FIGURE 7.2
Second, the data generated are valuable. As organizations sensor and measure areas of their business, those sensor readings help them make better, more profitable decisions. The data generated by IoT, when analyzed by AI, will—and does—enable organizations to better run core business processes. This is true not only in the utility industry but also in oil and gas, manufacturing, aerospace and defense, the public sector, financial services, health care, logistics and transportation, retail, and every other industry I’ve seen.7
The third reason IoT will transform business is the power of Metcalfe’s Law—i.e., the value of a network is proportional to the square of the number of its members.8 In this case, the network is a company’s federated data image, and the members are the data points. With the proliferation of sensors in an enterprise’s value chain, there is a proliferation of data, both in volume and variety. More data, more value.
Consider an aerospace company looking to implement an AI-driven predictive maintenance solution for its fleet of jet aircraft. Unpredicted equipment failures mean less flight time—clearly a problem for any fleet operator. Most jets already have a wide array of onboard sensors, but many of the readings are not used to predict when maintenance is required. As a result, most organizations today rely on time-based maintenance—i.e., scheduled at predefined time or usage intervals—which leads to over-maintenance and cannot adequately predict when equipment requires service. Wasted resources and costs are the result.
Imagine an alternate approach that takes advantage of all data generated by the aircraft systems. Start with one data source: the jet engine forward vibration sensor. On its own, this does not provide enough data to build a comprehensive predictive maintenance application for the entire jet. It may not even provide enough data to predict jet engine failure with an acceptable level of accuracy. But imagine we use data from 20 sensors on each engine, gathered from every engine across a thousand-jet fleet. Using AI machine learning algorithms on this significantly larger and richer data set, we can predict engine failure with much greater accuracy, resulting in reduced aircraft downtime and more efficient use of maintenance resources—which clearly translate into economic value.
Now consider ingesting and analyzing sensor data from all the jet systems and components, not just engine sensors: We can now predict failure for every part of the aircraft—the engines, the fan system, the landing gear—before it happens, because we have sensor data for each part. But this is not the only implication. Because the aircraft is a unified system, the components—and the data—are interrelated. This means data generated by the jet engines may be predictive of failure rates for the fan system and vice versa. This is true for every pair of components. In this way, the value of the data increases exponentially with the volume and diversity of data.
By ingesting more data from more devices and increasing the richness and volume of our data set, the accuracy of our predictive maintenance application is now much higher—and more effective and cost efficient—than traditional methods. Additionally, predictive maintenance paves the way for more efficient management of inventory parts and supply chain operations—a compound benefit. Returning to the old, scheduled aircraft maintenance operations is inconceivable. In much the same way and for the same reasons—data volume, value, and Metcalfe’s Law—IoT will transform business processes across industries.
As another example, IoT has significant impact on agriculture. A potato farmer in the Netherlands now runs one of the world’s most advanced potato farms because of IoT.9 Multiple types of sensors on his farm—monitoring things like soil nutrients, moisture levels, sunlight, temperature, and other factors—provide large amounts of valuable data, enabling the farmer to use his land more efficiently than other farms. By connecting every piece of the farming process through IoT, he knows exactly which parts of his land need more nutrients, where pests are eating leaves, or which plants are not getting enough sunlight. Equipped with these insights, the farmer can take the right actions to optimize his farm’s production.
The business world is in the early days of capturing the value that IoT—in conjunction with cloud computing, big data, and AI—can unlock. Our Cambrian Explosion of IoT is still ahead of us. But one thing is clear: IoT will change business in a major way. The question remains how. I argue it will profoundly change three fundamental aspects of business: how we make decisions, how we execute business processes, and how we differentiate products in the marketplace.
First, decision-making will change, with data-driven decision-making in particular taking on an entirely new meaning.10 Algorithms will become an integral part of most, if not all, decisions. This is particularly true for the day-to-day decisions that keep a business running. Think of decisions made on the factory floor, inside a fulfillment warehouse, or even in a bank’s lending division. With product usage information, equipment health data, and environmental measurements, problems can be assessed in real time and recommendations can immediately be surfaced to operators. This means less reliance on simple but suboptimal “rule-of-thumb” practices. It also means less reliance on operational expertise. Human expertise is only required when the AI-produced output seems off-base. If it is, the system can learn from human intervention to better address similar cases in the future. This means fewer staff, less human involvement, and superior business results. Sensored value networks enable fact-based, AI-driven predictive decision-making.
Second, IoT will change how business processes are executed, resulting in faster, more accurate, and less expensive decision-making. Instead of consulting one’s own intuition and experience and doing “what feels right,” operators will consult an algorithmic recommendation that clearly explains why it suggests a certain course of action. The onus will be on employees to override the system, but this will happen only in a fractional number of cases. Employees will be freed up to focus less on operational minutiae and more on adding strategic and competitive value.
Third, IoT will change the way products are differentiated in the marketplace. We will see a new level of individualization of product behaviors. Smartphones already adapt to how their owner speaks or types. Smart thermostats learn the temperature preferences of residents and automatically accommodate them. In health care, intelligent glucose monitors equipped with algorithms can automatically adjust insulin delivery via an implanted pump.
This is just the beginning—IoT changes the relationships we form with physical objects. IoT gives manufacturers unprecedented visibility into how customers use their products. Not only does this enable companies to know their customers better and thus make better products, but it also enables new warranty and equipment rental models—such as guaranteeing products under certain usage constraints and deactivating rented items when the customer stops paying a subscription fee.11 These models may not seem appealing to end users at first glance, but they can radically change the economics of owning or renting certain products, and customers absolutely do respond to better pricing models. IoT has unlocked new possibilities in decision-making, operations, and product differentiation, and it will continue to do so—often in ways we have not yet imagined.
How much will all this IoT-driven change impact the economy? With the total number of connected devices projected to grow from about 20 billion today to 75 billion by 2025,12 analysts expect IoT will contribute up to $11.1 trillion in annual global economic value by 2025.13 That is a staggering amount, equivalent to approximately 11 percent of the global economy, based on the World Bank’s projection of $99.5 trillion in global GDP in 2025.
FIGURE 7.3
Significant workforce displacement will be a byproduct of IoT adoption and the corresponding automation it enables. I expect the level and timing of job displacement to vary dramatically across industries, but the aggregate statistics are undeniable.
Almost half (47 percent according to the Economist) of American jobs are at risk due to automation, and a substantial portion of this automation is due t
o IoT. In Britain and Japan, the numbers are similar: 35 percent and 49 percent of jobs are at risk, respectively.14
Individual firms also weigh the impact of automation. UBS CEO Sergio Ermotti predicts that automation caused by new technology adoption could cause the company to downsize by 30 percent. At UBS alone, that is almost 30,000 employees.15 Former Deutsche Bank CEO John Cryan predicted the firm will cut its 97,000-person organization in half due to automation.16 Goldman Sachs estimates that self-driving cars could destroy 25,000 driving jobs every month in the U.S. alone.17
However, job displacement doesn’t mean people will no longer work. New jobs will emerge even as traditional jobs disappear. As I noted in chapter 6, advanced technologies will create more jobs than they will eliminate, and this change will happen quickly.
FIGURE 7.4
New opportunities will abound for those with immediately relevant skill sets and new types of roles will emerge that we cannot even imagine today. In 2018, 7 of the top 10 fastest-growing job titles on LinkedIn were data science and engineering roles.18 This growth in data science will continue for the foreseeable future: In 2020, there are expected to be about 700,000 job openings for data scientists and similar roles in the U.S.19 Similarly, operational roles for those who manage IoT devices of all sorts, as well as new types of IT, networking, and telecommunications jobs, are likely to grow. These high-value jobs will often be cross-disciplinary, grounded in business and technical knowledge.
There is every reason to be optimistic. But as we’ve noted, employers, governments, and schools will need to train and retrain millions of people for these new jobs. Millions of existing workers will have to find new jobs. I believe we in the commercial sector have a responsibility to promote training and education for these new roles. IoT, when paired with AI, is causing a structural shift in the employment landscape. IoT and the technologies it enables will affect our world at a scale that is difficult to overstate.
How IoT Creates Value
A broad and growing range of use cases is driving the tremendous impact of the internet of things. All these involve different parts of the technology stack, ranging from the actual hardware of connected devices to services, analytics, and applications. From a customer’s or end user’s perspective, IoT’s real value comes from services, IoT analytics, and applications, while the rest of the technology stack serves as an enabler with lower value and growth potential.20 Ultimately, organizations using IoT technologies (factory owners, operators, manufacturers, etc.) will capture most of the potential value over time.
For business leaders to adopt IoT solutions, they need to know how these offerings will add value to their organization by solving critical business challenges: reducing asset maintenance costs, optimizing inventory, increasing revenue through better demand forecasting, increasing customer satisfaction and product quality, and more.
By focusing on these concrete business problems, IoT offerings can be rolled out rapidly across industries and gain widespread adoption. In the following section, I will discuss some of the most promising use cases and how they can add value to businesses.
IoT Use Cases
Smart Grid
As previously noted, the utilities sector was one of the first industries to make use of IoT on a large scale. By deploying millions of smart meters across their operations, utilities have created what we today call a smart grid.
Enel, the large utility based in Rome, manages more than 40 million smart meters across Europe. These meters generate an unprecedented amount of data: more than 5 billion readings per day. IoT phasor measurement units (PMUs) on transmission lines emit power quality signals at 60 Hz cycles (i.e., 60 times per second), with each PMU generating 2 billion signals per year.
The combination of inferring the power consumption, production, and storage capacity of each customer in real time (using cloud computing and AI)—in conjunction with the network effect of interconnected customers, local power production, and storage—is the essence of the smart grid. The more connected sensors and the more data available for analysis, the more accurate the deep learning algorithms will be, resulting in an increasingly efficient smart grid. Recall Metcalfe’s Law: Increasing the number of connected sensors drives exponential value in the network. All this sensor data provides near-real-time information on the condition of the grid—its status, equipment issues, performance levels, etc. This allows the algorithms to adapt their predictions and recommendations in near real time. Enel estimates the application of AI algorithms on all this smart grid data across its entire network will yield more than €600 million in annual economic value.
Predictive Maintenance
Companies can apply AI to data captured with IoT technologies—sensors, meters, embedded computers, etc.—in order to predict failure of equipment before it happens. This reduces unplanned downtime and enables flexible work schedules that extend equipment life and drive down service labor and parts replacement costs. A wide range of industries—from discrete manufacturing, energy, and aerospace to logistics, transportation, and health care—will be able to capture these benefits.
Consider, for example, Royal Dutch Shell, one of the world’s largest energy companies with 86,000 employees operating in more than 70 countries, and annual revenue of more than €300 billion. Shell is developing AI applications to address numerous use cases across its global operations, which span a vast number of assets throughout more than 20 refineries, 25,000 oil and gas wells, and over 40,000 service stations. Several AI applications are live in production, with numerous others to follow.
In one use case, Shell developed and deployed an application to predict failure of hydraulic power units (HPUs)—which are critical in preventing well blowouts—for 5,000 gas wells at its Australian QGC operation. The AI-powered application ingests high-frequency data from multiple sensors at these remote, hard-to-access well sites in order to predict both when an HPU will fail and the root cause of the issue. This enables maintenance teams to proactively address issues before failure occurs. The demonstrated benefits of the application include higher asset runtime, optimized resource utilization, reduced operational costs, and millions of dollars in realized annual revenues.
In another example, Shell developed and deployed an application that predicts performance deterioration for more than 500,000 valves operating at refineries around the world that produce gasoline, diesel, aviation fuel, lubricants, and other products. Valves of multiple types are critical components in controlling the flow of fluids throughout refineries. A valve’s performance deteriorates over time based on multiple factors, including the type of valve, its historical usage, exposure to heat, pressure, rate of fluid flow, etc. Multiple sensors continuously record various measures of a valve’s operation and status.
For this use case, Shell automated the training, deployment, and management of more than 500,000 AI models—one for each individual valve. The application ingests more than 10 million sensor signals at very high frequency; applies valve-specific AI models to predict deterioration; and prioritizes the most critical valves for operator attention. The application enables Shell’s operations to shift from being reactive and rule-based to predictive and prescriptive. Based on these changes, which reduce maintenance costs and increase operating efficiency, deployment of the application for this use case alone is estimated to deliver several hundred million dollars in annual economic value.
Inventory Optimization
From manufacturing to consumer packaged goods and many other industries, companies across the world struggle to get inventory planning right. In fact, many large organizations with complex supply chains and inventory operations miss their “on time in full” (OTIF) delivery targets 50 percent of the time when employing inventory optimization solutions from traditional enterprise resource planning (ERP) providers. IoT-based solutions—combined with AI-based big data analysis—can help companies dramatically improve their OTIF achievement rates, significantly increasing processing speed while red
ucing response times, stock-outs, and inventory pileups.
For example, a $40 billion U.S. discrete manufacturer of large complex machines uses an AI-powered inventory optimization solution to balance optimal inventory levels and minimize inventory costs. The company’s machines select from 10,000 different options and as many as 21,000 components in each bill of materials. Unlike traditional ERP-based inventory solutions, the AI application applies advanced stochastic, AI-based optimization on top of the company’s large IoT-based data set. As a result, this manufacturer is able to reduce inventory costs by up to 52 percent on its $6 billion holding inventory—freeing up $3 billion of capital that it can deploy and invest elsewhere.
Patient Care
The potential for IoT in health care is vast. IoT gives doctors the opportunity to track patient health remotely in order to improve health outcomes and reduce costs. By harnessing all these data, IoT supports doctors in predicting risk factors for their patients.
For example, pacemakers are a kind of IoT sensor—they can be read remotely and can issue alarms to doctors and patients, warning if a heartbeat is irregular. The wearable industry has given people the ability to easily track all sorts of health-related metrics: steps taken, stairs climbed, heart rate, sleep quality, nutrition intake, and so on.
Digital Transformation Page 15