Harnessing AI’s Potential to Optimize T&D^*1 Grid Operations

Energy Solution Lab (ESL), a division within Hitachi America Ltd.'s R&D department, was launched in 2017. Our team focuses on developing next-generation digital energy technologies. Based in California, we actively collaborate with federal and state governments, as well as utilities and research partners across the U.S., further strengthening Hitachi’s position as an innovative leader in the energy industry and ecosystem development.

At ESL, one of our primary goals is to address the technology gap created by emerging trends in the energy sector. Solar energy, in particular, is revolutionizing power generation by being highly distributed and close to end-users, fundamentally changing how utility grids operate. However, existing utility software systems, which were developed 20 or even 30 years ago, were not designed to manage the complexity of today’s increasingly distributed energy resources.

To bridge the gap between legacy systems and evolving industry trends, and to facilitate more advanced grid planning and operation, ESL partnered with the California Energy Commission to develop GLOW—a powerful cloud-based platform that simulates the interactions of power distribution system components to support grid planning and analysis. Since its inception, GLOW has expanded to include three modules: Glow.AI, Glow.Ops (operations), and Glow.Edge.

Glow.AI helps utilities and grid operators process data, extract valuable insights, and make informed decisions in increasingly complex operating environments. In addition to leveraging advanced AI for decision-making and data analysis, we are also exploring the use of large language models (LLMs) and generative AI to further enhance decision-making, enabling the energy industry to make faster, more effective, and cost-efficient decisions.

While load growth^*2 in the U.S. has remained relatively stable in recent decades, the key drivers for both current and near-term load growth are emerging technologies, particularly electrification and the expanding demands of data centers. As AI technology continues to advance, the need for massive computing power to support AI training is driving significant growth in data center loads. This shift presents exciting opportunities for innovation in energy solutions. However, as data center demand escalates, there are concerns about potential capacity challenges. Major tech companies’ data centers already require power on the gigawatt scale, highlighting the need for strategic planning and innovation in grid capacity.

In parallel, the electrification of transportation is expected to drive substantial growth in EV charging loads. As more consumers and businesses embrace electric vehicles, the demand for charging infrastructure and electricity to power these vehicles will continue to rise rapidly. Although recent policy shifts may have temporarily slowed the pace of EV adoption in some areas, the long-term trajectory remains positive. The continued expansion of EV charging networks and the increasing adoption of electric vehicles will create new opportunities to optimize energy use and integrate innovative solutions into the grid, driving sustainability and energy efficiency.

Additionally, the energy industry faces challenges such as the backlog in generation integration^*3, with new transmission generation projects taking up to five years—too slow to meet the rapidly evolving load demands. One significant challenge in California is the "duck curve," a V-shaped pattern where midday solar power generation dominates power supply, but load peaks in the early morning and late afternoon, when solar energy is minimal. This shift demands more flexible, cost-effective peaker generation capacity^*4, which is often more expensive than base load generation.

In response to these challenges, the energy industry is embracing more interconnected grid systems to accommodate load growth while keeping costs manageable. On the retail side, Virtual Power Plants (VPPs)^*5 are increasingly playing a pivotal role by integrating distributed solar energy and energy storage, creating a more flexible and resilient grid.

AI technology holds immense potential in addressing these challenges by enhancing the control and management of energy systems. Digital solutions powered by AI will become a key differentiator for energy vendors. Utilities are actively seeking AI-driven solutions that can operate grids more efficiently, autonomously, and at a lower cost, leveraging IoT^*6 to create smarter, more responsive grid systems.

*2

Load growth refers to the increase in demand for electricity over time.

*3

Generation integration refers to the regulated process to integrate new generation to the utility grids.

*4

Peaker generation capacity is power plants used only during periods of high demand.

*5

VPP refers to a mechanism or application of operating various small-scale energy resources such as solar power and energy storage near energy consumers, as if they were one power plant.

*6

IoT stands for Internet of Things.

While generative AI has made significant strides in natural language understanding and open-ended use cases, challenges remain, particularly with accuracy and hallucinations in LLMs. In the energy sector, core operational software requires a level of safety, precision, and regulatory compliance that generative AI was not originally designed to provide. For instance, decision-making in grid operations involves processing and analyzing structured data from field measurements, and recent benchmarks show that generative AI struggles to achieve 100% accuracy, particularly when handling straightforward queries. The energy industry is also highly regulated, and AI, despite its advancements, has limitations in generating content that fully adheres to stringent legal and regulatory requirements.

Another challenge is the scarcity of domain-specific training datasets. Most generative AI and LLMs are trained using publicly available, general-purpose data, which contrasts sharply with the proprietary datasets common in the energy sector. This lack of accessible, high-quality data creates significant barriers for AI solution providers looking to develop accurate, tailored models for energy applications. Furthermore, many vendors lack the resources or expertise to build their own specialized LLM models. While it's possible to customize general-purpose models to suit specific needs, this raises additional concerns around maintenance, insurance, and partnerships.

Despite these hurdles, I remain a strong believer in the transformative potential of advanced AI technologies. The challenges we face today are not insurmountable, and rapid advancements in AI research suggest that solutions will continue to evolve. As AI models improve in accuracy, reliability, and domain-specific adaptability, they will play an increasingly pivotal role in industries like energy. The development of specialized AI models and datasets for the energy sector will help address many of the current barriers related to proprietary data and regulatory compliance. As energy organizations collaborate and share best practices, AI tools will be refined to meet the sector's stringent needs.

I firmly believe that over time, AI will bridge gaps in operational efficiency, grid management, and decision-making, fostering a more sustainable, responsive, and resilient energy infrastructure. The potential for AI to revolutionize the energy industry is immense—enabling smarter grid operations, reducing costs, and optimizing energy use in ways that were previously unimaginable. As we continue to navigate these challenges, the promise of AI to unlock new levels of innovation, efficiency, and sustainability remains a powerful driving force.

In the future, VPPs are expected to play an increasingly prominent role on the retail side of the energy market, supported by the growing adoption of real-time pricing mechanisms. However, VPP models vary significantly by region, shaped by the nature and scope of local regulations. Technology vendors must stay attuned to these regional market dynamics to remain competitive.

At ESL, we are pursuing a dual-track strategy—balancing immediate needs with long-term innovation. In the short term, we are prioritizing traditional operational technology (OT) software as the primary engine for reliability and performance, with AI playing a complementary role. When OT systems are functioning effectively, there is often little incentive for immediate overhaul. However, AI offers clear value in enhancing user experience, particularly in human-software interaction. For instance, through our GLOW.AI solution, we’ve introduced a prototype feature called “Live Assist,” which simplifies data analysis via text- or voice-based commands. Meanwhile, critical backend operations continue to rely on robust OT systems to ensure safety and stability.

Looking ahead, we foresee domain-specific LLMs being trained to support technical tasks and real-time decision-making for grid operators. Given the complexity of the energy sector and its deeply embedded real-world challenges, a single LLM is unlikely to be sufficient. Instead, we anticipate a more modular approach—one that leverages a collection of AI agents, specialized modules, and diverse AI techniques to solve distinct problems.

While the AI industry is currently experiencing unprecedented levels of capital investment, the key challenge remains monetization. Controlling the cost of AI technologies will also become increasingly important. The energy industry, though cautious, is actively exploring practical use cases where AI can deliver tangible value. For solution providers like Hitachi, this represents both an opportunity and a responsibility—to develop business models that are commercially viable and technologically sound.
One of Hitachi’s core strengths lies in our unique capability to bridge IT and OT. On the OT side, Hitachi Energy has built a strong reputation, underpinned by extensive market reach and trusted relationships. This solid foundation positions us well to expand into digital solutions, leveraging the GLOW suite to drive innovation across the energy value chain.