The Robot Has Left the Laboratory. It Is Now on the Factory Floor.

Real work. Real robots. Real displacement. The boardroom conversation has barely started.

Audio summary: https://youtu.be/p45SBCD6G0c

Sign up for my Substack daily AI newsletter here.

See my AI Training course portfolio for corporate Business Leaders here.

Article link: https://open.substack.com/pub/johanosteyn/p/the-robot-has-left-the-laboratory?r=73gqa&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/

For the past two years, the dominant conversation about AI and the future of work has taken place in boardrooms, consulting reports, and conference keynotes — and it has been almost exclusively about knowledge workers. Which professional roles will AI augment? Which analytical tasks will be automated? Which white-collar functions will shrink as large language models absorb the work that graduate recruits used to do? These are important questions. They are also the wrong questions to be leading with. Because the first and largest wave of AI-driven disruption to the global workforce is not arriving in the office: It is arriving on the factory floor, in the distribution centre, and at the logistics dock — and it is running years ahead of the boardroom conversation that should be accompanying it.

CONTEXT AND BACKGROUND

The evidence of what is already happening in physical work environments is no longer speculative. At Auto China 2026, held in Beijing in late April, the world’s largest automotive exhibition showcased not just new vehicles but a fundamental restructuring of how those vehicles are made. Chinese automaker Xiaomi’s AI factory in the Beijing suburb of Yizhuang produces a new electric vehicle every 76 seconds, with 700 robots and 180 autonomous mobile transport units having largely taken over the work previously performed by human assembly workers, with central computers and AI agents controlling the majority of processes. This is not a pilot programme. It is not a proof of concept. It is a functioning production line, operating at scale, producing vehicles at a pace and cost that human-staffed factories cannot match.

China’s national strategy makes the direction of travel explicit. The 2026 Government Work Report proposed creating a new form of intelligent economy for the first time, explicitly requiring the accelerated promotion of new-generation intelligent terminals and agents, with autonomous vehicles and humanoid robots named as policy priorities alongside AI-integrated manufacturing. At the same show, companies including XPENG, Chery, and Geely showcased humanoid robots developed not despite being automotive manufacturers but because autonomous driving and embodied AI share the same core technology stack — sensor fusion, environment mapping, and real-time decision-making. Geely’s founder has announced his intention to make the company the largest robotics manufacturer in the world, repurposing technology developed for autonomous driving to control physical robots. The vehicle factory and the robot factory are converging into a single physical AI ecosystem — and it is being built at a national scale under explicit government direction.

The global humanoid robotics market is accelerating to match this ambition. Industry analysts estimate 50,000 to 100,000 humanoid robot shipments in 2026 alone, with unit costs dropping rapidly toward the fifteen to twenty thousand dollar range as manufacturing scales. Over seven billion dollars in venture capital flowed into humanoid robotics companies between 2023 and 2025. Modern humanoid robots can now operate sixteen to twenty-hour shifts before recharging, making them viable for two-shift factory and warehouse operations. The International Federation of Robotics confirms that automotive manufacturing, warehousing, and logistics are the sectors seeing the fastest humanoid deployment — the three sectors that together employ the largest concentrations of lower-skilled workers globally.

INSIGHT AND ANALYSIS

The conventional wisdom that physical work was relatively safe from AI disruption rested on a reasonable observation: robots had historically struggled with the dexterity, adaptability, and environmental variability that human workers handle instinctively. That observation is no longer accurate. AI-powered dexterity, driven by large language models and vision-language-action models, now enables robots to understand verbal instructions, adapt to novel situations, and handle irregular objects — capabilities that were impossible just three years ago. AI-driven imitation learning enables robots to copy human movements directly from video demonstrations with increasing accuracy, compressing what was previously years of programming into days of observation. The technical barrier that protected physical work from automation has not merely lowered. It has effectively collapsed for the categories of task that dominate manufacturing, warehousing, and logistics employment.

The academic evidence on what this means for employment is accumulating with uncomfortable speed. Peer-reviewed research published in the Journal of Open Innovation: Technology, Market, and Complexity in March 2026, based on a decade of quarterly employment data and a structural vector autoregression model, found that positive AI shocks trigger an immediate and statistically significant contraction in manufacturing employment — and that AI shocks account for nearly 30 per cent of forecast-error variance in manufacturing employment over a ten-quarter horizon. This is not a projection about what might happen when humanoid robots become widespread. It is a measured finding about what has already been happening — in economies at every level of development — as AI has been integrated into manufacturing processes over the past decade. The deployment of humanoid robots at scale accelerates this dynamic rather than initiating it.

McKinsey Global Institute's landmark research on workforce transitions estimates that between 400 and 800 million individuals could be displaced by automation and need to find new jobs by 2030, with up to 375 million workers — roughly 14 per cent of the global workforce — needing to switch occupational categories and learn new skills entirely. Those estimates were considered alarming when first published. The deployment of humanoid robots at scale in 2026 — a development McKinsey's researchers could not fully anticipate — means the upper end of that range is now the more plausible scenario, not the outer limit.

The sectors McKinsey identified as facing the earliest and deepest disruption are the same sectors now seeing the fastest humanoid deployment: manufacturing, warehousing, agriculture, and food service — not financial services, professional services, or the knowledge economy roles that have dominated boardroom AI risk discussions. This inversion of the expected disruption sequence is the critical planning assumption that most organisational AI strategies have still not absorbed. The workers AI will reach first are not the ones most boards are thinking about.

What makes this particularly significant for global business leaders is the competitive dimension. China is not simply automating its own workforce. It is using AI-integrated manufacturing to produce goods faster, cheaper, and at higher quality than competitors operating with more human-intensive production models — and then exporting those goods into the same global markets. Francois Roudier, secretary general of the International Organisation of Motor Vehicle Manufacturers, the global federation of automotive trade groups, told reporters at the Beijing Auto Show in April 2026 that Chinese automakers are now so advanced they are upending the global car industry — and that what is unfolding is not a transition but a revolution. The AI factory advantage compounds annually through continuous machine learning improvement. Every manufacturer still operating human-intensive production lines is competing against a benchmark being reset by physical AI faster than traditional industrial strategy can respond to.

IMPLICATIONS

For boards and executive teams, the planning implication is direct and uncomfortable. The AI strategy most organisations have developed over the past two years was built around a specific threat model — AI augmenting and displacing knowledge work — and has produced governance frameworks, reskilling programmes, and risk assessments calibrated to that model. That model is incomplete. An organisation with manufacturing, logistics, warehousing, or supply chain operations that has not yet assessed the physical AI disruption timeline for those operations is not behind on an emerging trend. It is behind on a transition that is already underway in the most competitive facilities in its industry.

The workforce implications extend beyond individual organisations to the societies in which they operate. The workers most immediately at risk from physical AI disruption are those with the fewest alternative options — lower-skilled, lower-wage workers in manufacturing and logistics who cannot easily transition into the knowledge economy roles that AI is simultaneously creating at the other end of the labour market. As the Human Sciences Research Council’s research on AI and employment in developing economies notes, the distribution of AI’s productivity gains in the absence of deliberate policy intervention tends to widen rather than close existing inequalities. For business leaders who understand that social stability is a condition of long-term commercial viability, this is not a concern that can be delegated to government. It is a material consideration for how physical AI is deployed, at what pace, and with what investment in the transitions it requires.

The policy gap is equally significant. China’s physical AI strategy is explicit, funded, and embedded in national industrial policy. Most other governments have published AI principles documents that do not address the specific disruption to physical work environments, the competitive pressure from AI-integrated foreign manufacturers, or the workforce transitions required before 2030. The gap between a principles document and an industrial strategy is the gap between acknowledging that physical AI is transforming the global economy and actually preparing for what that means. Most of the world currently occupies that gap — while the factories of the future run on without waiting for the policy frameworks intended to govern them.

CLOSING TAKEAWAY

The robot has left the laboratory. It is clocked in, on shift, and producing output at a pace and cost that changes the competitive mathematics of every industry it enters. The boardroom conversation about AI disruption has been sophisticated, well-resourced, and largely focused on the wrong wave. Manufacturing, logistics, and warehousing are not the sectors where AI disruption will eventually arrive. They are the sectors where it has already begun — in facilities producing electric vehicles every 76 seconds, in warehouses where autonomous systems have replaced the majority of human pickers, and in supply chains being restructured around physical AI capabilities that did not exist three years ago. The executive teams that recognise this transition now, build it into their competitive strategy, and plan honestly for its workforce consequences will be the ones whose organisations are still relevant when the second wave — the knowledge work disruption they have been so carefully preparing for — finally arrives.

Author Bio: Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net