Artificial intelligence is moving beyond office software and into the physical infrastructure of the global economy. A new agreement between ADNOC, OpenAI and G42 aims to connect advanced AI systems with energy production, field development, refining, logistics and trading, highlighting the growing role of AI in industrial decision-making.
◉ ADNOC will deploy OpenAI’s advanced AI technology to connect operational and commercial decisions across its energy business.
◉ The partnership will combine industrial data, AI models and UAE-based computing infrastructure to support complex energy-sector decisions.
◉ The agreement highlights a wider investment theme: AI growth increasingly depends on electricity, data centres and the infrastructure needed to operate them.
A New Phase of Industrial AI
On November 2, Abu Dhabi National Oil Company (ADNOC), OpenAI and technology group G42 announced a multi-year partnership to deploy advanced AI across ADNOC’s operations.
The agreement, announced at the ADIPEC energy conference in Abu Dhabi, is designed to connect intelligence across different parts of the energy value chain rather than treat each operation as a separate system.
According to the announcement, the partners will develop a decision-making capability using OpenAI’s Frontier platform, ADNOC’s industrial data and expertise, and G42’s AI computing infrastructure based in the United Arab Emirates.
The financial terms of the agreement were not disclosed.
The partnership is significant because it moves AI from individual productivity applications towards decisions involving complex physical assets, capital allocation and commodity markets.
From Oilfields to Commercial Decisions
The initial deployment will focus on two applications.
The first involves integrated field development planning. Energy companies must coordinate information about underground reservoirs, wells, surface facilities, engineering requirements and project economics before deciding how to develop an oilfield.
ADNOC says the new system will connect decades of data across thousands of wells, multiple reservoirs and hundreds of surface facilities. The stated objective is to prepare and optimise complex development plans in hours rather than months.
The second application focuses on the commercial value of energy production. Interconnected AI agents are intended to assess information across production, processing, refining, logistics, sales and trading.
By combining operational forecasts with market conditions, costs and expected margins, the system is designed to help decision-makers evaluate how a change in one part of the business could affect the wider operation.
These are announced objectives, not independently verified financial results. The partnership’s eventual value will depend on implementation, data quality, operational reliability and measurable improvements in business performance.
AI Infrastructure Becomes Strategic
The agreement also illustrates a broader challenge facing the AI industry: increasingly capable software requires substantial physical infrastructure.
Data centres need reliable electricity, cooling systems, specialised chips, network connections and access to capital. Industrial AI introduces another layer of complexity because its outputs must work within real operational, engineering and safety constraints.
The International Energy Agency reported that global data-centre electricity consumption rose by 17% in 2025, while electricity use at AI-focused data centres increased even faster. Its analysis projects that total data-centre electricity consumption could approach 950 terawatt-hours by 2030.
These projections underline why energy supply has become a strategic issue for the technology industry. At the same time, energy companies are exploring how AI can improve planning, asset utilisation, maintenance and commercial decisions.
The relationship runs in both directions: AI requires energy infrastructure, while energy companies increasingly want AI to make their existing infrastructure more productive.
What It Means for Investors
The partnership connects several major investment themes without guaranteeing that any particular company will benefit financially.
OpenAI represents the AI technology layer, while G42 provides the computing infrastructure described in the agreement. ADNOC brings the industrial assets, operational expertise and proprietary data needed to apply the technology to real-world energy systems.
For public markets, the wider theme extends to semiconductor manufacturers, data-centre infrastructure providers, power-equipment companies and energy producers. Companies exposed to electricity generation, grid expansion, cooling systems and advanced computing may also be relevant to the broader AI infrastructure story.
However, the commercial implications differ considerably across these sectors. A rise in AI adoption does not automatically translate into higher profits, and increased infrastructure spending can create financing, execution and valuation risks.
For industrial users, the key question is whether AI can deliver measurable improvements in productivity, reliability and capital efficiency. For technology providers, the challenge is converting advanced capabilities into durable enterprise revenue.
The Bigger Picture
ADNOC’s agreement with OpenAI and G42 is part of a wider shift towards AI systems that coordinate complex business processes rather than simply generate text or answer questions.
Energy is a particularly important testing ground because decisions involve large capital commitments, technical uncertainty, volatile commodity prices and infrastructure that must operate reliably.
The partnership also highlights the United Arab Emirates’ ambition to strengthen its position in both energy and AI infrastructure.
Ultimately, the significance of this development will depend on execution. If AI can connect industrial data and improve decisions across an entire energy value chain, the model could become relevant well beyond oil and gas. If deployment costs, technical limitations or operational risks outweigh the benefits, adoption may prove slower than the industry’s ambitions suggest.
