As digital solutions underpinned by artificial intelligence make their presence felt across energy, petrochemicals and heavy industry segments, a fascinating, fast moving and potentially lucrative facet of this shift is coming into the foreground – agentic AI.
To the uninitiated, AI comprises of two strands – generative and agentic. Generative AI platforms – now considered ubiquitous in our digital world – use models to learn from inputted data and structures to produce tailored results. However, agentic AI extends beyond being merely a platform that’s reactive to instructions but one that’s proactive and capable of complex problem-solving with limited human intervention.
Industrial theorists, coding experts and software developers have been working on various iterations and use cases of agentic AI with energy companies for nearly a decade, even though it wasn’t called that prior to 2024.
But it truly entered the energy sector’s collective consciousness in November that year after Abu Dhabi’s state-owned energy company ADNOC announced it had started a pilot to deploy agentic AI, in partnership with Microsoft and AIQ. Several majors, including the world’s leading integrated energy companies, followed suit in announcing plans and pilot programmes of their own.
The three core drivers behind their embrace of agentic AI are greater efficiency, scalability, and sustainability integration and decarbonisation.
Going agentic across the industrial landscape
While use cases for agentic AI extend across the industrial landscape, experts at the recently concluded OPTIMIZE 26 Conference in Houston in the United States told Energy Connects that deployment across four silos is expanding rapidly. These include autonomous grid management, predictive maintenance, field operations and asset management, and energy trading and management.
“We are routinely deploying tools including machine learning and predictive analysis for production efficiencies and business flexibility. These deployments are aligned with our transformation plans in Europe where we are restructuring our base chemicals business.”
- Adriano Alfani, CEO of Versalis, subsidiary of Eni
Emanuelle Brechet, Vice President of Data Technologies at TotalEnergies, said it’s all about value creation. “Our core reason for leveraging AI applications is creating value across operational silos. This takes many forms, ranging from boosting operational excellence to expanding revenue, reducing emissions to safety improvements. AI is at the heart of our strategic digital programmes – including integrated power management, digital plants, and digital solutions for health, safety and training,” he told Energy Connects.
Over the next two years, TotalEnergies will invest $1 billion to “deploy AI at scale” to unlock additional revenue across its operations, Brechet added.
Adriano Alfani, CEO of Versalis, a chemical subsidiary of the Italian energy major Eni, said it was routinely deploying tools including machine learning and predictive analysis for production efficiencies and business flexibility. “These deployments are aligned with our transformation plans in Europe where we are restructuring our base chemicals business.”
It involves developing new value chains for biochemicals, oilfield and circularity products in Europe, where high energy costs and operational complexities were seen to be impacting Versalis’ headline performance.
A competitive differentiator in a VUCA world
In a volatile operating climate, energy and process industries have come to view technology and agentic AI-premised solutions as a competitive differentiator, said Claudio Fayad, CTO of Emerson’s AspenTech business which develops of industrial software and AI products.
In an exclusive interview with Energy Connects, Fayad said: “We have accelerated our investment in industrial AI, AI-driven asset performance management tools and the agentic environment over the last 15 months. This amplification of agentic AI – with appropriate guardrails – is designed to serve as the ultimate accelerator of process efficiencies for end customers.”
Explaining its critical need, Fayad noted: “Industrial agentic AI feeds off operational technology (OT) data, which is fundamentally different from IT data. Generic AI is not equipped for 24-7 plant operations unlike industrial AI for which context and complex OT data processing is non-negotiable.”
This is particularly true as energy and heavy industries move from unit-based process optimisation to enterprise-wide optimisation. That’s why most, including the world’s top 20 energy companies by market capitalisation, continue to pour billions into agentic AI and “co-innovate” with the software industry.
Spending billions on AI to unlock trillions?
It has led to lofty market projections. An aggregation of various industry assessments (e.g. Fortune Business Insights, GVR, Dimension Market Research, Capgemini, and more) intertwined with industry anecdotal evidence suggests a current agentic AI market valuation in the range of $7 billion to $10 billion.
That’s around 5% of an overall industrial AI and automation market valued at $200 billion in 2025. However, while the headline market is expected to grow at a compound annual growth rate (CAGR) of around 12% to $570 billion by 2034, forecasters predict the agentic AI market could be worth anywhere between $150 billion and $200 billion, with a CAGR of over 40%.
Such projections deserve a healthy dose of scepticism, but the direction of travel is hard to dispute. Emerson’s acquisition of AspenTech last year, which followed AVEVA’s by Schneider Electric in 2023, were both big-ticket deals by major industrial vendors to bolster their industrial AI product footprints.
“Industrial agentic AI feeds off operational technology (OT) data, which is fundamentally different from IT data. Generic AI is not equipped for 24-7 plant operations unlike industrial AI for which context and complex OT data processing is non-negotiable. … Moving forward with advances in industrial AI, we see over $1 trillion in value unlocked by 2036.”
- Claudio Fayad, CTO of Emerson’s AspenTech business
On May 11, AspenTech fired the industry’s latest agentic AI salvo with the launch of its platform AVA. Fayad said AVA brought "agentic, domain-aware AI capabilities" that can be “embedded directly and safely in operations."
It’s platform agnostic and omnichannel. As such AVA connects to large language models but more importantly carries specific industrial knowledge and skills. “That makes the large language models operate as an industrial persona curated from the domain knowledge of AspenTech, Emerson and the users’ own operations.”
It is expected to evolve to “create, deploy and orchestrate agents and specific advisors” pertaining to the end users’ operations.
Accelerating AI adoption
Nuno Pacheco, Senior Optimisation and Control Engineer, at Repsol, who is leading the tool’s pilot at the company, said: “It is a practical way to accelerate AI adoption to deliver repeatable and scalable impact across our operations.”
“Having deployed AVA, we see both this tool and our AI journey as an amplifier of the industry knowledge we have at Repsol. Essentially, it’s about bringing together engineering first principles and wider quality data to feed it into the AI for more efficient and beneficial outcomes.”
Other vendors in the market from Honeywell to Siemens, ABB to Yokogawa are also constantly bringing their competing, and often platform agnostic products to market. Fayad said the cycle of innovation, product updates and introduction of new products continues to shorten in the cutthroat world of industrial AI.
TotalEnergies, Versalis and Repsol joined many of their peers in noting that billions spent on agentic AI will likely result in trillions worth of throughput gains and efficiency savings. Fayad said $500 billion in real economic value was created over the previous 12 years through automation and optimisation. “Moving forward with advances in industrial AI, we see over $1 trillion in value unlocked by 2036.”
So, while there may be calls for being disciplined on the AI hype and an understanding of its limitations, with much at stake in a competitive climate, the industry isn’t holding back on an agentic AI-driven transformation.
Energy Connects includes information by a variety of sources, such as contributing experts, external journalists and comments from attendees of our events, which may contain personal opinion of others. All opinions expressed are solely the views of the author(s) and do not necessarily reflect the opinions of Energy Connects, dmg events, its parent company DMGT or any affiliates of the same.
