Honeywell Technologies has launched Forge Plant Cognition, an industrial AI platform developed with NVIDIA AI technologies and Dell Technologies infrastructure, with Dangote Petroleum Refinery in Lekki, Nigeria, selected as its first deployment site. The platform is designed to help refining, petrochemical and liquefied natural gas operators identify production bottlenecks, anticipate operational problems and optimize existing assets. Honeywell estimates the technology could generate more than $60 million in gross annual operational benefits for the Dangote facility at full-capacity throughput, although the projection remains preliminary and excludes implementation costs.
Honeywell Brings Industrial AI to Dangote Refinery
Industrial AI is moving beyond predictive analytics toward systems that combine plant data, engineering knowledge and physics-based models to help operators make decisions about complex physical processes. Honeywell Technologies is advancing that approach with Forge Plant Cognition, a platform designed to connect AI capabilities with the process expertise required to operate large-scale industrial facilities.
Built on Honeywell’s Forge Cognition platform, the new offering combines industrial automation, process technology knowledge and AI infrastructure from NVIDIA and Dell Technologies. Its initial deployment is planned at Dangote Petroleum Refinery in Lekki, Nigeria, where the companies aim to improve throughput, operational performance and the capabilities of local engineering teams.
The launch illustrates an emerging direction in industrial AI: using AI systems not just to identify patterns in operational data, but also to interpret those patterns against engineering constraints and help translate findings into practical decisions.
From plant monitoring to AI-assisted optimization
Industrial facilities generate large volumes of operational information, but identifying the cause of a production bottleneck can require expertise spanning equipment, process conditions, product specifications and interconnected plant systems.
Forge Plant Cognition is designed to combine real-time operating data with Honeywell UOP’s process expertise, trusted calculations and physics-based models. According to Honeywell, this approach can help operators identify constraints, anticipate problems and evaluate actions intended to improve throughput, yield and product value.
The platform targets three broad operational priorities: uncovering hidden production margin, protecting output by detecting emerging equipment or process issues, and making specialized engineering knowledge more accessible to operators.
These capabilities address a longstanding industrial challenge. Conventional monitoring tools can flag abnormal conditions, but determining what caused an anomaly and which corrective action is appropriate may require experienced engineers. AI-assisted workflows could help accelerate that investigation by connecting observed conditions with technical knowledge and model-based analysis.
The distinction is important, particularly in refineries and petrochemical facilities, where an operational adjustment can affect multiple interconnected processes. Recommendations must account for equipment limits, process safety, product quality and other operating constraints. Honeywell’s emphasis on engineering expertise and physics-based models reflects the need to ground AI-generated guidance in the realities of industrial operations.
The announcement does not specify how much decision-making authority the platform will have over plant equipment or whether particular optimization actions will be executed autonomously. Its described capabilities center on identifying constraints, generating insights and supporting operational decisions.
NVIDIA and Dell provide the AI infrastructure
Forge Plant Cognition is powered by Dell AI Factory with NVIDIA and supported by Dell Services. Honeywell is combining several NVIDIA technologies to support different parts of the industrial AI workflow.
NVIDIA Nemotron open models are intended to power agents that interpret operational data. NVIDIA Omniverse libraries support the creation of digital twins of the plant and its processes, enabling simulation-based training and planning.
The platform also incorporates NVIDIA Cosmos models and the NVIDIA Metropolis Video Search and Summarization Blueprint for vision AI applications. Honeywell says these capabilities enable agents to detect anomalies and generate real-time alerts. NVIDIA PhysicsNeMo provides a framework for building and training physics-based AI models that can support digital-twin-integrated plant optimization.
Together, these technologies are intended to connect operational data, visual information, simulation and process modeling. A digital twin can provide a virtual representation of a physical facility, allowing engineers to investigate scenarios and test approaches without relying exclusively on live plant conditions.
The actual value of this architecture will depend on the quality of the underlying data, the accuracy of its models and how well its outputs fit existing operational workflows. Industrial deployments also require careful validation before AI-generated recommendations influence safety-critical processes.
Honeywell projects more than $60 million in annual benefits
Honeywell estimates that AI-enabled optimization through Forge Plant Cognition could deliver more than $0.29 per barrel in operational benefits at the Dangote refinery. At full-capacity throughput, the company projects gross benefits exceeding $60 million annually for this specific facility.
These figures are preliminary estimates, not independently verified savings. Honeywell states that the projection excludes implementation costs and is not guaranteed. Actual outcomes will depend on factors including refinery configuration, operating conditions, crude mix, utilization, baseline performance, data quality and product margins.
The estimate therefore should not be interpreted as a confirmed financial return or a result already achieved at the facility. It does, however, show how industrial AI providers are framing the commercial opportunity: even relatively small improvements in per-barrel economics can become significant when applied across a large refinery operating at high throughput.
Supporting Dangote’s expansion and local engineering workforce
Dangote Petroleum Refinery will integrate operating data with Honeywell’s process expertise and AI-based optimization capabilities. The stated objectives include identifying constraints, accelerating troubleshooting and enabling more informed decisions.
Dangote CEO and Managing Director David Bird said the company selected the platform to help optimize performance and develop local technical capabilities. The refinery also intends to use AI-enabled access to specialized process knowledge to support engineering and operational staff as the workforce expands.
The companies say this effort aligns with Dangote’s goal of developing a workforce that is 70% Nigerian, with the longer-term aim of reducing reliance on external expertise.
The deployment builds on a longstanding relationship between the companies. In a separate announcement in November 2025, Honeywell said it was helping Dangote pursue a capacity increase from 650,000 to 1.4 million barrels per day within three years. That figure represents the previously announced expansion plan, not confirmation that the target has been reached.
The wider industrial AI opportunity
Forge Plant Cognition enters a market in which industrial technology providers are seeking to connect AI models with operational technology, process engineering and asset management. The competitive challenge is not simply to deploy a capable model; it is to make its recommendations useful, traceable and safe within an industrial environment.
Honeywell’s advantage is its combination of process technology and industrial automation expertise with AI infrastructure from NVIDIA and Dell. The deployment at Dangote will provide an important setting for evaluating how this combination performs in a large, complex operating environment.
For now, the headline financial figures remain projections. Demonstrating sustained improvements in throughput, reliability and operating costs will be essential to establishing the platform’s value.
The broader direction is clear: industrial AI is increasingly being designed around the decisions operators need to make, rather than data analysis alone. Forge Plant Cognition is Honeywell’s attempt to bring that model into refinery operations, where even incremental improvements can have substantial economic consequences.
Market Landscape
Industrial AI is evolving from isolated predictive maintenance and analytics applications toward integrated systems that combine operational data, engineering knowledge, simulation and AI-assisted decision support.
Honeywell’s Forge Plant Cognition targets this convergence by connecting its industrial process expertise with NVIDIA AI technologies and Dell infrastructure. Its initial Dangote deployment focuses on refinery optimization, where throughput, equipment reliability and process efficiency have direct commercial implications.
The market’s central challenge is proving that AI insights translate into sustained operating improvements. Digital twins, vision AI and physics-based models can support analysis, but results depend on reliable plant data, validated models, operational integration and human oversight. Honeywell’s projected financial benefits should therefore be evaluated separately from independently measured outcomes.
Top Insights
- Industrial AI deployment: Honeywell is launching Forge Plant Cognition at Dangote’s Nigerian refinery to help identify bottlenecks and improve operational decisions.
- Integrated AI infrastructure: NVIDIA models, Omniverse libraries, Cosmos, Metropolis and PhysicsNeMo support data interpretation, digital twins, vision AI and physics-based modeling.
- Projected financial benefits: Honeywell estimates more than $60 million in annual gross operational benefits at full-capacity throughput, excluding implementation costs.
- Engineering expertise remains central: The platform combines AI capabilities with process knowledge and physics-based models to help ground recommendations in industrial operating conditions.
- Results remain to be demonstrated: The financial figures are preliminary company estimates, and the announcement does not establish independently verified performance gains.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
