Computers are useless. They can only give you answers.
Introduction
It is common knowledge that a number of major firms such as Amazon, Google and Uber have transformed industries using digital technologies. All of these firms are founded on business model innovations: offerings that involve systemic changes to how value is created and captured (Appio et al., Reference Appio, Frattini, Petruzzelli and Neirotti2021; Bharadwaj et al., Reference Bharadwaj, El Sawy, Pavlou and Venkatraman2013). There are many newspaper articles and blog posts providing advice to managers on how to deal with the opportunities and challenges created by digital technologies. Digital technologies can offer new forms of value creation while posing a threat to conventional means of customer value propositions (Aversa et al., Reference Aversa, Formentini, Iubatti and Lorenzoni2020; Bresciani et al., Reference Bresciani, Huarng, Malhotra and Ferraris2021; Velu, Reference Velu2016). Digital technologies can help to transform organisations, opening up possibilities for new firms to emerge while destroying the viability of others (Berman, Reference Berman2012; Velu et al., Reference Velu, Pooya and Dalzell-Payne2019; Berman and Dalzell-Payne, Reference Berman and Dalzell-Payne2018; Broekhuizen et al., Reference Broekhuizen, Broekhuis, Gijsenberg and Wieringa2020). Established firms who are considering implementing digital technologies need to decide how to do so, as it can profoundly alter their prospects for the future.
Today, several digital technologies and critical infrastructures, such as AI, cloud computing, additive manufacturing and 5G, are developing rapidly and may transform the economic landscape in which firms compete. The transformation may be compared to several general-purpose technologies from the past, including steam and electricity (David, Reference David1990; Rosenberg and Trajtenberg, Reference Rosenberg and Trajtenberg2004; Velu, Reference Velu, Nambisan, Lyytinen and Yoo2020; Dosi, Reference Dosi1982).Footnote 1 Electric motors were introduced around 1879 to replace steam engines, yet productivity initially declined slightly up to 1920, before beginning to rise rapidly (David, Reference David1990). The rapid productivity growth in US manufacturing may be attributed to new organisational structures and business models that only emerged in the 1920s. The new business models involved putting multiple electric motors where they are needed and leasing them from external firms with specialist support services. This enabled productivity growth through lower energy consumption, improved production flows and greater resilience.
Studies suggest that there are multiple stages or phases to digital change, ranging from relatively simple to comprehensive changes (Verhoef et al., Reference Verhoef, Broekhuizen, Bart, Bhattacharya, Dong, Fabian and Haenlein2021). These phases can be classified into three broad categories: digitisation, digitalisation and digital transformation. Digitisation is the encoding of analogue information into a digital format (i.e., into zeros and ones) such that they can be stored, processed and transmitted by electronic means (Yoo et al., Reference Yoo, Henfridsson and Lyytinen2010). Digitalisation refers to how digital technologies can be used to alter existing business processes (Pagani and Pardo, Reference Pagani and Pardo2017). Digital transformation refers to a more comprehensive change across the firm that contributes to new business models (Iansiti and Lakhani, Reference Iansiti and Lakhani2014).
Digital technologies enable products and services to be delivered electronically and hence provide new value propositions to customers (Andal-Ancion et al., Reference Andal-Ancion, Cartwright and Yip2003). Moreover, digitally enhanced products and services encode much more information and are significantly more customisable. Hence, firms could better engage with customers and also obtain information about their processes on a real-time – or near real-time – basis. Digital technologies enable a more integrated set-up, both within and across firms. This enables the disintermediation of some existing firms but reintermediation with different roles, depending on the flow of material and information required to develop and deliver propositions to customers. Such an integrated approach implies that firms need to change their business models from primarily “make and sell” products and services to “sense and act” based on the information flow between the customer and firms across the network (Koebnick et al., Reference Koebnick, Velu and McFarlane2020). Hence, digital transformation can be challenging for firms. In this chapter, we provide an overview of the theory behind digital technologies and the implications for digital transformation. We then discuss some of the recent developments in digital technologies based on developments in artificial intelligence. Following that, we preview a new and emerging general-purpose technology based on quantum mechanics that is likely to change the landscape of computing and the Internet in years to come, as well as provide a basis for major digital transformation opportunities. Finally, the conclusion section provides some of the societal challenges related to business model innovation in the digital economy.
4.1 The Evolution of Digital Technologies and Transformation Challenges
Information-technology-driven competition that has shaped strategy can be classified into three waves (Porter and Hepplemann, Reference Porter and Heppelmann2014). The first wave, in the 1960s–1980s, involved the automation and digitisation of individual activities across the value chain, such as order processing, procurement and manufacturing resource planning. This involved standardising processes, and the challenge for firms was how to differentiate effectively. The second wave came with the rise of the Internet in the 1990s and 2000s, which enabled the integration of activities across suppliers, customers, channels and geography, enabling more globally distributed supply chains. In the third wave, which is happening today, IT is being embedded into products and hence changing the nature of products and services. Products with embedded sensors, increased connectivity and cloud-based data storage and analysis enable new product features and service functionalities. This could potentially reshape the value chain dramatically. There is a good example of this from the agricultural context (Porter and Hepplemann, Reference Porter and Heppelmann2014). Imagine a tractor that is used on a farm to move earth and cultivate crops. Such a tractor could be embedded with sensors, which inform the farmer about its performance and how to optimise its use. In the next stage, the tractor could be connected to other farm equipment such as tillers, planters and combine harvesters. This would make it a product system; hence, the basis of competition shifts away from being a product to a product system. The manufacturer of the tractor could alter its business model from being a farm equipment manufacturer to farm equipment optimisation as a service proposition. The next stage could be expanding beyond a product system to a system of systems proposition, whereby the connection is not only to farm machinery but to irrigation systems, weather, crop prices and seed-management systems. Moreover, the connected processes enable increased granularity in segmenting customer needs and require enhanced flexibility of value-adding activities to meet these needs. Such features will require firms to transform their business models from the traditional “make-and-sell” into a “sense-and-act” business model (Koebnick et al., Reference Koebnick, Velu and McFarlane2020). Such a connected system of systems proposition blurs industry borders and throws up major challenges for firms in designing the appropriate business models to create competitive advantage.
Smart products could change how consumers consume the product from buying outright to pay-per-use. Alternatively, the firm could go into a hybrid model, whereby product ownership is transferred to the user but there is a performance-based service contract to ensure that the product performs to a certain specification, such as uptime (Martinez et al., Reference Martinez, Neely, Velu, Leinster-Evans and Bisessar2017). This fundamentally changes the incentive structure of firms. For example, Rolls-Royce traditionally might have had a business that gains from selling spare parts and service contracts on its aircraft engines (Grubic and Jennions, Reference Grubic and Jennions2018). However, such a business model might lessen the incentives to design engines that are reliable and durable. Hence, following pressure from its customers, the firm decided to include sensors in its aircraft engines and to sell a service contract by the hour, known as power-by-the-hour, which pays based on the hours that the engine is in the air (Smith, Reference Smith2013). This is often known as servitisation (Baines and Lightfoot, Reference Baines and Lightfoot2013; Martinez et al., Reference Martinez, Neely, Velu, Leinster-Evans and Bisessar2017; Smith et al. Reference Smith, Maull and Ng2011; Neely, Reference Neely2008; Neely et al., Reference Neely2005). Such a trend is happening in many industries, from personal home and office space to personal and public transportation systems. For example, automotive firms such as Scania, a Swedish commercial vehicles manufacturer, have connected 500,000 vehicles on the road to manage the optimisation of routes, predictive maintenance and more efficient staffing of trucks as a service (Bjorkdahl, Reference Bjorkdahl2020). However, the generation of technological shifts from product systems to systems of systems might prove challenging for truck companies. The truck industry is facing a technological shift from diesel engine powertrains to electric road system (ERS) powertrains, whereby there is continuous electric power transfer to trucks via the road surface (Tongur and Engwall, Reference Tongur and Engwall2014). Although, from the customers’ perspective, there might be little change, this represents a radical change from the truck manufacturers’ perspective. Such ERS powertrain systems will be more integrated into other subsystems, such as roads, fuel supply and billing systems, than the diesel-engine-based powertrain system. Truck manufacturers’ core capabilities need to be transformed from engine technology to knowledge of customer behaviour, logistics and fleet-management services as customers adapt to wanting transportation services. Hence, the challenge for firms is to decide the scope of the firm, what core competency it would need and who it needs to partner with to enable complementary assets to come together to provide the product and service offering to the customer (Leonard-Barton, Reference Leonard-Barton1992).
4.2 Theory of Digital Technologies
Digital technologies are increasingly prevalent in enabling new sources of value. The Von Neumann architecture of digital technologies enables the instructions to manipulate the data and the data itself to be stored in the same format and location, which enables reprogrammability (Langlois, Reference Langlois2007). Digital technologies enable the decoupling of information from its related physical form or device, which has been termed resource liquefication (Normann, Reference Normann2001). This enables digital technologies to have two features. First, there is the separation of content from the medium (Yoo et al., Reference Yoo, Henfridsson and Lyytinen2010; Faulkner and Runde, Reference Faulkner and Runde2013). For example, in the physical world, the content of a book in terms of, say, its story is tied to the physical copy of the book. However, with the digital medium, the content of the book can be read or listened to in various ways, such as the phone, computer or tablet, among others. The second feature is the separation of form and function. For example, a private car has often been used as a personal vehicle, and hence the form and function are relatively fixed. However, with the advent of Uber, the personal car is being used as a taxi and reverting back to being used for private purposes. Hence, digital technologies enable the procrastinated binding of form and function, whereby new capabilities can be added to a proposition once they have been designed and created (Zittrain, Reference Zittrain2006). These, in turn, provide new ways to recombine resources and enable generativity, the capacity to produce unprompted change and innovation from uncoordinated and heterogeneous audiences (Yoo et al., Reference Yoo, Henfridsson and Lyytinen2010).
Moreover, digital technologies can create and shape physical reality, a phenomenon called ontological reversal (Baskerville et al., Reference Baskerville, Myers and Yoo2020). Typically, the physical reality takes precedence over the digital reality. For example, as one drives, the blue dot representing the location of the car on Google Maps follows the physical location of the car. Ontological reversal happens when the digital version of an object is created first, before the physical version. For example, in the case of autonomous cars, the blue dot that represents the car is created first and is computed to move from point A to point B on Google Maps. The real car then moves based on the movement in the non-material digital blue dot. Hence, the non-material digital object,Footnote 2 the blue dot, comes before the material real object, the autonomous car. In this way, digital reality takes precedence over physical reality and the ontology is reversed. This ontological reversal could have profound implications for human experiences and how value is created and captured. The vertical and horizontal scope of the firm, in terms of the activities within its hierarchy, as well as across the network of the value chain, will determine the boundary of the firm. These features of digital technologies enable new combinations of digital and physical assets to redefine the boundaries of the firm.
The transaction costs theory of the firm helps to provide an initial framework to understand the boundary of the firm. Transaction costs theory suggests that each actor will require payment of its resources’ marginal productivity in order to provide incentives to recombine resources (Townsend and Busenitz, Reference Townsend and Busenitz2008). The transaction costs theory of the firm focuses on the balance between the costs of transacting in the market compared to the firm. Coase (Reference Coase1937) argued that firms exist when the transaction costs of using the markets are higher than those within the hierarchical structure of a firm. Transaction costs in markets can arise because of uncertainty; asset specificity – how specific the assets are to the required transactions; and the degree of opportunism (Williamson, Reference Williamson1985). Economists start with the premise that markets are an efficient way of coordinating economic activity in cases in which a pure exchange economy exists (e.g., spot contracts) (Williamson, Reference Williamson1985). Markets, however, prove to be inefficient if business dealings require coordinated investment where the composition of the assets is very specific to the transactions, and if the business dealings are of a non-contractable nature as a result of uncertainty (Grossman and Hart, Reference Grossman and Hart1986). Non-contractability can arise from asymmetric information between trading partners, bounded rationality and difficulty in specifying and measuring output (Gibbons, Reference Gibbons2005). Non-contractability is also an issue in team-based production, which faces challenges measuring the input of each party precisely, making it difficult to apportion the value created (Hart and Moore, Reference Hart and Moore1990). This gives rise to the metering problem, which is caused by shirking by the parties in the team, which is a form of moral hazard (Alchian and Demsetz, Reference Alchian and Demsetz1972). Organising team-based production is often therefore more efficient within the hierarchical structures of firms than markets.
Modularity is important in understanding how transaction costs determine the boundary of the firm (Baldwin, Reference Baldwin2007). Transactions are defined as mutually agreed transfers between actors with compensation. The costs of the output being transacted need to be defined and measured, which are called mundane transaction costs. The mundane transaction costs are minimised at the thin crossing points of the task network, which creates module boundaries. Tasks within such a boundary are called transaction-free zones, and transactions occur between modules. Therefore, tasks that correspond to problem-solving, as well as routine activities where transfers are dense, need to be located within these modules. These modules are called transaction-free zones, where the cost of the transactions is less than the value of the transfer. Transactions need to be located at module boundaries where mundane transaction costs are minimised; this corresponds to the thin crossing points of the task network and where the boundary of the firm lies. Hence, transaction costs theory explains how tasks are bundled into modules in such a way that transactions occur across modules, known as thin crossing points, in order to minimise costs – the breakpoints where firms and industries may split apart (Baldwin, Reference Baldwin2007). An alternative methodology to the transactions costs approach to determining the boundary of the firm is to examine the capabilities and resources (including knowledge generation) that maximise the resource portfolio of the firm, as well as the market demand for new customer value propositions (Jacobides and Billinger, Reference Jacobides and Billinger2006; Nikerson and Zenger, 2004; Priem et al., Reference Priem, Wenzel and Koch2018).
A valuable way to understand how such organisation of activities and tasks might be affected by digital technologies is to consider the firm as a multi-agent system with an identifiable boundary and goal, in which an individual agent’s efforts are expected to make a contribution (Puranam et al., Reference Puranam, Alexy and Reitzig2014; Kretschmer and Khasabhi, Reference Kretschmer and Khashabi2020). In this context any functioning organisation involves solving two fundamental and inter-linked problems, namely, the division of labour and integration of efforts. The division of labour involves how to divide the task – task division – and how to allocate the tasks to individuals or teams – task allocation. The integration of efforts involves the provision of appropriate incentives and motivation to complete the tasks – the provision of rewards – and what information needs to be provided in order to execute the tasks and coordinate the actions – the provision of information.
Digital technologies fundamentally alter the division and integration activities. First, digital technologies can provide significantly more information as a result of the liquefication process that we outlined earlier. Moreover, digital technologies could create a new wave of critical elements for creating customer value, and hence certain new activities might need to be added while others are eliminated. For example, Cemex, a Mexican-based global cement firm, was able to eliminate a substantial proportion of its human-based inspection tasks for cement supply because it could monitor inventory levels in real-time at client sites using sensors (Kretschmer and Khasabhi, Reference Kretschmer and Khashabi2020). Second, such a process requires the regrouping of tasks, not only in how they are to be divided but also in the integration process. The regrouping of tasks needs to be done by putting tasks that are complementary together in order to augment their combined value, as well as optimising coordination (Milgrom and Roberts, Reference Milgrom and Roberts1990). Such complementarities might arise by working with other firms. For example, Vodafone, a UK-based telecommunications firm, worked with Microsoft to provide the underlying technology infrastructure for its digital transformation efforts (Correani et al., Reference Correani, De Massis, Frattini, Petruzzelli and Natalicchio2020). Vodafone adopted Microsoft’s conversational autonomous interface using neural networks processing natural language when redesigning its customer care services. Vodafone is able to interact with its customers through multiple channels such as voice, social networks and home assistance, in order to listen, understand and assist customers.
Digital technologies also affect information flow for communication, monitoring and coordination. Research shows that information technologies and communication technologies have different impacts on autonomy and control (Bloom et al., Reference Bloom, Garicano, Sadun and Van Reenen2014). On the one hand, information technologies, such as enterprise resource planning (ERP) systems, increase the autonomy of employees and hence enable decentralisation. On the other hand, technologies that improve communications, such as data intranets and electronic communications systems, could potentially decrease the autonomy of employees and encourage centralisation. Hence, these two forces could impact how organisations divide and integrate activities, both within and across firms. Such division and integration as a result of digital technologies will affect the design of business models.
4.3 Realising the Full Potential of Digital Transformation
A report by the Conference Board titled “Realizing the Full Potential of Digital Transformation” noted that firms that have successfully executed digital transformation share two common characteristics (Hao et al., Reference Hao, Hicks, Popper and Velu2020). First, these firms have the discipline to undertake digital initiatives only if they are aligned with the overarching corporate or business unit strategy. Second, they can enable business model innovation through new value propositions for their customers through the adoption of digital technologies. The research study was based on interviews with executives such as chief digital officers, chief IT officers and R&D directors from fourteen companies across ten industries in the United States and Europe. The study provides three insights on digital transformation: first, digital transformation must be integrated within the business strategy; second, business model innovation is often a key part of digital transformation and leads to creating competitive advantage for firms; and, third, firms measure and manage digital transformation using a multifaceted approach. We discuss these insights from the Conference Board research report below.
The first insight of the Conference Board report highlights that of the biggest challenges that firms face is ensuring that their digital adoption fits the overall corporate strategy. Often firms might be tempted to adopt a new digital strategy, such as blockchain, sensors or additive manufacturing technology, because it is a novel technology, without necessarily thinking through how it will improve customer solutions. Digital technology adoption often requires continuous adjustment of activities and processes that needs to be closely aligned with the strategic planning process. Moreover, the adoption of digital technologies might inspire a pivot in an existing strategy or a shift to an entirely new one. However, the continuous alignment between digital technology adoption and corporate strategy needs to be maintained by the firm. This is especially challenging when firms are organised across functions and geographies where there could be a disconnect between digital technology adoption and the corporate strategy as a result of the misalignment of interests and rewards.
A US-based car-rental firm was able to learn from the acquisition of a new car-rental firm that rents cars by the hour and which had superior on-board electronics monitoring systems in order to manage the car-rental system. The firm learnt that having on-board electronics car technology on its regular longer-term rentals would be helpful in monitoring the performance of its cars and replacing vehicles if the system predicted a possible breakdown. The rented vehicle is normally replaced by exchanging another car with the faulty vehicle wherever the rental vehicle has been taken, for example, to a holiday destination. The firm learnt that this increases service quality and contributes to customer loyalty. With the increasing possibility of autonomous vehicles, the car-rental firm is looking to develop a business for fleet-management services as the result of on-board digital technology management systems – digital transformation that inspires a new business model.
There are several ways that firms can align their corporate strategy with digital technology adoption. The first approach involves the definition of a major customer solution outcome based on the technology trend. A large European personal care company has a strategy of helping customers to visualise and personalise their beauty products. Once this has been done, using cross-functional teams, IT develops feasible, cost-effective digital solutions. De Beers defined the need to have end-to-end traceability of its diamonds, because of the risk posed by lab-manufactured diamonds, and, with partners, it developed a blockchain system to do this, from the digital exploration of diamonds to omni-channel sales and distribution. The second approach involves organising cross-functional teams to develop the capabilities to achieve major strategic goals. At a major US-based semiconductor manufacturer, the firm identified thirteen competencies such as demand creation, product information management and order fulfilment to grow the market share and profitability of the firm. These major competency areas are then organised and led by a cross-functional team consisting of senior management from the various divisions of the firm, which reports to the executive leadership team. A third approach is followed by BBVA, the Spanish multinational financial services firm, which established a corporate venture fund to observe and invest in promising fintech start-ups that could be integrated into the bank’s business strategy. The fourth approach is to develop trends from the bottom up. For example, at 3M, a multinational conglomerate based in the United States, business units are encouraged to draw up strategic plans that address megatrends in technology and how they are likely to impact the business. Through a recursive process these are aligned with the corporate strategy and the strategy is refreshed when required.
The second insight of the Conference Board Report emphasises that digital technology adoption is most beneficial when firms re-examine how they create and capture value (Christensen et al., Reference Christensen, Bartman and van Bever2016; Cennamo et al., Reference Cennamo, Dagnino, Di Minin and Lanzolla2020). For example, a leading US publishing company has transformed the delivery of books from a physical to a digital format, which enables it to collect information on usage patterns, which can be used to provide additional services such as tutorials to improve students’ exam performance. Moreover, the digital model has also allowed the firm to explore new approaches to developing value, such as selling products directly to consumers and disintermediating distributors and developing subscription models for students, professors and institutions.
Two major challenges that firms face when leveraging digital technologies to change their business models are the piecemeal syndrome and incentive misalignment (Velu, Reference Velu, Nambisan, Lyytinen and Yoo2020). The piecemeal syndrome arises from the fact that managers are typically risk-averse, according to prospect theory, whereby disutility from failure is higher than the utility from gain. Hence, in most firms, the incentives are typically aligned to conduct incremental changes at the process or functional level, with less regard for the systemic changes needed for business model innovation, as the latter is often riskier than the former. This piecemeal adoption effect was seen with the introduction of PCs to replace mainframe computers (Steinmueller, Reference Steinmueller2000). Many firms treated PCs similarly to mainframes, with users continuing to procure services from the IT department as a centralised function while not fully benefiting from the more customisable opportunities provided by the PC. Hence, PCs took a long time to benefit firms, and they only did so when they facilitated the growth of more generic software, for word processing and manufacturing operations control, which was more customisable and could be upgraded, thus quickly displacing dedicated computers.
The second challenge relates to incentive structures that are misaligned with the new business model. For example, 3M sold furnace filters that keep air filtered and clean in buildings. The emergence of sensor-based technologies enabled 3M to develop a servitised business model based on selling “clean air” to its customers on a subscription basis; the company operates on the basis of promising a certain quality of clean air and is responsible for monitoring and cleaning – or replacing – the filters when necessary. The new subscription-based business model demands the minimisation of filter changes. However, the older business model requires maximising product sales and manufacturing large volumes to get the benefit of economies of scale to reduce the costs of production per filter. These conflicts often prevent firms from adopting new business models enabled by digital technologies.
The third insight of the digital transformation highlighted by the Conference Board report relates to how firms measure and manage digital transformation using a multifaceted approach consisting of input, throughput and output. Input involves measuring the capability and readiness of the organisation for digital transformation. For example, some organisations measure the digital capability of their staff in both soft and hard skills and ensure sufficient diversity among others between the sciences and the humanities. Gurbaxani and Dunkel (Reference Gurbaxani and Dunkle2019) proposed a six-dimension framework to measure firms that are gearing up for digital transformation. These six dimensions are: clarity of the strategic vision; the degree of alignment of the vision with investments; the suitability of the culture of innovation; the possession of sufficient intellectual property assets and know-how; the strength of digital capabilities; and the use of digital technologies. Firms can benchmark themselves on these dimensions against other firms in four broad categories: behind, on par, slightly ahead, and significantly ahead. Throughput requires organisations to systematically report on how effectively and efficiently their business model delivers the right customer value. The business model coherence scorecard (BMCS) discussed in Chapter 3 might provide a useful framework to identify and manage the efficiency and effectiveness of business models as part of the digital transformation effort. Finally, digital transformation success should be measured against achieving the business strategy. Such output measures call for a balance between financial and non-financial measures (Kaplan and Norton, Reference Kaplan and Norton2004; Velu, Reference Velu, Nambisan, Lyytinen and Yoo2020). Firms should use digital technologies to have real-time scorecard systems with customer-centric metrics across the different levels of the organisation that need to be fully aligned with the overall strategy of the firm.
4.4 The Age of Data and Artificial Intelligence
An increasing number of firms rely less on traditional business processes operated by customer service representatives, engineers and managers and more on algorithms to make decisions (Iansiti and Lakhani, Reference Iansiti and Lakhani2020; Bader and Kaiser, Reference Bader and Kaiser2019: Brock et al., Reference Brock and von Wangenheim2019). Artificial intelligence (AI) based algorithms set the prices for Amazon, recommend songs on Spotify, suggest films on Netflix, match cars with passengers on Uber and approve loans on Ant Group. These AI-based algorithms are able to learn and update in order to improve the decisions being made over time (Brynjolfsson and McAfee, Reference Brynjolfsson and McAfee2017). Many traditional firms are also beginning to embrace the use of AI to transform their business models (Cappa et al., Reference Cappa, Oriani, Peruffo and McCarthy2021; Del Giudice et al., Reference Del Giudice, Scuotto, Papa, Tarba, Bresciani and Warkentin2020).
One of the implications of liquefication, the ability of digital technologies to decouple information from its related physical form or device, is the abundance of data that becomes available (Normann, Reference Normann2001). Data has been defined as sign tokens or marks used to describe, represent or perform reality or index other marks (Alaimo et al., Reference Alaimo, Kallinikos and Aaltonen2020). Studies have claimed that data as sign tokens or marks operates at a far more granular level of reality than higher-order functional entities such as components (Von Krogh, Reference Von Krogh2018). These sign tokens or marks in the form of data are a key resource used in the digital economy to create value; the value is created from both recombination and reinterpreting through a process of encoding, aggregation and computation of data (Alaimo et al., Reference Alaimo, Kallinikos and Aaltonen2020).
Since John McCarthy coined the term AI (artificial intelligence) in 1956 at a conference in Dartmouth, there has been steady improvement in the techniques used to analyse and compute data. The ability to collect large amounts of data resulting from an increase in computer power has given rise to this era of AI advances. Artificial intelligence describes a set of advanced general-purpose technologies that enable machines to perform highly complex tasks effectively (Hall and Pesenti, Reference Hall and Pesenti2017; Ransbotham et al., Reference Ransbotham, Khodabandeh, Fehling, LaFountain and Kiron2019). AI technologies aim to reproduce or surpass abilities (in computational systems) that would require “intelligence” if humans were to perform them – the science and engineering of intelligent machines (McCarthy, Reference McCarthy1999). Machine learning is a type of AI in which a machine approaches a problem not by being explicitly programmed but by learning from data; it tackles a task by making data-driven decisions and predictions (Ghahramani, Reference Ghahramani2015). There are broadly three ways in which machines can learn: supervised learning, unsupervised learning, and reinforcement learning. In the case of supervised learning, algorithms learn to predict the output from the input data where all the observations in the data set are labelled.Footnote 3 Unsupervised learning involves observations in the data set that are unlabelled, and the algorithms learn the inherent structure from the input data. Reinforcement learning is based on interactions with the environment whereby the algorithm learns how to optimise based on the presence or absence of some reward.Footnote 4
The application of AI may be classified into a hierarchy of analytical tasks, namely, descriptive, diagnostic, predictive and prescriptive (Davenport and Harris, Reference Davenport and Harris2007). Descriptive AI involves the process of gathering and interpreting data to describe what has occurred. Diagnostic AI involves the process of gathering and interpreting data sets to detect patterns and anomalies and determine relationships in order to explain why something has happened. Predictive AI involves the use of descriptive and diagnostic analysis from the past and identifying the likelihood of particular outcomes in the future. Finally, prescriptive AI involves using descriptive analysis for what has happened, diagnostic analysis for why, predictive analysis for timing and a form of reoccurrence in order to infer actions to influence future desired outcomes. For example, in the case of Uber, descriptive relates to understanding the customer profile and the rides that customers have taken in each car. Diagnostic involves understanding why customers have taken those rides, for example, as a result of weather conditions. Predictive involves using historical data to predict which customers are going to need cars, when and for which destinations. Finally, prescriptive informs drivers about which passengers are likely to need cars, and when, and informs drivers to wait at the right time and in the right place for passengers. This could also involve suggesting to passengers that they might want to take an Uber for a particular ride. In the case of industrial applications, such as the use of sensors in manufacturing operations, descriptive might involve understanding the breakdown of machines in terms of type and timing. Diagnostic would include an understanding of why they broke down. Was it due to overheating, stress on the rotating parts or other factors? Predictive involves an analysis of which machines are likely to break down, and when, based on the past pattern of usage. Finally, prescriptive provides information for making suggestions about what should be done to minimise or avoid losses. Most applications of AI are predominantly descriptive, diagnostic and predictive, while the use of prescriptive is still in its infancy. Recent advancement in generative AI (such as Chat GPT) can create novel content from audio, text, images and video which can augment human creativity and potentially complement predictive AI (Eapen et al, Reference Eapen, Finkenstadt, Folk and Venkataswamy2023).
Many studies have highlighted the notion that AI should be used for augmentation rather than pure automation. Automation implies that machines take over a human task, whereas augmentation implies that humans collaborate closely with machines to perform tasks. However, Raisch and Krakowski (Reference Raisch and Krakowski2021), in their review paper, mentioned that augmentation cannot be neatly separated from automation because these dual AI applications are interdependent across time and space, creating a paradoxical tension. They provide two use cases of AI to illustrate their point about time and space. Let us consider the time element first, whereby the use of automation will have an impact on augmentation in the future, and there will be continuous interaction between them. JP Morgan Chase is increasingly using AI to help set the criteria for the selection of candidates to hire. By removing the human element, the idea is to make the system fairer and more efficient in choosing candidates for interview. However, over time it is likely that digitalisation might require employees to have data science skills, which do not play a major role in the extant criteria for selecting candidates. Hence, the initial use of AI for automation would require subsequent human involvement, with a focus on augmentation, which allows humans and machines to work together by altering the context and adjusting their models accordingly.
Let us consider the space element next, whereby the use of automation will have an impact on augmentation in adjacent processes. For example, Symrise, a major global fragrance company, adopted an augmentation approach to generate ideas for new fragrances using a database of 1.7 million fragrances based on customer requirements. This model was reached through a combination of master perfumers’ knowledge and customer purchases. The system initially consisted of automated searches for new novel fragrance formulas faster than humans were able to do. However, this use of AI to automate the idea-generation stage has an impact on the preceding objective-setting stage, whereby master perfumers must identify and input customer objectives and constraints in order to allow the automated generation of fragrance formulas matching these requirements in the idea-generation stage. This continuous iterative process of AI for automation and augmentation will enable new business models.
As machines become more intelligent using AI methods, they tend to have feedback systems to continuously improve performance, which in turn affects the organisational design (Galbraith, Reference Galbraith1974). This feedback is likely to affect business model design through the content of activities of the organisation, the structure in terms of the relationship between activities, and the governance with respect to the responsibility for decision-making. Studies have highlighted that AI today is particularly focused on machine intelligence, which tries to solve well-defined problems. AI is less suited to defining the problem, as this requires a sense of self, motivation and purpose (Raisch and Krakowski, Reference Raisch and Krakowski2021; Von Krogh, Reference Von Krogh2018). However, one of the challenges in redesigning the business model, as machines become more intelligent, is to identify the appropriate problems that need to be solved to enable the business model to evolve. Such types of problem identification might require more general or human intelligence (Davenport and Kirby, Reference Davenport and Kirby2015; Shevlin et al., Reference Shevlin, Vold, Crosby and Halina2019). Hence, cooperative methods for managers to use AI to enhance general intelligence would be needed for the business model to evolve appropriately to act as a complementary asset to intelligent machines in order to fully benefit from AI. Moreover, AI increases connectivity across the value chain. For example, AI is being used to develop digital twins, the generation or collection of digital data representing a physical object, which enables closer linking of product development, production, distribution and aftersales performance in the automotive industry, which calls for co-evolution of the business models of firms across the ecosystem. Hence, the business model needs to be seen as a complex system that co-evolves with the entire business ecosystem in which the firm is embedded as machines become more intelligent and connected (Burström et al., Reference Burström, Parida, Lahti and Wincent2021).
4.5 Next-Generation Digital Technologies – Quantum Technologies
One of the key issues of digital technologies for firms is not just considering technologies that have been commercialised but also looking into the future at emerging technologies. History shows us that the benefits of new technologies can take many years, if not decades, to materialise, as this requires new social, legal and organisational approaches (Geels, Reference Geels2002). Hence, it is appropriate to look forward in time to see forthcoming technologies that might have a profound effect on productivity and economic growth. One such technology is based on quantum mechanics, which is beginning to show incredible promise as a general-purpose technology that might have profound implications for all walks of life in years to come (Knight and Walmsley, Reference Knight and Walmsley2019; van Dam, Reference van Dam2020; MacQuarrie et al., Reference MacQuarrie, Simon, Simmons and Maine2020; Vedral, Reference Vedral2011). Quantum technologies are expected to enable business model innovations and contribute to productivity growth as well as economic and societal well-being (Velu et al, Reference Said, Page, Salter, Velu, Gallitto, Massi and Harrison2022; Velu and Putra, Reference Velu and Putra2023).
In the early 1980s, Nobel-prize-winning physicist Richard Feynman challenged computer scientists to develop a quantum computer (Cusumano, Reference Cusumano2018). Feynman’s challenge was based on his intuition that one could not solve problems in physics using classical digital computers with bits or binary digits, which are always either 0 or 1. This is problematic for physicists when representing a particle that could be in multiple states. For example, a particle (e.g., electron or photon) such as a qubit could simultaneously be in two states, both 0 and 1. It would be relatively straightforward to simulate with a single qubit, as it is in state 0 or 1. With two qubits, we have the possibility of having both in state 0, both in 1, one in 0 and the other in 1, or vice versa, which results in a total of four probabilities. This implies that N qubits could be in 2n states. Therefore, with 10 qubits, 1,024 probabilities are needed, and with 20 qubits, 1,048,576 are needed. The information states required for over 300 qubits would soon surpass the number of particles in the universe. However, the systems that scientists investigate may have a number of particles that is many orders of magnitude larger, and hence the number of probabilities becomes unimaginably large – for example, simulating molecules for drug discovery. Feynman published a paper in 1982, Simulating Physics with Computers, where he postulated that to simulate quantum systems one needs to build quantum computers, as the problem cannot be solved by simply scaling or applying principles of parallel processing with classical computers (Feynman, Reference Feynman1982). Feynman went on to remark (1982, p. 486): “Nature isn’t classical, dammit, and if you want to make a simulation of nature, you’d better make it quantum mechanical, and by golly it’s a wonderful problem, because it doesn’t look so easy.”
Physicists, computer scientists and engineers have come a long way since 1982 in trying to harness the power of quantum technologies (Pirandola and Braunstein, Reference Pirandola and Braunstein2016; Wehner et al., Reference Wehner, Elkouss and Hanson2018). In particular, there are three properties of quantum technologies that have significant implications for solving some of the most intractable problems of the world and significant societal challenges, such as the need for better medicines and therapies, understanding nitrogen fixation for making ammonia, discovering low-carbon technologies to address climate change and greater security around information, among others (Ball, Reference Ball2019; Government Office of Science, 2016). First, quantum technologies display superposition, where a particle can be in two or more states at once. Second, when two particles could be remotely connected – the entanglement principle – their state needs to be described jointly as a system and not according to the components separately which results in a correlation between spatially separated particles. Hence, when a measurement of the state is made on one particle, the outcome of the other entangled particle is known, regardless of the distance between them. Third, there is the uncertainty principle, whereby it is not possible to know the state of the particle until it is measured, when it will collapse to one state or the other, nor can an arbitrary quantum state be copied exactly (the “no-cloning principle”). These properties of quantum technologies have significant implications for how they might offer new solutions to problems that existing classical technologies (e.g., digital computers) will be unable to solve. We discuss below each of these properties and their implications for business model innovation in the case of quantum computers, sensing and imaging and communications.
4.5.1 Quantum Computing and Simulation
Quantum superposition, as a result of qubits being able to be in many states at once and entanglement enables complex calculations to be done much faster than classical digital computersFootnote 5 (Cusumano, Reference Cusumano2018). Hence, a quantum computer can process many inputs simultaneously instead of going through them one at a time, like a classical computer that can only be in binary states of 0 or 1. For certain classes of problems, this can mean a much faster solution – solving complex combinatorial problems or simulating material at the atomic or molecular structure (Sodhi and Tayur, Reference Sodhi and Tayur2022; Ruane et al., Reference Ruane, McAfee and Oliver2022). In order to harness the benefits of quantum computers, the qubits need to be realised and manipulated, which is not easy, as they are not stable. Qubits can be realised and manipulated using several classes of technology, such as superconductors, silicon spins or trapped ions among others, all of which are being developed as possible underlying technologies for quantum computers. Recently, there has been progress in achieving quantum supremacy, whereby a quantum system was able to do a calculation faster than a classical system (Ball, Reference Ball2020). For example, Google claimed in 2019 that its quantum computer carried out a specific calculation that would have taken a classical supercomputer 10,000 years to compute (Gibney, Reference Gibney2019).Footnote 6 A group of scientists based in China were subsequently able to demonstrate this using beams of laser light to perform computations that have been shown to be mathematically impossible for normal computers (Ball, Reference Ball2020). These are still the very early stages for comprehensive quantum computers. Nonetheless, although quantum computing is unlikely to replace classical computers, it might be very helpful in complementing them to tackle complex problems. The current state of art quantum computing is called noisy intermediate-scale quantum computers, as they experience errors for a particular number of operations and hence are not yet advanced enough for fault-tolerance. The vision is to build a fully fault tolerant quantum computer sometime in the future.
Quantum computers will have an enormous advantage in addressing two major classes of problems, namely, factoring large numbers and optimisation or simulation. We discuss the impact of factoring in the section on communications. Many commercial activities rely on optimisation or simulation; for example, complex products in aeroplanes are optimised using computer models before any real parts are manufactured, drugs are discovered by simulating new molecule combinations, new materials are created for batteries, financial portfolios are optimised, and routing is managed for logistics operations, among others. Moreover, many of the solutions using artificial intelligence use ever-larger data sets, where quantum computing might play an increasingly critical role in the future in speeding up machine learning capabilities. These optimisation- and simulation-based problems would result in more dynamic personalisation, for example, in the delivery of goods using logistics, where there is a much faster sense of responding to new requirements that are specific to individual preferences or in personalised medicine.
4.5.2 Sensing and Imaging
Quantum entanglement enables particles to be connected to one another, whereby the state of one of them provides information on the other (National Science and Technology Council, 2016). This unique property, which has no classical equivalent, together with the very sensitive nature of quantum mechanical systems can be used for various imaging and sensing applications.
Most cameras and imaging technology today are restricted to visible light and capturing images in two dimensions. However, the real world is three-dimensional and has multiple light spectrums going from radio waves and ultraviolet to infrared. Quantum technologies can leverage these multi-spectrums of light to capture images. Conventional cameras rely on recoding light that bounces back from an illuminated object, which requires the wavelengths of the light illuminating the object and the light captured by the camera to be the same. However, quantum imaging enables the illumination of an object with one type of light, such as infrared, but recording of the image using visible light. This is possible because the two types of light emitted by, say, an ultraviolet laser are entangled. Hence, the camera can build an image of an object based on the visible light, but it is the non-visible light that is projected onto the image because these two types of light are entangled. This type of quantum imaging is called ghost imaging and it is particularly useful for gas-emission detection in hazardous areas or for analysing chemical compositions.
Quantum sensors are being developed to provide more effective and non-invasive health and medical imaging such as brain tumours and cancer detection using entanglement principles. Quantum sensing is also very valuable for surveying hazardous environments such as mineshafts and also more efficient in the maintenance of infrastructure such as rail networks (Stray et al., Reference Stray, Lamb and Kaushik2022). Moreover, the use of quantum sensors can improve position, navigation and timing (PNT) systems to much-improved levels of accuracy compared to global positioning systems or global navigation satellite systems. These quantum sensors are particularly helpful for positioning and route navigation for autonomous cars, trains and other vehicles, especially in densely built environments, underwater, underground and in tunnels. Moreover, these PNT applications of quantum technology provides more accurate timing for energy management or financial transactions recording that calls for very fine-grained time stamping.
4.5.3 Communications
One of the most important areas of the application of quantum technologies is in communications (Cacciapuoti et al., Reference Cacciapuoti, Caleffi, Tafuri, Cataliotti, Gherardini and Bianchi2020; Wehner et al., Reference Wehner, Elkouss and Hanson2018). The first application relates to the security of communications and the second to the development of the quantum enabled Internet. Networking digital computers to create the original internet helped to transform business models and accelerate growth across many industries, and we believe that a quantum-enabled Internet will spur such new developments.
In order to ensure secure communications, cryptography is built into many communication systems, including commonly used web browsers and mobile phones. Cryptographic keys are analogous to keys in the physical world – they are used to encrypt and decrypt data from sender to receiver. Factoring is very important because it is behind the most common form of cryptography, which is used to protect private and sensitive data. The encryption commonly used today relies on factoring large prime numbers, which a classical computer is unable to do sufficiently quickly. However, a quantum computer could easily break this kind of cryptography, solving large prime number factoring, which could render global communications systems vulnerable to hacking. One possible solution is to use quantum key distribution, where it would not be possible to effectively intercept the message because any tampering would be detected through the principle of uncertainty. This ensures the security of communications.
The quantum-enabled Internet, a network connecting devices through quantum links together with classical ones, is often envisioned as the final stage of the quantum revolution (Cacciapuoti et al., Reference Cacciapuoti, Caleffi, Tafuri, Cataliotti, Gherardini and Bianchi2020), which could open fundamentally new communications and computing capabilities such as blind quantum computing which allows fully private computation. The quantum-enabled Internet would need to produce entanglement on demand between any two users. This might involve sending photons through both fibre-optic networks and satellite links that can extend the reach of entanglement by relaying it from one user to another along intermediate points. This would require routers and repeaters to transport quantum information to enable communication between distant quantum computers – a phenomenon called quantum teleportation.
Overcoming the class of problems outlined above by leveraging the properties of quantum technologies in the case of computers, sensing, imaging and communications could create new customer value propositions. Making such customer value propositions valuable to society calls for new means of organising the value creation, value capture and value network, and hence the development of new business models. Firms in some industries, such as financial services and healthcare, are already beginning to experiment with quantum technologies to solve problems that existing classical computers are unable to address effectively. The development of new business models to fully benefit from an emerging digital technology, such as quantum technologies, will require new cognitive models and leadership from both start-ups and established firms.
Conclusion
Organisations adopt different approaches to their digital transformation journeys (Hao et al., Reference Hao, Hicks, Popper and Velu2020). For example, in the financial services industry, with a vibrant fintech market and with start-ups disrupting the established ways of banking, some banks, such as Barclays and BBVA, have taken venture capital and incubation approaches to observe promising start-ups and drive innovation from an outside-in perspective. In the healthcare sector, health service providers have been trying to consolidate client data at each point of the life cycle and health journey to enable digital technologies to provide a holistic service and enable new business models. In other sectors, a performance-target-driven approach has been adopted. For example, a target of achieving 80 per cent deployability of aircrafts, which is often achieved by commercial airlines, enabled the US military service to implement digital transformation by adopting more data-driven analytical approaches for aircraft utilisation and maintenance.
Studies on digital transformation show that it is important to maintain customer-centricity and to keep the transformation journey agile by building cross-functional teams, working with external partners and ensuring the development of data proficiency among staff (Hao et al., Reference Hao, Hicks, Popper and Velu2020). Another area that has been emphasised as important in enabling digital transformation is the notion of the digital mindset. The digital mindset relates to employees having the perception that digital technologies are a strategic pillar of the organisation, and hence the firm having structures and processes in place for knowledge sharing, experimentation, continuous improvement, innovation and flexibility (Solberg et al., Reference Solberg, Traavik and Wong2020). Studies show that there could be two major considerations in shaping the digital mindset (Solberg et al., Reference Solberg, Traavik and Wong2020). The first relates to individual personal resources, where individuals see themselves as having either a fixed or a growth mindset. A fixed mindset relates to individuals who believe that their competencies are relatively fixed and who tend less towards embracing new opportunities to learn and grow. A growth mindset relates to individuals who believe that intelligence and ability can grow and hence look for ways to learn and grow. The adoption of digital technologies provides opportunities to experiment, learn and grow and might therefore be conducive to the growth mindset.
The second consideration relates to situational resources and whether people believe that resources are finite or expandable. On the one hand, when resources are finite, gains for one party correspond to losses for others. On the other hand, when resources are expandable, gains are possible for all parties. Therefore, it is important for management to understand the digital mindset of employees across these two dimensions when proposing digital transformation projects in order to position the opportunities and address the challenges appropriately. For example, when the Royal Bank of Scotland, a major British bank, was adopting an advanced data analytics system that could take over the core duties of financial advisors (primarily to sell ad hoc financial products), the bank created a new position of “journey manager” for advisors to take up a new format of customer service. The role entailed facilitating customers’ journeys through their major financial moments with the help of data analytics (Solberg et al., Reference Solberg, Traavik and Wong2020). This approach enabled the change to be accepted and the digital transformation to be implemented more efficiently.
The emergence of digital technologies also creates a challenge for firms designing business models that does not provide firms with an unfair advantage over their competitors. Digital technologies could result in unfair practices. For example, there is a practice known as the “last look” in foreign exchange markets (Anderson, 2005). This is where banks have the option to pull prices that they have offered on an electronic trading platform in order to protect them against toxic trade, with faster electronic trading venues being able to trade milliseconds faster because of the latent advantage, to the detriment of the banks. However, it has been known for banks to use this “last look” facility to pull a trade even when the trade is less profitable as a result of regular market price movements, to the detriment of customers. Such a practice could become prevalent in other fast-moving markets that are using ever-more sophisticated artificial intelligence and machine learning algorithms. These practices could result in business model designs that embed unfair advantages for certain firms over others and hence threaten the integrity of the marketplace.
There are several areas in the field of digital technologies and business model innovation that require further study, covering the identification of opportunities, phases of digital transformation and measuring the success of the transformation. First, what processes do firms have in place to identify digital transformation opportunities? We have discussed various ways in which firms seem to identify digital transformation opportunities. The question of whether or not there are certain ways of organising to identify new digital transformation opportunities that work under certain conditions compared to others requires further investigation. Second, do firms need to follow a phased approach to digitisation, and then digitalisation, followed by digital transformation? When does such a phased approach become effective and what should firms do to ensure that they are able to migrate seamlessly from one phase to the next? Third, how should firms measure the success of their digital transformation efforts? Whether or not there are particular metrics that are dependent on the type of digital transformation is an area that could be further explored as part of a research study.
This chapter has provided an overview of the opportunities and challenges of digital technology adoption and business model innovation and proposed some outstanding research issues on this topic.
