Hitachi Digital Services: Will AI Impact the Energy Sector?

AI can boost efficiency, help to integrate renewables, enable predictive maintenance and may even generate billions in economic value in the energy sector. 

Rajesh Devnani is Vice President – Energy & Utilities Vertical and Strategic Accounts at Hitachi Digital Services.

He shares his thoughts with Energy Digital.

What are the main challenges in adoption and scaling of AI and Gen AI? 

The adoption and scaling of AI, including Generative AI (Gen AI) and now AI Agents, in the Energy and Utilities sector presents both promising opportunities and complex challenges. The industry is currently facing pressures to reduce costs, enhance efficiency, improve sustainability and adapt to changing regulatory and environmental standards. AI and Gen AI can significantly contribute to these goals, but the path to effective implementation and scaling requires overcoming several obstacles and adopting tailored strategies.

Running multiple siloed pilot initiatives without a comprehensive digital strategy (should instead assess on feasibility and impact) is a key adoption challenge. The strategy of ‘Let a thousand flowers bloom’ or a scattershot approach won’t yield the results. We are likely to witness the ‘purgatory of the pilot’ which was quite rampant in the early days of IoT adoption. Before jumping on to ad hoc Proof of Concepts, it is important to prioritise high impact use cases that will potentially deliver a high ROI based on feasibility.

The key challenges associated in scaling AI and Gen AI adoption revolve around data quality/complexity, limited talent availability, regulatory and security constraints, existing technical debt around legacy systems, uncertain ROI and the ability to manage change.

Energy organisations need to approach these challenges thoughtfully to scale AI in an agile, impactful manner. Adopting a platform and ecosystem driven mindset leveraging existing technology and a modular approach and investing in modernising the data and cloud infrastructure and adopting a future-proof Hybrid AI environment will be quite key. AI and specifically Gen AI with it’s far reaching democratisation potential has far reaching impacts across diverse stakeholder groups and hence it is imperative to build strong cross-functional teams and eliminate internal silos. Given the importance of the industry as Critical infrastructure, it is equally important to prioritise AI Governance including privacy, security, transparency and ethics.

How is AI being used across various functional domains?

AI is significantly transforming the Energy & Utilities industry by enabling smarter decision-making, optimising operations and enhancing predictive capabilities. AI and Gen AI can be applied across various functional domains, including asset management, demand forecasting, customer service, energy trading and sustainability efforts. At Hitachi, we have been engaged in helping our clients across a multitude of Industrial AI use cases. These include:

• Guided Repair/Predictive Maintenance for fleets and how GenAI is further accentuating the value delivered
• Intelligent Infrastructure Monitoring for power utilities
• Sewer Pipeline Infrastructure Compliance management for Water utilities

What emerging AI trends should energy leaders be aware of? 

• AI Agents are a revolutionary technology and a step change to the current technology. Think of the paradigm shift from typing commands to the Graphical User Interface or the advent of the mobile apps which spawned huge ecosystems, revolutionised computing and enabled access for the masses. AI agents are the killer use case that Gen AI was waiting for.  
• System 1 to System 2 transition for models: Models are increasing to moving to the Left-Brain architecture of the 2-system thinking model proposed by Daniel Kahneman. Models are becoming stronger on the reasoning aspect much beyond the original concepts of next token prediction. There will be a stronger emphasis on concepts like Chain of thought reasoning, inference time compute etc. 

• Open-source models are reaching parity with proprietary models. We can expect a similar battle play out as in the erstwhile space of public vs private clouds. 
• Multi-modality will become table stakes although the future will be dominated by specialized models like Small Language Models (SLMs) which will be focused on specific industry domains or functions. 
• Market Consolidation like in the world of IoT Platforms
• Multi-layered stack for Gen AI (GPUs/Chips, LLM, Platform & Tooling, Applications) with LLM layer becoming table stakes and differentiation is being driven at the application layer (from being a mere wrapper to cognitive architectures encompassing routing to LLMs, Graph/Vector DBs for RAGs, core application logic/APIs for embedding model into core business functionality). 
• Last but not the least, Agents will massively disrupt the SaaS market. We are fast moving from the current paradigm of Software as a Service to Service as a Software and Eventually Agents as a Service where everything including Websites, Mobile Apps etc. are up for disruption.