New research from Berkeley Research Group (BRG) shows the industry is moving beyond experimentation. At the same time, it highlights a more fragmented reality: adoption is accelerating, but not along a single, consistent path.
That shift comes as energy systems face mounting pressure. Rising electricity demand, driven by data centers, electrification, and digital infrastructure, is forcing companies to rethink how they forecast, trade, and manage supply. In that context, AI is becoming less of a back-office tool and more of a core operational capability.
Energy companies are no longer treating AI as a side project. BRG’s survey finds that 95% of executives report at least moderate implementation, with around 40% already applying AI to core commercial and operational functions.
In trading environments, AI is helping process large volumes of market data faster, improving both speed and accuracy in decision-making. Forecasting tools are also advancing, enabling more precise predictions of supply and demand—an increasingly critical capability as renewable generation introduces more variability into the grid.
Beyond market-facing functions, companies are seeing gains in areas like cybersecurity and internal operations. AI is being used to flag anomalies, streamline workflows, and reduce manual workloads. These improvements are incremental on their own, but together they point to a broader shift: AI is becoming embedded in day-to-day decision-making rather than sitting on the periphery.
Despite widespread adoption, the way AI is being used differs sharply between fossil fuel and clean energy companies.
Clean energy firms are still building out capabilities in areas where AI could deliver significant value. Fewer than half have implemented AI for resource forecasting, and adoption remains limited in energy storage and delivery optimization. Given the variability of renewable generation, these gaps represent both an opportunity and a challenge tied to data complexity and system integration.
Fossil fuel companies, meanwhile, appear to be moving more cautiously in certain operational applications. AI deployment in asset operations remains relatively limited, particularly in technically complex areas such as reservoir modeling and well-performance optimization. These use cases require specialized datasets and domain expertise, which can slow broader adoption.
The result is an industry evolving along parallel tracks. Rather than a single adoption curve, AI is developing in ways that reflect each segment’s technical constraints, asset base, and immediate priorities.
Even as adoption expands, several constraints continue to slow progress.
Cybersecurity and data privacy concerns are among the most frequently cited challenges, reflecting the sensitivity of both operational systems and market data. As AI becomes more embedded in critical infrastructure, these risks are becoming harder to ignore.
Data-related issues remain another major obstacle. Many companies still struggle with fragmented systems, inconsistent data quality, and limited accessibility. Without a strong data foundation, even advanced AI models have limited impact.
At the same time, a shortage of skilled talent is making it harder to scale AI initiatives. Organizations are competing for professionals who can bridge technical development with real-world energy applications. Regulatory uncertainty adds further complexity, with many companies still adapting their AI governance frameworks to evolving rules.
Taken together, these factors suggest that while AI adoption in energy is accelerating, its long-term impact will depend on execution.
