Hong Kong has always been a city of crossings — of goods, people, and ideas. For me, it is also personal: as the child of Chinese and English parents, it felt like a living bridge between East and West. Its fortunes have shifted with protests, lockdowns, and geopolitical tensions, yet Hong Kong remains a gateway — not just to trade, but to the future of industrialization. Almost everything we touch daily — from smartphones to medicines to furniture — has, in some way, been conceived, financed, manufactured, or shipped through the Greater Bay Area.
Industrialization is often dismissed as the unglamorous machinery of progress, but it is the substrate of our lives. And today, it is being reimagined. The rise of the industrial internet promises to transform not just how things are made, but how societies endure. If the consumer internet gave us attention and convenience, the industrial internet is about survival and resilience.
A decade ago, the buzzword was Industry 4.0. Born in Germany, it was less a global revolution than a modernization plan for the Mittelstand — the network of small and mid-sized firms that powered exports but lagged in digital adoption. The vision was compelling: factories filled with sensors, machines connected by data, production made flexible and personalized. The reality was mixed. Isolated “lighthouse factories” proved the concept, but most systems remained siloed. Machines talked to operators, not to each other. What emerged were pockets of excellence, not ecosystems of intelligence.
Today, the challenge — and the opportunity — is different. The future of industrialization lies not in smarter individual factories, but in intelligent networks of suppliers, logistics hubs, and energy systems: ecosystems coordinated by AI agents, digital twins, and real-time data flows. In other words, a new kind of keiretsu for the 21st century — one that runs on code.
In postwar Japan, the keiretsu was more than an industrial structure — it was an operating philosophy. Networks of suppliers, manufacturers, and distributors, bound together by cross-shareholdings and long-term trust, gave companies like Toyota an edge that no single firm could match. The power lay not just in efficiency but in collaboration: decades of tacit knowledge shared across a web of partners.
That model faded as globalization, outsourcing, and financialization pulled supply chains apart. But the logic of the keiretsu is reemerging, this time not through shareholdings or geography, but through data, digital platforms, and AI agents. As I discuss in the video below, in this new form — the digital keiretsu — value is created by networks of autonomous organizations and intelligent systems working in concert, as if they were parts of a single organism.
This shift is not theoretical; it is already visible in practice. Across industries, some of the world’s most influential companies are experimenting with new forms of coordination — weaving together suppliers, platforms, and machines into networks that behave less like supply chains and more like living systems. Their approaches differ, but taken together they illustrate the contours of a new industrial order.
Walmart’s competitive advantage was once brick-and-mortar muscle: vast distribution centers, fleets of trucks, and ruthless cost optimization. Today, its edge lies in intelligence.
At the core is Scintilla (previously known as Luminate), Walmart’s data platform that integrates sales, logistics, and supplier flows into a unified ecosystem. AI agents embedded in the system forecast demand spikes, trigger automatic procurement, and adjust promotions across stores and digital channels. In warehouses worldwide, drones and other physical AI systems are beginning to scan shelves and feed real-time data back into the loop.
For suppliers, connecting to Walmart’s nervous system is less about compliance and more about co-creating value. What emerges is a retail keiretsu bound by data, where Walmart and its partners share visibility and optimization across the entire chain.
If Walmart’s keiretsu is about scale, Shein’s is about speed. The Chinese fast-fashion giant has built a supply network that looks more like a neural net than a traditional chain.
Every day, Shein’s AI engines scour social media and search data to spot micro-trends. These insights are instantly translated into micro-orders — sometimes as few as 50 garments — pushed out to thousands of small factories clustered around Guangzhou.
Factories don’t negotiate with human managers; they answer to algorithms. Those that deliver on time and at quality get more orders; those that miss targets vanish from the system. In place of decades of tacit trust, Shein enforces algorithmic trust. Compliance, payments, and scheduling are dictated by the platform.
The result is swarm-like: thousands of semi-automated producers behaving as one agile organism, guided not by hierarchy but by code.
Amazon’s fulfillment network may be the closest thing to a living keiretsu at scale. Its warehouses hum with millions of autonomous robots, while AI systems orchestrate picking, packing, and routing with precision.
The real innovation lies in how suppliers plug in. Through Fulfillment by Amazon (FBA), businesses of every size effectively outsource logistics to Amazon’s AI brain. Each partner’s data feeds into the system, and every optimization — from a smarter picking algorithm to a better routing pattern — propagates instantly across the network.
Amazon doesn’t own most of the goods it ships, but it owns the intelligence that moves them. In doing so, it has created a platform keiretsu: decentralized in ownership, but unified by a common AI operating system.
Tesla’s Gigafactories aren’t just factories — they are software platforms in physical form. Before a single robot is installed, each production line is modeled as a digital twin, stress-tested by AI, and optimized in simulation. Once live, the same AI agents continuously tweak operations, rebalancing flows in real time.
Suppliers don’t just deliver parts; they integrate into Tesla’s digital ecosystem. Data is the ticket of entry, and interoperability is non-negotiable. In effect, Tesla has rebuilt the keiretsu in digital form: an ecosystem where suppliers orbit not around equity ties, but around Tesla’s factory-as-operating-system.
Across these examples, a new industrial logic is taking shape. The digital keiretsu is not defined by proximity, but by data: APIs and digital twins knit together partners scattered across continents. It is coordinated less by managers than by machines: AI agents handle the rhythms of production and logistics, while humans set direction and strategy. It is sustained not by ownership but by platforms: firms remain independent, yet bind themselves into shared digital infrastructures. And it is optimized not only for efficiency but for resilience: where traditional supply chains chased cost, these new ecosystems prize adaptability and collective problem-solving.
The deeper shift is this: industrialization is no longer about efficiency, it is about civilizational resilience. A century ago, the assembly line reshaped consumer society. Today, the digital keiretsu may determine our capacity to survive and thrive as a species. Climate change, pandemics, energy shocks, food insecurity — these are not consumer problems, they are industrial ones. Solving them demands networks of intelligence that extend beyond the walls of any single company.
The consumer internet may have shaped how we connect and communicate. The industrial internet will shape how we endure. In the greater story of what lies ahead, it may be this quieter, deeper transformation that will matter most.