SWITCHED ON
The daily technology series nobody asked for but everyone needed
The New Factory: Advanced Manufacturing and the Industrial Tech Revolution
3D printing, AI-driven production, digital twins on the factory floor, and the reshoring push that is trying to rebuild manufacturing capacity that took decades to offshore. The factory is being reinvented. The politics of who benefits are as complicated as the technology.
The first industrial revolution moved production from homes and workshops into factories. The second mechanised those factories. The third computerised them. The fourth is making them intelligent, adaptive, and in some cases capable of producing customised goods at the economics of mass production. The disruption to trade patterns, employment, and geopolitical advantage is going to be significant. It is also, in most mainstream coverage, almost entirely invisible.
Yesterday we went into crypto's second act — Bitcoin's institutional moment via spot ETFs, DeFi's genuine innovation alongside its spectacular hack record, the TerraUSD and FTX collapses as fraud and design failure rather than technology failure, stablecoins and CBDCs, and the honest accounting of what blockchain actually does well for users not motivated primarily by ideology or speculation. Today we are going somewhere considerably more physical. The factory. Specifically, the factory as it is being transformed by a convergence of technologies that together constitute what manufacturers and economists are calling the fourth industrial revolution: advanced robotics, 3D printing at production scale, AI-driven quality control and process optimisation, digital twins of production lines, and the reshoring push that is attempting to bring manufacturing capacity back to Western countries after decades of offshoring — and discovering that it is considerably more difficult and expensive than the political announcements implied.
01 — What Additive Manufacturing Has Become
3D printing — more precisely additive manufacturing, since the technology encompasses far more than the desktop plastic printers that most people associate with the term — has matured from a prototyping tool into a production technology deployed in aerospace, medical devices, automotive, and consumer goods manufacturing. The range of materials that can be printed has expanded from plastics to metals, ceramics, composites, and biological materials. The scale of parts that can be produced has expanded from small components to structural elements and entire assemblies.
GE Aviation produces fuel nozzles for its LEAP jet engine using additive manufacturing — the nozzles are printed as a single component that previously required assembling twenty separate parts. The printed nozzle is twenty-five percent lighter and five times more durable than its predecessor. Airbus uses additive manufacturing for structural brackets in the A350. The medical device industry uses patient-specific implants — hip replacements, cranial plates — produced by additive manufacturing from patient imaging data, with geometry optimised for each individual. These are not prototype or experimental applications. They are production manufacturing at commercial scale in safety-critical industries.
The economic logic of additive manufacturing is the opposite of conventional manufacturing's logic. Conventional manufacturing gets cheaper per unit as volume increases, favouring mass production of identical parts. Additive manufacturing costs the same per unit regardless of volume, and costs nothing extra to customise each unit. It favours personalised production, complex geometry, and low-volume high-value parts — a fundamentally different economic regime.
The limitations that prevent additive manufacturing from displacing conventional manufacturing for most applications are real and are not close to being overcome. Production speed is significantly slower than injection moulding or machining for most part geometries and volumes. Surface finish typically requires post-processing. Material properties can be anisotropic — stronger in some directions than others depending on print orientation. Cost per unit for high-volume standardised parts remains higher than conventional alternatives. The technology is genuinely transformative for specific applications and genuinely unsuitable for others, and distinguishing between them requires more precision than most technology coverage provides.
02 — Industry 4.0 on the Factory Floor
The "Industry 4.0" framing — coined at the Hannover Messe industrial trade fair in 2011 and subsequently adopted as a policy framework by multiple governments — describes the integration of digital technologies into manufacturing: sensors on every machine, real-time data collection and analysis, AI-driven process optimisation, and the connectivity between machines, supply chains, and customers that digital infrastructure enables.
The practical manifestation varies enormously between manufacturers. At the most advanced end — exemplified by facilities like Siemens' electronics manufacturing plant in Amberg, Germany, and the network of "lighthouse factories" identified by the World Economic Forum as exemplars of Industry 4.0 implementation — production lines run with near-zero defect rates, predictive maintenance flags equipment issues before failures occur, and digital twins of the factory floor are used to optimise production scheduling and test process changes in simulation before deploying them in the physical plant. These facilities are genuinely extraordinary demonstrations of what digital-physical integration in manufacturing can achieve.
At the other end of the spectrum — which represents the vast majority of actual manufacturing globally — small and medium-sized manufacturers are operating equipment that ranges from cutting-edge to decades old, with varying degrees of digital connectivity, limited data analysis capability, and workforces whose skills were developed for a different production paradigm. The gap between the lighthouse factories and typical manufacturing operations is vast, and the technology transfer required to close it is a major policy and industrial challenge that receives far less attention than the lighthouse showcases.
03 — Reshoring: The Politics and the Reality
The political consensus in the US and Europe has shifted decisively toward reshoring — bringing manufacturing capacity back from China and other low-cost production locations — driven by the supply chain vulnerabilities exposed by COVID-19, the semiconductor geopolitical concerns we covered in S2 EP02, and a broader reassessment of the strategic risks of depending on geopolitical rivals for critical manufactured goods. The Inflation Reduction Act, the CHIPS Act, and the EU's industrial policy initiatives represent the most significant government intervention in manufacturing location decisions in decades.
The economic reality of reshoring is considerably more complicated than the political announcements suggest. Manufacturing that offshored over decades did so because labour costs in low-income countries were a fraction of those in the US and Europe. Wages in China and other manufacturing centres have risen significantly over the past two decades, reducing but not eliminating the labour cost differential. Advanced manufacturing technology — robots, automation, AI-driven process control — reduces the labour intensity of production and therefore reduces the importance of labour cost differentials, making reshoring more economically viable than it was when the offshoring wave occurred.
But the supply chains, workforce skills, supplier ecosystems, and institutional knowledge that supported manufacturing in specific locations took decades to develop and cannot be rebuilt quickly. TSMC's Arizona fab delays, which we discussed in the semiconductor episode, are a high-profile example of a broader pattern: companies wanting to reshore find that the enabling ecosystem — trained workers, specialised suppliers, logistics infrastructure — does not exist in the destination location and takes years to develop. The political timeline for reshoring announcements and the economic timeline for reshoring reality are not aligned, and the gap is producing frustration on all sides.
04 — Robotics and the Manufacturing Workforce
The relationship between advanced manufacturing technology and manufacturing employment is one of the most politically sensitive and empirically contested questions in industrial economics. The optimistic framing: automation increases productivity, reduces costs, makes domestic manufacturing competitive, and creates new categories of higher-skill employment in operating, maintaining, and programming automated systems. The pessimistic framing: automation eliminates the entry-level and mid-skill manufacturing jobs that provided stable employment for workers without university degrees, and the new jobs it creates require skills that the displaced workers do not have and are costly and difficult to acquire.
The evidence is genuinely mixed, and the outcome appears to depend heavily on the pace of transition, the availability and quality of retraining, and the broader labour market conditions into which displaced workers are released. The German model — in which strong unions, works councils, and coordinated industrial policy have managed automation transitions with significant investment in worker retraining — has produced different outcomes from the US model, where automation-driven job losses in manufacturing communities have correlated strongly with the political and social disruption that characterised much of the 2010s and 2020s. The technology is not the determinant. The institutional context in which it is deployed is.
05 — The Supply Chain Lesson That Has Not Yet Been Learned
The COVID-19 pandemic exposed the fragility of just-in-time, geographically concentrated supply chains with brutal efficiency. The semiconductor shortage that lasted from 2021 to 2023 cost the global automotive industry alone hundreds of billions of dollars in lost production. The concentration of production for critical goods — semiconductors, pharmaceuticals, medical devices, rare earth processing — in single countries or single companies was revealed as a systemic vulnerability that no amount of logistics optimisation could protect against when the underlying production capacity was unavailable.
The policy response — reshoring, supply chain diversification, strategic stockpiling — is underway but incomplete. The economic incentives that produced concentrated supply chains in the first place — cost efficiency, specialisation, economies of scale — have not disappeared. Maintaining strategically redundant production capacity costs money and reduces the efficiency gains that motivated concentration in the first place. Companies and governments are genuinely wrestling with the trade-off between efficiency and resilience, and the outcomes will shape industrial geography for decades.
The honest conclusion is that the supply chain lesson has been partially learned and is being partially acted on, on a timeline that is considerably slower than the rhetoric of industrial policy suggests and that will produce genuine but incomplete change in the geographic distribution of manufacturing over the coming decade. The factory is being reinvented. Where it is located, who works in it, and who captures the value it produces are questions whose answers are still being written by the interaction of technology, policy, and the economic and political forces that shape both.
Tomorrow we are going somewhere that has been waiting since Episode One — the internet itself. Not the applications built on it, not the platforms, not the data practices, but the physical infrastructure: the undersea cables, the internet exchange points, the routing protocols, and who actually controls the pipes through which all of this flows. See you then.
Switched On is a daily technology series covering the ideas, systems, and arguments shaping the digital world. Opinionated. Witty. Occasionally wrong. Always worth the argument.
