Onboarding artificial intelligence can secure the future of world’s critical electric systems through improvements in efficiency, innovation and decision-making. Power companies must adapt to thrive, says Gary Wong, Global Segment Leader of Power, Utilities, and Infrastructure at AVEVA
The future of power and infrastructure hinges on algorithms, not concrete or cables. Companies that master data-driven decision-making will build the grids and networks of tomorrow—whether through new projects or by squeezing efficiencies from congested and aging systems.
But catering to the growing global population and the push for decarbonization requires new approaches in infrastructure. Not only do aging systems need to be upgraded, but utilities must become resilient and agile to withstand business uncertainties and freak weather events.
Power grids require $3.1 trillion in investments before 2030, according to research firm Rystad Energy. The lack of investment is showing up in inefficiencies, reliability issues and missed opportunities in integrating renewables. In the US, for example, nearly 2,600GW of total generation and storage capacity—almost entirely zero-carbon—was awaiting connection even before Donald Trump’s inauguration. Across the pond, aging and outdated systems hinders National Grid Energy System Operator battery use in the UK, wasting renewables and increasing reliance on fossil fuels.
Relying on traditional models—manual controls, siloed approaches and static planning—is no longer sufficient. That’s where algorithms come in.
Why AI is integral to tackling infrastructural issues
Power infrastructure is becoming smarter and greener, led by industrial artificial intelligence (AI) applications, and supported by the internet of things (IoT) and cloud computing.
Together, these technologies play a multi-faceted role in modern energy management, to take just one infrastructure example.
Using IoT data in the cloud, AI algorithms can predict energy demand with precision and optimize the integration of intermittent renewables such as wind and solar to adjust energy production, reduce waste and improve reliability. In contrast with traditional methods that rely on expansive assumptions and periodic schedules, these innovations deliver measurable improvements for cash-strapped utilities, with real-time feedback loops enabling human operators to optimize decisions and prevent issues before they arise.
On just one front, onboarding digital technologies today could save $1.8 trillion in global grid investment over the next two decades, according to the International Energy Agency.
How companies are transitioning to AI-driven infrastructure
Duke Energy knows how algorithms can save costs and improve performance. Its single-window Monitoring and Diagnostic Center monitors diverse assets from renewable energy to coal and gas facilities across 60 plants in seven US states. Using low-cost sensors and AI algorithms, Duke tracks more than 500,000 data points and runs over 11,000 automated models across its fleet. Its operators now have real-time performance visibility and are alerted to problems before or as they occur. They have saved more than $250 million through predictive interventions so far, with one single early catch avoiding costs of $34 million.
Data-led technologies in the cloud likewise support cross-industry collaboration. Companies can jointly meet market needs by sharing operations information over a digital backbone. The US-based Western Energy Imbalance Market (WEIM) is a wholesale energy trading market that enables participating organizations to buy and sell energy. An advanced digital suite brings together its 22 members within a connected ecosystem, providing access to real-time information on market performance, business trends, and potential misalignments so that operators can minimalize financial settlement risks. Efficiency and decision-making have improved, and each WEIM member additionally benefits from flexibility and scalability, enabling competitiveness in an evolving market.
Elsewhere, innovative grid management is tackling the proliferation of new devices such as EVs, smart meters and solar panels are plugged in. UK Power Networks (UKPN) delivers electricity to 19 million homes in one of the world’s largest population centers. Recognizing the limits of reactive maintenance, it developed a data-intensive platform that processes four billion data points daily. AI-powered analytics predict failures before they happen, reducing outages and improving efficiency. With a centralized hub and streamlined operations, system operators can focus on network resilience, load balancing and strategic infrastructure decisions.
These are just some of the ways data and AI are proving their worth across key performance indicators such as grid reliability, cost reduction and energy efficiency in ways traditional methods can’t match.
Why human insight is the backbone of AI-enabled utilities
But while AI will speed up grid modernization, it can’t do so without humans. Only when experienced professionals program, monitor and fine-tune digital systems in line with business and social goals, can they yield superior outcomes.
AI-generated insights can guide long-term planning and real-time algorithmic adjustments can address immediate challenges, but neither can replace human expertise.
That’s why the majority (75%) of power producers in key global markets will prioritize investments in AI for over the next 12 months, according to the AVEVA Industrial Intelligence Index 2024. Speaking about key target investments with the greatest potential to drive opportunity for their organization, nearly half (43%) of executives look to invest in digital platforms that enable their workforces to share data and collaborate with internal and external partners.
A five-step transition to an AI-first power utility
AI will be indispensable in driving utilities’ transition. But adopting AI isn’t a single-step upgrade. Whether in power or other sectors, companies must take a phased approach to go from outdated systems to intelligent, self-optimizing networks.
- The first step is building a digital foundation after a comprehensive audit of existing infrastructure and data collection practices, value chain inefficiencies and operational challenges. By assessing where improvements are needed, companies can pinpoint target areas for AI-driven solutions.
- Next, establish the necessary digital infrastructure for AI. Typically, this is a mix of cloud and edge computing to ensure applications run efficiently and effectively. The cloud delivers scalability, while real-time processing happens at the edge—critical to immediate decisions around predictive maintenance or load balancing.
- Pilot programs with clear success metrics are crucial to test AI’s effectiveness and minimize risk. Well-defined pilots allow strategic recalibration and can secure buy-in.
- Next come scaling and refining AI solutions across all aspects of operations, including deploying customer-facing AI tools in line with business goals. Constant fine-tuning using live operational data is essential to get the most from AI tools.
- As AI systems mature, the most essential step for long-term success is continuous innovation. This occurs by way of real-time performance monitoring, regular retraining of AI models with new data to improve accuracy and adaptability, periodic comparisons with KPIs and benchmarks, and adjusting goals based on performance insights.
Why stakeholder alignment helps future-proof the power sector
The rise of digital-first utilities is only a matter of time. With the AI in energy market growing at a compound annual rate of 36% through to 2030, data-driven technologies will play a seminal role in building secure and decarbonized power and infrastructure systems to support humanity’s future. However, for AI-led digital transformation to be truly effective, it must be guided by clear policy frameworks, strategic partnerships with technology providers and investors and open dialogue with customers.
Algorithms can cement power’s future, but only if all stakeholders are on board. The question is no longer whether the sector will adopt AI; it’s whether it will do so fast enough to remain competitive.
