Many types of artificial intelligence (AI) capabilities have accelerated in recent years due to tumbling costs of data storage and processing, rapidly expanding data availability, and improved data storage and modelling techniques. In general, analytical AI can analyze historical data and make numeric predictions, while generative AI (gen AI) allows machines to produce new outputs similar to human-generated content. Gen AI, in particular, has been building momentum since 2017 and hit an inflection point at the end of 2022 when applications such as ChatGPT became publicly available.
It’s no surprise, then, that AI adoption has surged across industries. For instance, in 2023, a third of respondents taking part in McKinsey’s annual global survey on the state of AI indicated that their organizations regularly use gen AI in at least one business function, and 60 percent of organizations that have adopted analytical AI said they are also developing gen AI use cases.1
Historically, the rail industry faced challenges in adopting digital technologies due to limited data availability and quality, regulatory considerations, and lack of standardization. Today, analytical AI and gen AI provide an opportunity for companies across the railway value chain to further embrace digitization.
A recent report, The journey toward AI-enabled railway companies, produced by the International Union of Railways (UIC) in partnership with McKinsey, examines the adoption of analytical AI and gen AI in the rail industry, and the business potential that these new technologies can bring. The report finds that railway companies have already begun to implement various AI technologies for around 20 key use cases. Greater adoption could unlock an estimated $13 billion to $22 billion in impact a year, globally.
At present, only a few railway companies and OEMs are implementing multiple use cases at scale. The report identifies use cases that have been deployed, or have the potential to be deployed, and looks at success factors for implementation.
Railway companies are focusing their efforts on about 20 use cases
Although there are more than a hundred potential use cases, railway companies’ efforts are mostly focused on a few analytical AI use cases. Some gen AI use cases were noted but gen AI was not defined as the preliminary focus of the study and is still nascent in most of the cases. Use cases tend to target business priorities relating to four KPIs: on-time performance, customer engagement, safety, and operational performance. These KPIs are aligned with the top four criteria that passengers, across geographies, use when choosing their mode of transport. A 2022 report by UIC and McKinsey, Boosting passenger preference for rail, identified these criteria as price, safety, reliability, and convenience. Exhibit 1 summarizes the key areas of potential for railway companies looking to leverage AI.
While the range of potential applications is substantial, for most railway companies AI is only an emerging trend—few have implemented any kind of AI at scale with success. Around 25 percent of companies have implemented multiple use cases at scale, and roughly 35 percent of companies have one or two use cases at scale, with other use cases being in pilot stage.
Use cases vary in terms of the maturity of the technology and their adoption by the rail industry. There are around 20 common use cases, at different stages of maturity, across the four groups of business activities: Railway undertakings; infrastructure management; passenger experience; and support functions.
Exhibit 2 plots the most common use cases in terms of maturity and adoption. Use cases higher up on the curve are likely to have been adopted by all the major railway companies. Use cases lower down have been adopted by fewer companies.
In some instances, use cases are identified as being mature but not yet deployed at scale—often when the use case was pioneered in an adjacent industry that helped mature the technology. Take, for example, revenue management systems used in the airline industry. The technology and use case are mature, but the level of adoption in rail is relatively low as reservation systems work differently in each industry.
Would you like to learn more about our Travel, Logistics & Infrastructure Practice?
Railway undertakings
Railway undertakings are companies or entities responsible for operating and managing railway services, including the provision of train transportation. Here the most mature analytical AI use cases, in the process of being fully deployed in the field as well as those already deployed and capturing impact, focus on shift planning and energy efficiency. AI solutions that optimize crew planning and shift planning have been deployed across all business units that work in shifts including train drivers, onboard staff, and maintenance operators. In some instances, adoption has generated a 10 to 15 percent optimization in shifts as well as reductions in labor costs.
Use cases in pilot phase, that have shown ability to drive impact through proof of concept (PoC) and are currently being improved before being deployed at scale, include predictive maintenance for rolling stock. Depending on the type of rolling stock and the type of component, predictive maintenance has enabled a 15 percent increase in reliability, and a 20 percent reduction in maintenance costs.
Autonomous trains are currently at the PoC stage. A few railway companies are exploring the potential of semi-autonomous and driverless trains intended to improve capacity and efficiency.
Use cases still in the early stages of exploration include disruption management through AI-powered digital twins of real-time operations.
Infrastructure management
In the context of the railway industry, infrastructure management encompasses planning, operation, and maintenance of the physical and organizational components of rail networks, including tracks, stations, and signaling systems. At-scale use cases are focused on predictive maintenance for rail infrastructure and crew and shift optimization. Use cases in pilot phase span passenger flow management, capacity planning, and real-time traffic management.
Use cases at PoC include inventory management, and maintenance co-pilots. As seen in railway undertakings, nascent use cases involve AI-powered digital twins, in this instance for optimizing the design and construction of infrastructure projects.
Passenger experience
In the railway industry, this refers to the overall satisfaction and comfort of individuals using train services, encompassing aspects such as service quality, convenience, amenities, and customer interactions. At-scale use cases focus on revenue management, security, and providing real-time intermodal information. A quarter of the railway companies in the research sample have pursued the use of artificial vision and predictive algorithms that support security. Other use cases, mostly in pilot phase include passenger flow management and content generation.
Support functions
Support functions include essential non-operational activities such as HR, finance, communication, IT, and procurement that contribute to the overall efficiency and effectiveness of railway undertakings and infrastructure managers.
Most use cases are still nascent or in pilot phases such as people analytics, talent training, software development, and using gen AI to quickly access and understand complex documentation.
Implementing AI: The size of the prize
Overall, various AI technologies can support railway companies to better invest, build, plan, and deliver efficient operations and meet passenger needs. To illustrate, for a €5 billion rail company, AI could deliver around €700 million a year in value (Exhibit 3).2 This includes increasing revenue through revenue management solutions and infrastructure capacity use cases, as well as optimizing labor, maintenance, and corporate costs.
Implementation is key for realizing this value. Many use cases can be successfully designed and deployed at scale within 12 to 18 months to realize value. The journey to become a data-driven company, fully integrating analytical or gen AI use cases in ways of working and operating can be challenging. In fact, over 60 percent of companies across industries experience a stall at some point on their digital transformation journey.3
Railway companies can take inspiration from data-driven companies in adjacent industries. What these companies have in common is that they put six building blocks in place that are key to a successful digital and data transformation: strategic roadmap, talent, agile operating model, technology, data, and adoption and scaling. Companies interested in exploring the power of all AI technologies, and those continuing to innovate with AI at an enterprise level, can focus their efforts on these six key components.
Delivering on the promise of AI may not be easy. Many railway companies have not deployed use cases at scale, yet. For those that have, successful deployments are characterized by investment in dedicated capabilities and talent, and the definition of clear objectives—aligned with business priorities—which helped focus investment on a few game-changing use cases. While transformative, AI can bring a new set of risks that may need to be addressed from the beginning. Accordingly, organizations looking to adopt AI would do well to prioritize strong data governance and robust cyber security.
If this seems daunting it is worth remembering that railway companies do not need to act alone. There is a wide ecosystem of partners and vendors with deep technical and business expertise to support this journey.
