July / August 2018

The Fourth Industrial Revolution

The Fourth Industrial Revolution

Dr. Enno de Boer says rapid, major changes are on the horizon for just about everything—including manufacturing.

“We are entering the Fourth Industrial Revolution.”

Dr. Enno de Boer, McKinsey & Company Partner and Leader of Digital Manufacturing, made this intriguing declaration during his keynote presentation at the 2017 ISPE Annual Meeting & Expo in San Diego, California. Focusing on three principal elements—intelligence, connectivity, and flexible automation—de Boer invited attendees to consider the future of manufacturing across a range of industries, including pharmaceuticals. The opportunities surfacing in this transformative era of technological evolution fueled tangible excitement surrounding his presentation. 

De Boer expanded on the game-changing implications of this Fourth Industrial Revolution for the factory of the future in a follow-up interview with Pharmaceutical Engineering. Using the advancements of robotics and artificial intelligence (AI), he predicted that manufacturing centers will “eliminate dull, dangerous, and dirty work” and become “much more human-centric, bringing the best that humanity can bring: creativity, problem solving, and the entrepreneurial spirit. These workplaces will attract the best and the brightest.”

Dr. Enno de Boer
Dr. Enno de Boer


These improvements in manufacturing will also provide direct benefits to the consumer. In the pharmaceutical industry, de Boer explained, “there are a couple of dimensions. We are able to have a much more agile and reliable supply chain, so if demand patterns change, production systems can adjust more quickly. If there’s any kind of epidemic, a more agile, flexible production system can produce the needed medication in a shorter time. Innovations are translated to product scale more quickly. In terms of mass personalization, we have the chance to do it at lower cost, in smaller batches, even to the production unit of one. We can tailor products to customers. And lastly, we see a productivity increase in management functions, quality, logistics, maintenance, and on the production line itself.” 

Describing this progress in his keynote, de Boer explained, “Technology is advancing faster than ever. The Internet of Things (IoT) will make us even more connected than we are now. Connectivity and artificial intelligence, along with flexible automation, will allow us to move from a reactive shop floor to an autonomous, self-organizing factory.” Consider this: With over 8.4 billion connected devices worldwide, there are already more connected devices than people. Fifteen percent of production assets are connected today. The over 700 IoT platforms on the market all aim to change this in the coming years. We are poised to see rapid, major changes to the way just about everything works, de Boer said—and that includes manufacturing.


Evidence of this accelerating innovation is visible all around us. We might consider Moore’s Law, attributed to Intel cofounder Gordon Moore, who asserted in a 1965 Electronics1 magazine article that computer processing power would double every  year. Ten years later, Moore revised his projection to every two years, and in 2015 suggested that the rate of progress would slow in the coming decade.2 Nonetheless, according to a 2015 essay in Scientific American,3 Moore’s prediction has largely held true over the years.

De Boer underscored this rapid development, emphasizing that the last 36 months have seen incredible leaps in computing power. “During this time, we have become able to train our AI models 60 times faster at less than 1% of the previous cost. Our models are constantly improving, and AI speech recognition is already at the level of the human brain.” De Boer explained that in 2010, the error rate for computerized speech recognition systems was 27%, versus only 5% for humans. Three years later, the error rate for computers had dropped to 20%. By 2015, the rate was down to less than 6%, nearly matching that of humans. When it comes to image recognition, the advancement has been even more rapid. While the human error rate has held steady at 5%, the error rate for computerized systems has dropped quickly, from 28% in 2010, to 11% in 2013, to less than 5% in 2015—actually surpassing human performance.

Talking with Pharmaceutical Engineering, however, de Boer noted the difference between the existing potential for automation in manufacturing and how little has actually been realized. To explain the gap between the exponential growth in computing power and the more incremental growth we’ve seen in robotic systems, he offered this insight: “Moore’s Law applies to computer power, but robotics are much more physical systems. I’m still bullish on how robotics and automation will grow; the business case is sound. Up to 60% of tasks in manufacturing today can be automated. We have automated only a small fraction of that. There will be a race to close that gap.” 

That race may be played out on an innovation superhighway, according to de Boer—a technology infrastructure “autobahn” that is ready to be traveled at remarkable speeds. The obstacles to realizing this potential aren’t a matter of physical or computing infrastructure; rather, they lie with “people and systems.” The figurative autobahn is there, he explained. Manufacturers just need to learn to drive on it.


Augmented reality
Augmented reality helps operators view instructions, safety information, and data in real time. (Photo courtesy McKinsey & Co.)


Unfortunately, de Boer continued, “the majority of companies are in ‘pilot purgatory,’” a concept detailed in “The Next Economic Growth Engine: Scaling Fourth Industrial Revolution Technologies in Production,” a January 2018 World Economic Forum white paper4  he coauthored with Helena Leurent, a member of the World Economic Forum’s executive committee. In that report, the authors present challenges to adoption of technology, focusing on matters of “people and systems” as well as issues such as “lack of knowledge,” “lack of trust in scalability,” and “lack of leadership support.” The list is conspicuously devoid of problems related to any lack of available technology infrastructure, de Boer confirmed. 

And so the problem isn’t that the potential isn’t there, de Boer said—it’s that companies get hamstrung in the pilot phase and struggle to scale. “If you stay incremental, you’ll never get exponential growth. To get to Moore’s Law, you need to put the right scale-up engine behind it—scalable in terms of your data models. You need a scalable analytics model, a scalable people model—then you literally have the autobahn. Keep in mind this is the Fourth Industrial Revolution. It’s really about change management.” 


“Deep learning” is another broad technological breakthrough that started with machine learning in the mid-20th century. But deep learning is something radically different, said de Boer. “It involves learning based on pattern recognition rather than task-specific algorithms. With that, you train the model. The more data the model looks at, the more accurate the data; the more data the models ingest, the more they drive up accuracy. Keeping in mind that 90% of all data has been generated since 2015, it becomes clear why machine learning algorithms have become so much better.” Given this training, the algorithms actually improve themselves over time—assuming they have a steady stream of sufficient data. “That’s why it’s so interesting in manufacturing,” he added. “In terms of sectors, manufacturing is sitting on the biggest amount of data.”

Rapid as it may be, this processing power can’t come soon enough. Humanity has amassed a mind-boggling quantity of stored digital data, and the cloud is growing at a blistering rate. According to the Cisco Global Cloud Index,5  “globally, the data stored in data centers will nearly quintuple by 2021” from 2016 levels. 

De Boer’s keynote presentation emphasized the disparity between the vast (and growing) accumulations of production data and the tiny percentage used effectively in decision-making. Using an offshore oil rig example to illustrate this gap, he explained that of all the data captured by the rig’s various systems, only 40% is stored using storage infrastructure. Certainly, that’s a rapid drop-off, but the plunge is even more precipitous considering his explanation that only 1% of the data is streamed onshore through data management. Thus, only approximately 1% is monitored post hoc in the form of key performance indicators, and even less is sent back to the rig in the form of analytical insights. So how much is that, ultimately? “It’s one-half of 1% of all data that’s really helpful,”7  de Boer emphasized in his follow-up interview. “That will change dramatically with artificial intelligence.”

Only one-half of 1% of all data is helpful
McKinsey Global Institute. Reprinted with permission.
Only one-half of 1% of all data is helpful


All signs point to a paradigm shift in manufacturing. De Boer explained that more than half (51%) of all tasks globally can now be automated, and in the manufacturing sector the figure is higher, at 60%. To illustrate just how well the stage is set for robotic automation, de Boer offered a breakdown of how time is spent on various aspects of the manufacturing process, and then indicated to what degree each could be automated using current technology. 

His numbers, drawn from a McKinsey Global Institute Analysis, are striking. “Predictable physical” tasks comprise 34% of the manufacturing process, and de Boer stated that fully 87% of those tasks could be automated with today’s technology. Nearly 78% of data collection, which represents 22% of the manufacturing process, could be automated; likewise, 60% of data-processing tasks, which comprise 11% of the manufacturing process, could be automated. Of course, some elements are less adaptable: managerial aspects, expertise, interface, and unpredictable physical applications will see less automation. Nonetheless, the aggregated data suggest that more than half of the entire manufacturing process—fully 60%—could be automated now, according to de Boer. And that’s just based on current technology, which means, for all the reasons laid out above, that those potentialities will continue to increase—likely exponentially. The good news is that only 1% of all jobs can be fully automated, while the majority of today’s jobs will be augmented by automation.

Despite this substantial potential for automation, industrial robots currently have a surprisingly low penetration rate—less than 5%—in global manufacturing. According to the same aggregated data, only 180 industrial robots are at work per 10,000 manufacturing workers in the United States. That figure nearly doubles in Germany and Japan, where the ratio is 300 robots per 10,000 workers; in South Korea, it’s nearly triple—over 600 industrial robots for every 10,000. Given the anticipated rapidity of technological development, it stands to reason that these numbers will increase quickly.

This won’t come without consequence, of course, and it’s important to consider the potential difficulties inherent in this adaptation process. In a LinkedIn piece titled “How Technology Can Unlock Manufacturing’s Potential and $3.7 Trillion in Global GDP,”6 de Boer offered perspective on some inevitable effects of automation: “Let’s be frank: Turning factories into high-tech platforms will displace a significant number of workers. We cannot ignore the social and humanitarian consequences of automation. Governments, businesses, and civil society must take the lead in easing the transitions of workers by upholding social compacts and equipping current and future workers with the training and education they need. All sectors will have to reinvest in local economies and in new areas of growth.”

Enno de Boer speaks at the ISPE 2017 Annual Meeting, San Diego, California
Enno de Boer speaks at the ISPE 2017 Annual Meeting, San Diego, California


So where is manufacturing headed? “The future of manufacturing will be fundamentally different from what we have today. At the core we will have autonomous manufacturing—similar to autonomous or self-driving cars: a manufacturing shop floor that is self-organizing, self-optimizing, and self-healing. We will see the convergence of products and services. New value-added services and business models will be enabled,” said de Boer. 

Some of the technologies that have become highly relevant for making manufacturing autonomous are already in use and have matured at scale in cars, he noted. 

It is interesting to see how cars have become more and more autonomous over time. Early in the millennium (c. 2005), the automotive industry started introducing “assist systems” using technologies such as GPS/location services, sensors, video recognition, and connected control systems. By 2015, “automation” functionality was introduced, leveraging 4K video, pattern-recognition machine learning, predictive algorithms, augmented reality (AR) displays, and IoT-based vehicle-to-vehicle communication. The next phase, which started with semi-autonomous cars and is currently moving toward fully autonomous cars, involves big-data analytics, deep-learning algorithms, and vision systems that have been further improved. 

How will this Fourth Industrial Revolution play out? According to de Boer, the innovation will be realized through a series of paradigm shifts, each interwoven with the others amid an interplay of digital systems. With regard to manufacturing, he explained that technology will transform the future of production through four shifts:

  1. Autonomous manufacturing, embedded into an …
  2. End-to-end value chain, embedded into a …
  3. Supplier ecosystem, which enables …
  4. Value-added services and business-model innovations.

The key to this evolution, then, lies in the development of autonomous manufacturing. According to de Boer, this involves an interplay of connectivity, intelligence, and flexible automation. 


Connectivity will enable a new level of performance management and radically more efficient assembly operations. Through AR-guided assembly operations and real-time IoT-based performance management, manufacturing systems will see massively enhanced scale and speed. To illustrate this, de Boer offered data from successful deployments of intelligent factory systems: While the current performance-management systems enable the capture of one billion data points per day, fewer than 1% are able to be analyzed. By contrast, with real-time IoT-based performance management in place, systems can capture 10 billion data points daily, with the capacity to analyze 100% of it. Likewise, in terms of speed, data accessibility in current systems is measured in hours and days, whereas IoT-based functionality will measure this accessibility in minutes and seconds.

This connectivity will enable a new intelligence landscape, enabling predictive forecasting and digital supplier collaboration. Considering the immediacy of the “digital thread” that connects all aspects of the manufacturing process—from suppliers through manufacturing to dealers and on to customers—real-time analytics will transform any number of industries and provide notable benefits. Among these are reduced inventory and lead time at all steps, improved production planning (one example yielded a 33% better forecast accuracy), and step-change improvement in collaboration. All of this, according to de Boer, will significantly increase customer satisfaction. 

Ultimately, manufacturing will exist in a radically transformed end-to-end value chain, in which real-time data access enables intelligent, responsive performance management with substantially reduced waste and highly improved efficiencies. “Ecosystem” is an apt description for the potential reality de Boer describes: an interwoven, interdependent, highly evolved environment wherein each element of the end-to-end chain is connected efficiently and thus able to respond to changes organically and in real time. “This is where the value creation is happening,” de Boer said. “And how will we get there? The same way as the automotive industry. We have identified 40 digital applications that are ready for deployment. These are the stepping stones toward an autonomous production system.”

This reality is already emerging, and de Boer referenced two unnamed electronics manufacturers to illustrate the impact of these systems. One of these companies, located in China, saw issue resolution go from weeks and months to hours upon deployment of real-time IoT-based performance management. The other company, by implementing an almost fully automated assembly line, has realized a 200% increase in output and a 50% reduction in quality issues. With these improvements also came the ability to move production nearer to the consumer. 


Of course, while this Fourth Industrial Revolution is rooted in technological innovation, its implementation hinges on human leadership. Commitment to organizational change is an altogether human enterprise and requires a purposeful and collaborative effort from industry leaders and workers alike, from senior management to the shop floor. “It’s very important to not just exchange technology, but radically redesign the system,” commented de Boer. Achieving this “digital transformation” requires an intentional approach, which McKinsey has developed in collaboration with the World Economic Forum. Steps include:

  1. Mobilizing the organization
  2. Strategizing—setting the vision
  3. Sparking innovation by demonstrating the value
  4. Scaling up to achieve a full-value capture

Considering the rapid pace of technological development, time is of the essence. Organizations must respond to this changing landscape, de Boer added. “Technology fusion must happen quickly.” 

To facilitate this, a great deal of learning is required. “For this we need training,” he continued. “Over the last 12 months we have invested heavily in setting up capability centers—smart model factories that allow you to jump into a manufacturing environment.” He described the five newly opened McKinsey Digital Capability Centers, which, like the one in Chicago operated in collaboration with the Digital Manufacturing and Design Innovation Institute, offer “digital immersion workshops.” The workshops enable participants to explore experiential learning modules based on the production line. Meanwhile, they empower organizations with access to “an ecosystem of 50-plus technology partners, providing innovating solutions across the value chain.” In addition, the McKinsey “digital blueprint”7 helps management understand how to start, scale, and sustain the digital journey. 


While de Boer’s keynote presentation addressed digital technology across the manufacturing spectrum, he shared insights on the unique context of the pharmaceutical industry during our conversation. A significant factor at work in pharmaceuticals, he said, is the degree to which innovation needs to go hand in hand with regulatory compliance. Therefore, he noted, “It’s important to work in consortiums like [ISPE] to determine what you want to change in the processes and bring the regulators along with you. The disruption will happen—probably from the outside. So you have to be prepared for the future.”

Explaining further, de Boer stressed that for innovation to proceed optimally, manufacturers and regulators must “travel together on the innovation journey.” A partnership approach will facilitate travel on that figurative autobahn. “Technology will only unfold if we innovate the processes. We need collaboration. It’s not one or the other.” Referring to the interest in innovation that he’s seen among regulatory bodies, de Boer claimed, “Actually, regulators are for innovation. I see a lot of willingness on all sides of the table.” 

Digital innovation is constantly accelerating. We need only consider the fascinating technologies at work in our lives today on even the most pedestrian level—smart speakers and appliances, predictive advertising, voice recognition, intelligent and integrated navigation services, mobile boarding passes, and digital wallets. These are realities we now take for granted, and it’s easy to forget that they weren’t widely available—or, in some cases, available at all—just a few years ago. This same dynamic evolution is at work in the industrial sector, and if de Boer’s vision of the Fourth Industrial Revolution plays out as he’s described, the future of manufacturing is upon us.

As we concluded our interview, de Boer reiterated his enthusiasm about the near future of manufacturing: “I think the one thing I would really like to emphasize is this ‘facility of the future,’ which is an exciting workplace. There is a new workforce coming in, and we can make the work in the factories very attractive. Where else in the world will all these systems come together to drive change? We need to—and we will be able to—attract the best and the brightest of the next generation.” 

While de Boer’s vision certainly implies new challenges for the pharmaceutical industry, it also points to potentially groundbreaking levels of efficiency and innovation. Ultimately, we can hope, this evolution in manufacturing will yield radical advances that benefit both patients and workers.