The Return of Vertical Integration
Cars, clouds, and the turn away from the fissured firm.
In a previous post, I wrote about how frontier corporate strategies evolved from building it all to building nothing at all. The Fordist firms of the postwar era were vertically integrated giants, with production capabilities spanning every stage of the value chain. But beginning in the 1970s, the rise of shareholder value primacy, outsourcing, and the IT revolution transformed how leading firms operated. Labor- and capital-intensive production was offshored, while knowledge-intensive functions—like R&D, design, and IP management—were kept in-house.
Apple is a good illustration of the new model. It designs its products in California, retains tight control over its software ecosystem and intellectual property, but outsources nearly all of its manufacturing to contractors like Foxconn in China. From tech firms like Apple to consumer brands like Nike, firms in this era focused less on making things and more on extracting value through intangible assets: patents, licenses, platform fees, brand markups, and financial instruments.
This model of vertical fragmentation made many American firms fabulously profitable by allowing them to tap into cheap labor and deregulated markets abroad. But those profits came at a cost. This fragmentation gutted the American working class, who were once the anchor of Fordist production and shared in the rewards of rising productivity. And, as many have pointed out, decades of offshoring has eroded the nation’s capacity to produce at home.
Today, frontier firms are once again shifting their organizational strategies. From cloud giants to EV manufacturers, some of the most influential companies of the 21st century are turning back to vertical integration. However, this shift is not a response to the social costs of vertical fragmentation. Instead, it reflects a strategic pivot in the type of innovation these firms now prioritize.
In this post, I'll outline how frontier firms in cloud/IT and auto manufacturing—two sectors that were once poster children of the fragmented model—are pursuing vertical integration. Next week, I’ll dive into why some firms are turning to vertical integration, and what it means for innovation.
From the horizontal platform to the vertical cloud
During the Web 2.0 platform era, tech firms were universally seen as asset-light, agile, and defined by their software—not their hardware. It wasn’t that people ignored the underlying infrastructure these platforms depended on, but rather that building and maintaining that infrastructure was viewed as a low-margin, low-status task best left to less innovative players. Instead, the platform business model demanded scale, rather than control over every layer of the technology stack. The most competitive firms were asset-light, owning as little of the production stack as possible while focusing only on maximizing horizontal expansion.
This imperative for rapid expansion meant platform firms needed to decouple their software from underlying hardware, ensuring users could access their services across any operating system or device—whether Windows, Linux, or Mac; Android or iOS. It also shaped their approach to mergers and acquisitions, leading them to prioritize the purchase of horizontal competitors rather than complementary firms upstream. Facebook’s acquisitions of Instagram and WhatsApp, for example, were aimed at consolidating user bases and expanding reach rather than integrating new technological capabilities.
Over the past decade, some of the biggest players in tech have repositioned themselves around cloud computing and AI as the next major frontier of innovation. In the process, they’ve begun to reverse a decades-long trend toward horizontal sprawl by vertically integrating elements of the stack.
The leading cloud firms—Amazon, Microsoft, and Google—are now investing more heavily than ever in building physical infrastructure as a core source of competitive advantage. Their competencies extend well beyond the software layer (e.g., virtualization, data and AI services, and developer tools) and today includes building physical infrastructure such as custom chips, specialized server racks and cooling systems, and massive data centers. Even the energy required to power these facilities are no longer left to the market. Companies like Microsoft are investing directly in renewable energy projects to secure a stable, long-term power supply for their increasingly energy-intensive operations.
Initially, cloud infrastructure followed the logic of the platform era: fill data centers with off-the-shelf, commodity hardware and prioritize universal compatibility. The goal was to maximize flexibility such that any software could run on any hardware. But as workloads have grown more complex—especially with the rise of AI—the cloud's flexibility-first paradigm has started to show its limits. Performance bottlenecks, latency issues, or sheer limitations in hardware performance have pushed cloud firms to pay closer attention to the underlying hardware.
In response, the most competitive cloud providers are moving away from hardware-agnosticism and embracing full-stack control. Indeed, a key part of their competitive advantage now lies in the recoupling of software and hardware—developing specialized infrastructure tailored for specific workloads. Google’s machine learning library, TensorFlow, for example, is tightly integrated with its custom-made Tensor Processing Unit (TPU) to maximize performance and efficiency.
One key difference between the Fordist model of vertical integration and the cloud’s emerging version is how access is structured. In the Fordist era, vertical integration meant little transparency or access for outside actors. In contrast, cloud providers are increasingly vertically integrated as well but in a more flexible way. Clients aren’t restricted to using only the top-layer applications; they can plug into any part of the stack (above the virtualization layer), whether it’s compute infrastructure, databases, developer tools, or pre-built AI models.
Automaker's shift from "just-in-time" production to end-to-end control
The auto industry has undergone a parallel transformation. In the 1980s, U.S. automakers suddenly found themselves outcompeted by their Japanese rivals. Japanese car makers weren’t offering radically different vehicles, but they gained a decisive edge on price—an advantage made possible by organizational innovations at the firm level. Instead of following the Fordist model that had dominated U.S. (and to some extent European) auto manufacturing in the mid-20th century, Japanese automakers pioneered a lean, just-in-time production system that emphasized flexibility and cost-cutting by minimizing inventory and relying on tightly managed supply chains.1
Toyota became the most successful practitioner of this model, ultimately surpassing its American rivals to become the world’s largest automaker by the 2000s. And its influence extended well beyond the auto industry, shaping the strategies of firms like Apple, which adopted similar principles of outsourcing and tight control over a global supplier network.
Today, however, the pendulum is swinging back. In contrast to the vertically fragmented, just-in-time model pioneered by Toyota, the new frontrunners of the EV race—Tesla and BYD—are embracing end-to-end control. Tesla’s vertical integration spans multiple layers of its business—from in-house manufacturing in its highly automated Gigafactories to software development for vehicle systems and driver's assistance. The company has also invested in producing its own battery cells with strategic partners (such as Panasonic) which allows the company to bring in outside expertise while keeping control internal. Even in distribution, Tesla bypasses traditional dealerships, selling directly to consumers and maintaining full control over the customer relationship.
BYD, China's giant EV maker who surpassed Tesla in 2023 in EV units sold, is even more vertically integrated. Where Tesla's Gigafactories are highly automated, BYD utilizes China's supply of cheap labor to manufacture its cars. But its major competitive advantage is in its batteries, arguably the most critical component of an electric vehicle, which are made entirely in-house. Unlike Tesla, that still sources the majority of its batteries from external producers like CATL (despite producing their own batteries), BYD began as a battery company, supplying rechargeable batteries for mobile phones before expanding into the automotive sector.
Today, BYD not only produces its own EVs and batteries at scale, but it is vertically integrated across the value chain: from manufacturing their own semiconductors to building in-house software and electric motors. BYD's ambition for vertical integration even extends upstream into lithium mining and downstream into owning its own massive carrier ships for shipping its cars aboard.2 The company also has facilities to recycle end-of-life batteries.
Conclusion
From cloud computing to EV manufacturing, some of the most innovative firms in the global economy are embracing this neo-Fordist approach and bringing increasing portions of their supply chains under direct control. But this pattern isn’t limited to just clouds and cars.
In retail, companies like Home Depot and IKEA are expanding into logistics by chartering their own container ships, with IKEA even going a step further by purchasing its own shipping containers. IKEA is also extending its control upstream, having acquired a Romanian forest in 2015 and forestland in Alabama in 2018. These purchases are aimed at securing raw materials and building a more sustainable and resilient supply chain.
Even in media, Netflix has moved beyond content distribution into full-scale production, building its own studios and producing original programming. Conversely, legacy entertainment companies are moving in the opposite direction—expanding aggressively into distribution by launching their own streaming platforms, as seen with Disney+. Across sectors, vertical integration is re-emerging not as a relic of the past, but as a forward-looking strategy for resilience, coordination, and control.
However, this doesn’t mean all firms will follow suit in this direction. Vertical integration is capital-intensive and comes with high coordination costs. This type of organizational maneuver is also usually (but not always) done from a position of market dominance. In a global economy still shaped by robust supply chains—trade tensions notwithstanding—sourcing upstream components from around the world remains the more viable option for many firms, especially in low-margin sectors. For most firms, the benefits of tighter supply chain control simply don’t outweigh the costs.
(Part two of this series will explore how vertical integration has become a critical engine of innovation—driving process innovation rather than product innovation, and reshaping how firms compete in today’s economy.)
BYD started investing in its own carrier ships when third-party shipping costs started increasing. In 2024, daily charter rates hit $150,000 per vessel. The company is now building its own ships to fuel its ambitions in becoming an EV export powerhouse, announcing this year that it would spend $687 million to build its own fleet.
### The Perils of Vertical Integration and the Lessons of History
I am deeply skeptical of the current trend towards vertical integration, often referred to as the neo-Fordist approach. While you have listed several companies that are adopting this strategy, there is a notable lack of empirical evidence demonstrating why this model is superior for shareholders. It is crucial to consider the broader implications and historical context of such business strategies.
One glaring omission in your analysis is the case of NVIDIA, which has emerged as one of the most valuable companies in recent years despite being a fabless chip company. This stands in stark contrast to Intel, which pursued a more vertically integrated approach and has lagged behind competitors like ARM, who were willing to outsource production to TSMC. This example underscores the potential pitfalls of vertical integration and the benefits of leveraging specialized tier-1 suppliers.
Vertical integration can indeed be advantageous in certain contexts, particularly when the necessary supply chain for new products does not yet exist, as was the case for companies like Tesla and SpaceX. However, the tier-1 supplier approach often proves to be more effective and efficient. The move towards vertical integration may be driven by a desire to extract more value from the supply chain, but it is also possible that this trend is merely a fad, and that this generation of CEOs will need to relearn the lessons of the past.
The mid-20th century was marked by the rise and fall of sprawling conglomerates, which ultimately proved to be unworkable nightmares. It is equally bizarre that the populists of today are nostalgic for the pre-shareholder value maximization era, where the management class had free rein to pursue personal pet projects with company assets. This nostalgia overlooks the inefficiencies and misallocations of resources that characterized that period.
In conclusion, it is essential to approach the trend towards vertical integration with a critical eye, considering the historical context and empirical evidence. The lessons of the past should not be dismissed lightly, and the potential benefits of tier-1 supplier relationships should not be overlooked.