How artificial intelligence and automation can help transform power and utilities

Amid a cost-of-living crisis and the shift from fossil fuels to low-carbon alternatives, power and utilities face a seismic shift. According to the International Energy Agency (IEA), the world is on course to add more renewable capacity in the next five years than has been installed since the first commercial renewable energy power plant was built more than 100 years ago.¹ This transition, characterized by a departure from traditional reliance on large, dependable plants to a landscape marked by numerous renewable energy units, brings new challenges. These include intermittent output, system instability, reliability concerns and the escalating impact of climate change-induced events like floods, storms and wildfires. We expect that digital transformation, specifically emerging technologies like AI, will be essential for technology officers to navigate these complexities.

This transformation presents challenges, compounded by increasing demand from customers seeking reliable, always-on essential services that lower their greenhouse gas emissions. Some consumers have taken matters into their own hands, engaging in self-generation and storage. Meanwhile, natural gas networks are grappling with how to decarbonize their networks and what emissions reduction targets mean for their businesses.

According to the KPMG 2023 CEO Outlook,² nearly two-thirds (64 percent) of energy CEOs agree investing in generative AI is a top priority, with 48 percent expecting to see a return on their investment in three to five years. However, progress among many utilities is slow, stemming from a lack of capability, understanding or, in some cases, reluctance to embrace change. For example, many kinds of assets now generate an abundance of data — the global fleet of wind turbines alone is estimated to produce more than 400 billion data points per year³ — but without collection and organization this cannot be used to improve decision making. According to research by KPMG Australia, many utilities have a ’hidden debt’ of low workforce productivity which technology could improve by transforming processes and service delivery models.⁴

Generative AI has captured the world’s imagination with its ability to produce text and images in an uncannily human fashion. However, while generative AI holds promise in certain applications, power and utilities can benefit from exploring a wider array of digital intelligence and automation technologies, including various forms of AI such as machine learning and robotic process automation. The prominence of generative AI could serve to underscore the importance of investigating broader digital intelligence and automation.

source: KPMG International. ‘KPMG global tech report 2023.’ 2023.

How utilities can leverage AI

There are compelling instances where power and utilities can embrace intelligence and automation technologies, with a few utilities now scaling up projects and integrating them across entire organizations. This move holds immense potential for enhancing their benefits. These experiences show that their peers can consider the following specific applications:

• Investment decision making: Electric utilities can use AI to support their investment decision-making process. Generation and grid network assets have high costs and are likely to be in use for decades. The transition to renewables can make industrial batteries and other energy storage options more attractive, but business cases require robust predictions. Organizations can use AI for options analysis, scenario planning and modeling, helping to save them money and improve their planning in areas including regulatory work, assets, working with their ecosystem of suppliers and grid connections.
• Customer information and relationship management: UK-based Octopus Energy Group has developed a cloudbased digital platform called Kraken that applies intelligence and automation to customer information and relationship management. As well as supporting its own growth, Octopus has licensed it to other utilities in the UK and elsewhere and is also collaborating on developing wind energy. Octopus, which KPMG in the UK has advised on all of its capital raising in addition to several international acquisitions, uses the Kraken platform in generation as well as supply, making it an example of a utility that is using such technologies across the organization.
• Regulatory workload management: Utilities can use generative AI to manage regulatory workloads, including assisting with rate cases, the process through which US companies apply to state public utility commissions for rate increases to fund improvements. Rate cases can involve analyzing regulatory documents, sometimes totaling 10,000+ pages, which generative AI can summarize and provide footnotes. Getting humans to add a degree of structure, such as feeding the material to the AI system in chunks, can help it to avoid mistakes. Using a generative AI system that produces references can also allow for human checking.
• Validating schematics: Utilities are exploring the automated review of grid connection schematics, technical drawings submitted by developers of new buildings. At present, these can require extensive and time-consuming checks by the utility’s engineers, including the correction of basic errors. Generative AI could potentially be used to analyze symbols and images that could provide an automated first line of support. This analysis could spot potential errors and could accelerate the process. Engineers could then focus on other issues. This process could work as a shared service across the industry to help cover its costs.
• Enhance travel for field service engineers: Utilities can also look to enhance travel by field service engineers, using automation to help reduce the time spent driving between jobs known as ‘windshield time’. Doing this can potentially make the company more efficient, but there are opportunities to go further by linking the system to other sources of information. Using weather forecasts could help field service engineers avoid adverse conditions, coordinating with maintenance plans could help them visit when it is easiest to do their work, and data on specific components could be used to get these components checked when they are worn but yet to fail. It could also be used to schedule training on new components automatically before engineers encounter them for the first time.⁵

Utilities are also using AI for auditing and managing assets, remote monitoring of sites, including automated analysis of camera output, running digital simulations of the impacts of flooding and managing relationships with contract electricians. However, many projects are small-scale proofs of concept run by technology departments, which often end without plans to develop them.

How to help mitigate risks, build trust, and scale AI responsibly

The issue of scale is perhaps the most significant challenge for intelligence and automation technologies. Utilities tend to work at large scales, so technology projects should work at a similar size to make a significant difference. We believe the simple answer is for utilities to start small but think big, with plans to evaluate proofs of concept then putting successful ones into large-scale production, a top-down approach that will usually need board-level support. However, doing this means tackling other issues that can hamper such projects.

Utilities should seek to modernize technology architectures to help ensure their data is reliable and of high quality and can be accessed in real time. Many have developed these architectures incrementally, resulting in fragmented data sets trapped within departmental systems. These should be joined up through modular, scalable architectures that allow AI systems to access information across the organization.

In our view, utilities need better data to run effective digital projects. This can be achieved with better data governance controls, with projects to improve the quality and accessibility of older data, and by explaining why this matters, so that staff across the organization realize that keeping accurate data is important rather than an impediment to their jobs. Some utilities that have invested in technology over recent years, such as enterprise resource planning or human resources systems, are likely to find these provide good quality data that can be used with AI.

Customers worry about their personal data and what utilities will do with it, arguably in ways that many do not apply to smartphones and other personal technology. Utilities can help address this by being open about their uses of personal data and by applying techniques such as anonymization and strict access controls that can strengthen privacy while allowing data to be used effectively. This can be part of more general work to help improve relationships and trust with customers, who in some countries get treated as an afterthought rather than an organization’s focus.

Utilities tend to be both tightly regulated and culturally reluctant to move first in adopting technology, both of which can make it difficult to make the case for spending on digital technologies. Many also lack skilled internal talent. The cultural issues can be tackled through efforts to improve AI literacy to help reduce the fears of both executives and staff at other levels.

Using AI effectively can present specific challenges, including choosing what type to use. Generative AI is good at handling large amounts of text, while other AI systems work better with large amounts of structured numerical data. The latter would be a better choice for tasks such as planning work schedules, with generative AI being used to explain tasks in accessible language.

Another key issue is the responsible use of AI, particularly if it is used to advise on or make highly significant decisions such as choosing which area suffers a power cut. According to the KPMG global tech report 2023, 55 percent of organizations said progress toward automation has been delayed because of concerns about how AI systems make decisions.⁶ Similarly, 60 percent of energy CEOs agree that implementing generative AI can result in ethical challenges such as plagiarism, data protection, bias and lack of transparency.⁷ Effective human supervision of such decisions and documentation of what data an automated system uses are among the ways to reduce risks from automated decisions.

Read more about the KPMG Trusted AI framework here. 

Next steps: Build the case for AI and automation

In the rapidly evolving landscape of power and utilities, we believe strategic decisions regarding the adoption of intelligence and automation technologies are vital for enhancing operational efficiency and staying competitive. The following are steps that power and utilities can take to effectively navigate the ‘buy or build’ dilemma and cultivate a forward-thinking approach to technology integration.

By focusing on actionable steps and strategic approaches, power and utilities can effectively harness intelligence and automation technologies to enhance efficiency, reliability and sustainability in their operations.

1. Evaluate commercial products: Consider purchasing viable commercial products, especially for non-core areas, to help streamline operations and save time.
2. Assess proprietary data: For core areas involving proprietary data, building models within the organization may be more effective in maintaining control and helping to maximize utility-specific insights.
3. Enable quick decision making: IT organizations should facilitate ‘buy or build’ decisions promptly, helping to ensure they have the necessary capacity to manage the chosen outcome. This includes having staff capable of supporting users of purchased services.
4. Focus on value generation: Prioritize projects with the potential to generate significant value. Avoid investing in technologies solely because they seem trendy; instead, assess their potential impact on utility operations.
5. Embrace flexibility: Opt for digital technologies that offer flexibility and adaptability, allowing for a range of tasks, even those not currently envisaged. This can help prevent being tied down by rigid systems and enables scalability.
6. Establish an innovation center: Create a dedicated innovation center to systematically test new technologies and ideas, helping to foster a culture of continuous improvement and adaptation.
7. Forge strategic partnerships: Develop partnerships with digital technology providers through initiatives like in-house venture capital funds. This can help facilitate access to cutting-edge solutions and encourage collaboration in advancing utility operations.

Choosing to be a leader

We believe digitalization will happen and companies can choose to be leaders or followers. While both paths carry risks, trailing behind as a follower can potentially lead to being left in the dust by competitors. For those who dare to lead, the potential benefits can be immense, albeit requiring concerted efforts to integrate projects throughout the organization. This entails honing strategies for value generation, securing funding and engaging with boards and regulators. In our view, leaders need a bedrock of good corporate technology and data as well as an organizational culture that is open to change. This includes managing the risks of physical and data security, something which is a particular challenge for utilities as information and operational technology converge. It also involves navigating an evolving regulatory landscape on AI in ways that can balance risks and rewards.

When it comes to AI, establishing controls and governance can be crucial before diving in. Starting with small-scale pilots is often prudent, provided there is a clear pathway for project expansion. We believe it is essential to acknowledge that no single AI option works for everything.

The adoption of digital intelligence and automation is a journey, not a destination. Taking proactive steps to begin this journey helps set the stage for ongoing progress and adaptation in an ever-evolving landscape.

How this connects with what we do

KPMG firms can help power and utilities find the right technologies and partners, as well as support the business case development and direction of their implementation. We combine industry knowledge with a strong understanding of digital intelligence and automation technologies, and how they are used in power and utilities around the world. We can help companies explore possible innovations through use of our ignition centers.⁸ We have collaborations with many of the leading providers of technologies and have access to innovators that can help migrate legacy software to modern platforms.⁹ When we cannot join forces, we can build software. And we can help staff across power and utilities use new technologies, including through literacy programs, or help reorganize how power and utilities can fit technologies into their organizations.