Honeywell rolls out Forge Cognition, pitching NVIDIA-powered “physical AI” for buildings and factories

Honeywell Technologies is introducing Forge Cognition, a new AI and simulation platform aimed at helping operators run facilities and industrial operations more efficiently. The company announced the product in a market update dated October 14, 2026 from Charlotte, North Carolina, describing Forge Cognition as a self-contained, agentic artificial intelligence (AI) system that is paired with simulation capabilities for operational planning and execution.

In the announcement, Honeywell framed Forge Cognition as part of a broader push toward “physical AI,” a term generally used to describe AI systems that can reason about real-world environments and operational conditions, rather than only analyzing digital data. By combining agentic AI with simulation, Honeywell appears to be targeting workflows where decisions must account for physical constraints such as equipment behavior, building conditions, and production or maintenance schedules.

Honeywell said Forge Cognition was developed with technology from NVIDIA and Dell Technologies. The collaboration suggests Honeywell is aiming to bring together compute hardware and AI software building blocks with Honeywell’s own operational domain knowledge for facilities and industrial settings. NVIDIA’s role, based on the announcement wording, is tied to the AI technology stack behind the platform, while Dell is referenced in connection with enabling infrastructure for the solution.

While the announcement describes Forge Cognition at a high level, it does not provide additional implementation details in the material provided here. There is no disclosure of specific training datasets, model architectures, deployment form factor (for example, on-premises versus hosted), or how the system connects to building management or industrial control systems. Honeywell also does not specify what operational outcomes Forge Cognition is currently demonstrating, such as energy savings, reduced downtime, or improved throughput.

The company’s choice to emphasize a “self-contained” platform is likely intended to announcement that Forge Cognition is designed to run as a packaged solution, rather than requiring each customer to assemble disparate tools and services. In practical terms, that positioning can matter to facility operators because integrations, security reviews, and change management can be major sources of delay in AI rollouts.

Forge Cognition’s agentic framing also points to a shift in expectations for enterprise AI. Agentic AI refers to systems that can take multi-step actions toward goals, rather than only answering questions. In an operations context, that could mean moving beyond static recommendations toward workflows that initiate tasks, test scenarios in a simulation environment, and then translate results into operational steps. However, the announcement does not spell out which actions are automated and which remain under human control.

For NVIDIA, the development provides another example of how large AI vendors are being pulled into “edge-to-enterprise” use cases that extend past pure analytics. For Honeywell, it reinforces the company’s ongoing interest in applying AI to industrial and building operations, which is a market where reliability, safety, and integration with existing equipment often determine adoption.

One key limitation in the publicly available details provided here is that Honeywell did not outline performance metrics, pricing, contract terms, or a named customer rollout in the material reviewed. It also does not clarify whether Forge Cognition is available immediately, whether it is tied to specific Honeywell software offerings, or how general it is across different facility types.

What to watch next is whether Honeywell and its partners provide additional specificity, such as supported hardware and software environments, integration points, and quantified results from pilot deployments. Market participants will also likely look for clarity on how Forge Cognition handles data governance and operational safety, since agentic systems interacting with physical-world workflows can require tighter controls than conventional decision support tools.