This article is one of McKinsey’s contributions to the Global Lighthouse Network’s latest white paper, published on December 14, 2023. The first article in this series explored the evolution of AI and how leading manufacturers have harnessed it to propel leading-edge manufacturing from the learning phase to the doing phase of the Fourth Industrial Revolution (4IR). The second article examined what artificial intelligence looks like among today’s leading manufacturers. This final installment shows how the latest Lighthouses have built the capabilities to deploy AI and other 4IR technologies with both speed and scale.
In the mid-to-late 2010s, the Fourth Industrial Revolution—a group of cutting-edge technologies that could eventually redefine manufacturing—started to emerge. But this industry-shaping impact did not materialize, at least not at first. Even in 2018, when the Global Lighthouse Network was founded, over 70 percent of manufacturers reported being in ”pilot purgatory”—their use case pilots failed to have a significant business impact.
Nonetheless, the first Lighthouses began to crack the code, proving the potential impact of data-informed solutions such as automated guided vehicles (AGVs) and digital dashboards. But these early use cases were significantly less advanced than today’s—only 10 percent of top use cases in early cohorts relied on AI; by contrast, over 60 percent of today’s top use cases do. Moreover, early use cases took significantly longer to implement—averaging ten to 20 months, compared with the under six months that more than three-quarters of today’s Lighthouses have achieved.
Efficient implementation at scale is what sets today’s industry leaders apart. They no longer focus on piloting individual use cases; instead, they’ve built the capabilities to get new use cases right quickly, without trials. For the 25 companies with more than one Lighthouse site—and especially the ten advanced scalers, with more than four—entire factories serve as pilots for network scale deployment. Leaders now capture the value of 4IR technologies in ten to 50 factories at a go and have two or three times the ROI of organizations that are still working to find value in a single factory.
Scaling is a big challenge, and even leading companies struggle to achieve it. In a 2022 survey, only 11 percent of companies with a Lighthouse site said they had successfully scaled 4IR technologies across their production networks. The most critical bottlenecks they cited were fragmented data landscapes, legacy IT infrastructures, and a shortage of in-house talent.
As our colleagues Eric Lamarre, Kate Smaje, and Rodney Zemmel wrote in their recent book Rewired: The McKinsey guide to outcompeting in the age of digital and AI, achieving scale means “getting thousands of people across different units of the organization working together and working differently. It means bringing on new talent and developing accelerated learning loops that harness their skills and help them grow.” This is no small task. Yet Lighthouses, accelerating at just the right pace, have found that sweet spot. Some, in fact, have transformed sites in a mere matter of months. Let’s see how.
Lessons from leaders: Six steps to powering scale
Lighthouses, like digital leaders in other industries, have driven themselves up the adoption curve with a six-part approach to site transformation. (1) They first set the strategic road map. That’s like setting the GPS to guide the transformation toward an organization reimagined with technology and determining the route to prioritize and sequence the value at stake—including use cases (at the site level) and factories (at the network level).
These companies then build their delivery capabilities: the engine that powers the transformation. (2) The engine parts—its pistons, crankshafts, drivetrain, and timing belt—are programs to hire, train, and retain digital talent. (3) An agile operating model (often including digital studios) fosters speed, quality, and collaboration. (4) A technology backbone affords a clear, scalable, and distributed architecture for providing digital services and solutions easily. (5) The data architecture and governance enable critical decisions and ensure quality, easy consumption, and reuse.
But capabilities alone are not enough. (6) The final enabler—change management for adoption and scaling—is like the driver’s hands on the steering wheel. That is what maximizes value by ensuring the adoption and scaling of digital and analytics solutions. It involves building new skills and leadership characteristics and tightly manages the transformation’s progress and risks. And, of course, it validates the program’s impact by tracking key metrics in a standardized format. This enabler often includes a transformation office, which nearly 70 percent of Lighthouses cite as the most critical of the six to get right.
No less critical to the transformation’s functioning is the motor oil: ecosystem collaborations that keep the engine humming as universities, technology providers, innovation incubators, public entities, and many others shape best-in-class capabilities. Each Lighthouse in this recent cohort awarded in December 2023 has listed an ecosystem collaboration as a key enabler of the 4IR journey.
In all, the capabilities that pilot a factory transformation are the same ones that Lighthouses use to unlock network-level impacts at speed and scale. Johnson & Johnson’s latest Lighthouse (in Thailand), for example, implemented an energy management use case that was not only piloted but also scaled to 16 other sites within a year.
Setting your GPS with a strategic road map
The first enabler, the strategic road map, is like the GPS navigation system that plans a journey. Aligning senior leaders on the transformation’s vision and value, it involves a shared reimagining of business domains to deliver exceptional customer experiences and create competitive distance. We’ve observed that leading companies tend to fit one of three strategic archetypes as they plot their scaling course: IT-led, center of excellence (COE)–led, or “build and replicate” (Exhibit 1). Choosing the right scaling strategy depends on three factors: the diversity of your production processes, the staffing models of your sites, and the maturity of your enterprise information technology and operational-technology (IT/OT) stack.
For companies, such as Tata Steel, that have only a few but very large sites, it may make sense to swarm one factory at a time. That’s exactly what Tata did, first creating a Lighthouse factory in the Netherlands and then replicating the changes, one at a time, at two factories in India. Now, 80 percent of Tata Steel’s global production comes from one of the company’s three Lighthouse sites.
If a company has a large footprint, with diverse production processes and very site- or business unit–oriented staffing models, a capability-led approach to scale may be superior. Siemens, with hundreds of sites spanning dozens of manufacturing processes, tools, and systems, took this approach. Sites in Germany (Amberg) and in China (Chengdu) were chosen as digital-innovation hubs for disseminating use cases to other factories. COE teams across the network include leaders from multiple sites and are designed around specific technological capabilities, including a next-gen manufacturing execution system (MES) and digital twins.
Finally, companies with highly replicable processes across their production base and a strong, universal IT backbone may be best off scaling one use case or technology at a time. Contemporary Amperex Technology Co. Limited (CATL) is a prime example. Its dozens of factories have hundreds of battery production lines that all look and feel similar. The company’s IT and Ops departments, not its sites, host hundreds of data scientists, AI engineers, and other kinds of tech talent. CATL’s MES, enterprise resource planning (ERP), and other systems are fairly standardized, and it has invested in developing its own universal data profiles and name spaces to help standardize and accelerate rollouts. As a result, a new use case (such as a specific algorithm for set point optimization) can reach hundreds of production lines in just weeks—and sometimes less.
Building your 4IR capability engine
Today’s Lighthouses teach us that the principles explained in Rewired work for manufacturers. Much hinges on the capability engine, guided by an effective strategy: enablers two to five. Let’s open up the hood and check things out.
Digital talent
Leading organizations know that talent is just as vital as technology. Moreover, they know that each company tends to have unique talent needs: technology can typically be replicated once it becomes established, but skills and knowledge must be tailored. Learning must therefore be not only customized but also continually renewed and maintained. Front-runners know there’s never a time to rest. Companies can’t take the talent pipeline or their understanding of their skills and talent gaps at any given moment for granted.
LONGi, a Chinese solar-technology manufacturer, exemplifies a talent strategy focused on reskilling the existing workforce. To tailor the training of almost 1,000 employees, the company implemented an evaluation-training-certification method involving the identification of skill gaps, personalized learning, and value-based certification. A diagnostic informed a comprehensive talent road map. The 3F model—forums (learning), field (practice), and feedback (evaluation)—grounded all learning, especially for new digital roles, such as agile coaches and data scientists. A closed-loop, impact-driven certification process mapped the impact of the work of employees while recognizing and supplementing their capabilities (Exhibit 2).
Agile operating model
An effective capability engine must realize a challenging goal: bringing business and technology together to throttle up. This is a leadership challenge, and success depends on alignment across functions. An appropriate allocation of resources, clear incentives, and intelligently built teams are essential components of the agile operating model, which couples people resources with tech and data resources, setting up regular and consistent forward momentum (Exhibit 3).
ReNew is a prime example: its ReNew Digital (ReD) team of nearly 30 engineers and data scientists runs five or six agile pods, each focused on discrete use cases, located in its innovation studio. The team is responsible for leadership buy-in, including active sponsorship by the chief digital officer (CDO); aligns incentives with an independent profit and loss, funded in part by the impact of new use cases; and uses a multistakeholder staffing model that includes end users, internal business owners, data scientists, engineers, and process owners from all relevant departments.
Technology backbone
To achieve an at-scale transformation and unlock business value, a strong technology backbone is vital. It’s important to understand that this means more than just one or even a handful of substantial new technology investments. Those tend to be localized, which can promote fragmentation in the long run—creating, ironically, a force antithetical to transformation at scale. Rather, the technology backbone must be just that: the infrastructure needed to convey signals across the organization for at-scale deployment. It’s therefore necessary to think about at-scale deployment from day one, so leading organizations prioritize the accessibility and adaptability of their data environments by using decoupled architecture solutions (such as microservices), along with advanced development environments and tools.
Ingrasys demonstrates the benefits of starting small, but with a clear growth plan that intelligently integrates an ecosystem approach (Exhibit 4). Manufacturing servers require a significant number of black box vendor technologies, and Ingrasys addresses this need with clear design principles and vendor requirements that provide for security and scale. As a result, most (if not all) of the company’s vendors must open up their firmware to make data and controls accessible, design data outputs to mesh with the Ingrasys technology infrastructure, and collaborate for customized development and win–win solutions. In one case, an automated optical inspection (AOI) vendor collaborated on a new physical IT/OT plug-in or “media link” on the machine, so Ingrasys could deploy an in-house–developed AI inspection model that augmented the preexisting vendor solution. The vendor now offers this “bring your own AI” as a service solution to customers.
Data architecture
A solid data architecture works hand in hand with the technology backbone to empower talented people and help them work together in an agile way. The most advanced AI and analytics technologies emerging today thrive on massive, ever-expanding data sets. Front-runner organizations must therefore develop clear reference architectures and robust data pipelines to enable both business intelligence and machine learning solutions. Furthermore, they must have automated tools that actively support data quality and maintenance routines.
At CR Building Materials Tech, the data and analytics across 35 sites are powered by the Runfeng intelligent industrial internet platform, which sped up the deployment of digital use cases by 50 percent. The platform has four service layers, each making data accessible, accurate, and efficient. Within the application layer, an innovative microservices architecture improves tenant management, bolsters the system’s flexibility, and ensures maintainability. To ensure scalable storage and cloud management, the platform layer merges flexible configuration, enhanced deployment, and AI-driven analytics with a unified data ecosystem. The edge computing environment, which excels at managing industrial operations in real time, runs applications deployable across all 35 Runfeng sites (Exhibit 5).
Scale up with effective change management
Capabilities alone can’t drive a transformation: effective change management is needed to steer around potholes and roadblocks. Often, Lighthouses use transformation offices for this purpose; in fact, nearly 70 percent cite them as the most critical enabler for transformations. Their role is to track the impact of transformation programs, to provide appropriate financial incentives and proactive risk management, and to build critical digital skills and a strong culture of change across the organization. Many also support the active “assetization” of solutions for easy replication.
A networkwide outlook
Lighthouses have shown that to succeed with transformation efforts, companies must adopt a networkwide outlook right from the outset. Thinking too small won’t cut it. A pilot designed for site-level implementation might yield some benefits. But for a transformation to have a true business impact, companies must think, from day one, about at-scale deployment: building a strategy around a clear destination and then implementing a talent development approach that combines upskilling and new hiring to provide the capabilities needed. Coca-Cola, for example, used go-and-see programs, boot camp–style sessions, and training academies in its highly effective change management plan.
The collaborative ecosystem
The six enablers—including the four of the capability engine—work best when allowed to flourish in a collaborative environment. Innovation can and does happen in-house, but like a superior vehicle design, the best-performing capabilities harness the benefits of strategic partnerships. For industrial front-runners, this means building meaningful connections with universities, technology partners, and innovation centers. These not only are important sources of talent but also enable front-runners to share the knowledge and innovations that drive entire industries up the 4IR adoption curve.
This article series has focused on what the most advanced industrial manufacturers have done to take the world to the current inflection point of the Fourth Industrial Revolution—and where they are poised to take it next. By implementing the bold steps needed to innovate and climb the learning curve, Lighthouse organizations have shown what’s possible as they move beyond using 4IR innovations for site-level transformations and progress toward a network-level impact. Through a fervent commitment to capability building and a disciplined embrace of emerging machine intelligence technologies, Lighthouses continue to power up the doing part of the 4IR adoption curve, leaving the learning stage far behind. Next, they are poised to enter the optimizing stage and usher in radical levels of industry transformation at scale, effectively rewiring manufacturing across the value chain.
