Power in the Time of AI: How Intelligent Transformation is Reshaping Energy

The AI boom is projected to nearly triple data center power requirements by 2035, enabling amazing new capabilities while also placing the energy sector in a precarious position to meet these growth projections. Simultaneously, these players are grappling with the challenges related to rising security risks, aging infrastructure, increased adoption of distributed energy resources, and severe weather events.

Today’s transformation of the energy domain represents the largest operational and systemic overhaul since the days of Edison and Westinghouse, demanding solutions to enable a safer, more affordable, reliable, resilient, and sustainable energy ecosystem for everyone. These challenges are too vast for human teams and traditional automation systems alone.

For the last two years, AI has proven its business case in our sector, generating positive returns through pilots and proofs-of-concepts. But in 2026 the competitive line in the energy sector will no longer be drawn at AI adoption, but at the successful deployment and orchestration of multi-agent systems to address the industry’s biggest problems, transforming complex workflows into autonomous, self-optimizing systems. From safety to work management, customer engagement, asset management, and planning, multi-agent systems will transform how work is done.  Firms moving past simple chatbots and single-task automation to coordinated, specialized AI-driven systems can finally address the hurdles that have limited progress in the past, making energy safer, more affordable, reliable, resilient, and sustainable for everyone.

The Agentic-Driven Transformation of Energy

The adoption rate of agentic systems within the industry demonstrates the sector is increasingly moving past single-task automation. The focus is no longer on if agents can handle complex workflows, but on scaling and orchestrating them to manage complex planning, construction, operational, and back-office functions.

For energy companies, this means moving beyond simple tasks. An orchestrated multi-agent system can manage entire workflows, as companies like NextEra and Westinghouse are demonstrating:

• A Data Collection Agent collects, categorizes, and contextualizes data collected from a wide variety of field assets, internal databases, and third-party data sources.
• A Risk Agent evaluates collected data and identifies a potential failure point of an asset that may happen in the near future.
• A Dispatch Agent checks current operations, technician schedules, inventory, and weather conditions to determine the right response to the risk.
• A Trading Agent automatically adjusts power generation schedules and/or triggers hedges  to cover the anticipated loss of resources while the repair occurs.
• A Safety Agent determines the risk of the work and validates it against the technician’s training, vacation and sick leave schedule, and skill level, developing a custom, interactive safety briefing.
• A Regulatory Compliance Agent logs all decisions against regulatory mandates and creates an auditable report.
• A QA/QC Agent validates the work was performed properly and the system will operate as expected when put back into service.

This shift from single-task automation to full-workflow orchestration will become the bedrock of energy companies’ future capabilities. By mastering this foundational capability set on a stable, data-rich environment, the industry can tackle some of its most complex challenges, like managing interconnection queues to simultaneously enable new load growth and large-scale decarbonization, work management that can significantly reduce construction costs and times,  asset management that can enhance reliability and the bottom line, and hyper-personalized, proactive, multi-channel customer engagement. 

The next frontier is solving the engineering challenges underpinning the energy transition. The speed of discovery—from materials science for next-generation batteries and solar panels to optimizing the complex physics of carbon capture—has always been a critical bottleneck.

AI agents are transforming this by acting as "self-driving research partners," moving beyond simple data analysis and into the full lifecycle of an experiment – from setting an objective (e.g., "Find a new electrolyte combination with 10% higher energy density and 5-year cycle life"), to designing the experiment, executing simulations across a compute cluster, analyzing the results, and dynamically adjusting the next steps. Now, with proper oversight, AI is able to execute a process that would take human teams weeks or months.

By synthesizing information from thousands of studies and integrating it with real-time experimental data, specialized agents can accelerate deep-tech decarbonization, giving early adopters a significant competitive advantage.

Security and Trust As the Foundation for Regulated, AI-Enabled Energy Infrastructure

AI transformation can still succeed for highly-regulated companies like those in the energy sector, where critical infrastructure subject to stringent security requirements is central to their operations. My experience working with companies like Westinghouse showed me how even the most stringently regulated entities can harness the power of AI without compromising on safety or compliance. Westinghouse operates within a complex web of global regulatory frameworks, including oversight from the U.S. Nuclear Regulatory Commission (NRC) and adherence to strict standards like 10 CFR 73.55 for physical and cyber security. In this environment, where physical security can account for up to 20% of a site’s workforce, every digital tool must meet rigorous "Safeguards Information" (SGI) requirements and protect against cascading cyber-physical threats.

To meet these "non-negotiable" security demands, Westinghouse implemented a fully air-gapped version of Distributed Cloud (GDC). This allowed them to deploy a complete, managed cloud stack—including advanced AI and machine learning tools—in a standard data center that remains permanently disconnected from the public internet and any external cloud backbone. By providing this "secure-by-design" infrastructure, energy leaders can run specialized AI models like Westinghouse’s HiVE and Bertha, which process 75 years of proprietary nuclear data, within a sovereign environment that satisfies the world’s most demanding regulatory and export-control mandates.

As AI agents move from assisting with back-office paperwork to managing the operational state of the grid, absolute trust in the system's security, privacy, and auditable governance becomes a non-negotiable requirement. The complexity of multiple agents interacting with industrial Supervisory Control and Data Acquisition (SCADA) systems, energy market platforms, and consumer data introduces enormous risk if the foundation is flawed.

For this heavily regulated sector, the shift to agentic systems mandates a "clean data" strategy and a a secure-by-design, auditable, and compliant-by-default infrastructure, which includes:

• Data lineage and control: Ensuring full customer ownership and control over data, and guaranteeing proprietary information is never used to train the foundational models.
• Auditability: Implementing an orchestration layer that logs every decision and action, providing an ironclad audit trail for regulators and security teams.
• Threat immunity: Using autonomous security agents to constantly harden infrastructure against evolving cyber-physical threats, effectively fighting AI with AI.

The firms that prioritize this foundation now will be the only ones capable of scaling the high-value, autonomous operations and R&D of the future.

The energy industry is the perfect proving ground for the most sophisticated AI systems yet devised. The convergence of multi-agent systems with petabytes of operational data promises not just incremental gains, but the ability to accelerate energy security, revolutionize operational efficiency, and drive sustainable growth. The key to realizing this future lies in the implementation of complex, multi-agent systems that empower the grid to become autonomous, resilient, and adaptive. More than just a technological upgrade, mastering these systems means the energy sector is pioneering the practical, real-world steps toward the arrival of AGI, eventually turning over-hyped theory into reality.