3]. The caveat is that these savings are realized only where a facility is designed as lights-out from the outset. Where product-quality requirements impose their own environmental controls, such as the clean, dust-controlled conditions in semiconductor fabs or precision electronics, those loads persist regardless of human presence and can dominate the energy profile [3].

Two planning frictions deserve note. First, fire and life-safety codes are written substantially around human occupancy. Egress, corridor, and alarm provisions are keyed to occupant load, but automatic sprinkler requirements under standards such as NFPA 13 are driven by the combustibility and quantity of contents rather than by whether people are present, so a lights-out plant does not escape suppression obligations and may face them on largely the same terms as a staffed one; removing occupancy reduces some life-safety design drivers but not the core property-protection requirements [17]. Second, siting and land-use logic shifts: freed from the need to draw and retain a large local workforce, a lights-out plant can be located for proximity to materials, energy, or markets rather than labor pools, which has implications for regional development and for the local employment base that conventional factories anchor [13]. The local labor-market effect should not be overstated in the near term, because the facilities that could plausibly go dark are a small share of manufacturing and because they generate demand for higher-skilled technical roles even as they reduce line labor [3].

7. Outlook

The most probable near-term trajectory is the continued spread of the "dim" factory: extensive automation with lights-out shifts and cells, retaining a small resident crew for supervision, maintenance, and exception handling [3]. This projection assumes no step-change in general-purpose robotic dexterity and reasoning of the kind that would resolve the exception-handling and high-mix constraints; it also assumes continued declines in hardware cost and continued scarcity of skilled automation labor [3][16]. Under those assumptions, continuous lights-out plants will remain concentrated where products are standardized and volumes high, expanding incrementally rather than displacing the staffed factory. The principal variable that would accelerate the timeline is a demonstrated, deployable advance in machine reasoning and manipulation flexible enough to handle non-standard situations without human intervention; until such capability is proven at industrial scale, the fully dark factory remains an aspiration realized only at the margins [3].

References

  1. International Federation of Robotics. 2025. "Global Robot Demand in Factories Doubles over 10 Years (World Robotics 2025 Report)." Frankfurt: IFR.
  2. International Federation of Robotics. 2024. "Global Robot Density in Factories Doubled in Seven Years (World Robotics 2024 Report)." Frankfurt: IFR.
  3. Gray. 2026. "Shining a Light on the Lack of Fully Automated Dark Factories." Gray Insights, January 6.
  4. Null, Christopher, and Brian Caulfield. 2003. "Fade to Black: The 1980s Vision of 'Lights-Out' Manufacturing Is a Dream No More." Business 2.0 Magazine (CNNMoney), June 1.
  5. Siemens. 2015. "Fact Sheet: Amberg Electronics Plant." Amberg: Siemens AG, February 23.
  6. Capgemini Research Institute. 2020. "Discussion with Dr. Gunter Beitinger, SVP Manufacturing and Head of Factory Digitalization, Siemens."
  7. World Economic Forum. 2025. "Global Lighthouse Network 2025: World Economic Forum Recognizes 12 New Sites Driving Holistic Transformation in Manufacturing." Geneva: WEF, September 16.
  8. CommonWealth Magazine. 2024. "AI Robots Go to Work at Three Foxconn Lights-off Factories." September 18.
  9. GSMA. 2019. "Foxconn Lights-off Factory." GSMA Internet of Things Case Study.
  10. People's Daily Online. 2015. "First Unmanned Factory Takes Shape in Dongguan City." July 15.
  11. World Economic Forum. 2017. "This Company Replaced 90% of Its Workforce with Machines. Here's What Happened." February.
  12. CNBC. 2018. "Elon Musk Admits Humans Are Sometimes Superior to Robots, in a Tweet about Tesla Delays." April 13.
  13. Markoff, John. 2015. Machines of Loving Grace: The Quest for Common Ground Between Humans and Robots. New York: Ecco/HarperCollins.
  14. Assembly Magazine. 2012. "Automation Profiles: Robots Help Philips Shave Assembly Costs." June 1.
  15. Grand View Research. 2025. "Machine Vision Market Size and Share, Industry Report, 2030."
  16. ARK Investment Management. 2019. "Are Industrial Robot Costs Hitting an Inflection Point?"
  17. National Fire Protection Association. 2025. "Occupancy Classifications in NFPA 13." NFPA Blog, January 14.
  18. Aerospace Manufacturing and Design. n.d. "A Guide to Unattended or 'Lights Out' CNC Machining."
  19. SlashGear. 2024. "How Xiaomi's Dark Factory Robots Can Pump Out a Phone in Mere Seconds."
  20. FANUC. n.d. "FANUC Factories." FANUC Europe corporate site.

Read more