PepsiCo tests AI-driven digital twins to reinvent manufacturing

USA – PepsiCo has launched a multi-year, “industry-first” collaboration with Siemens and Nvidia to deploy advanced digital twin technology and artificial intelligence (AI) across its packaging and bottling operations, aiming to significantly improve line speed, flexibility and overall efficiency.

Early pilot projects are already underway at PepsiCo facilities in the US, with broader global rollout expected as capabilities mature.

At the heart of the collaboration is the integration of Siemens’ physics-based digital twin technology with Nvidia’s accelerated computing and Omniverse AI libraries.

Together, the partners are combining 2D and 3D digital twin data to create highly detailed, photorealistic virtual representations of PepsiCo’s packaging lines, enabling faster decision-making and predictive optimization.

John Nixon, vice president for Infrastructure, Life Sciences, Consumer Products and Retail at Siemens Digital Industries Software, says the partnership reflects a shift in how packaging operations are managed.

Digital twins, he notes, are evolving from engineering tools into live operational systems that reduce waste, improve sustainability metrics and enhance agility in high-speed, high-SKU environments.

Central to the initiative is Siemens’ Digital Twin Composer, which allows teams to model and visualize entire packaging and bottling lines, from filler and capper dynamics to downstream accumulation and palletizing.

By simulating complex interactions digitally, operators and engineers can test format changes, identify bottlenecks and optimize flow before making physical adjustments on the factory floor.

Nvidia’s role focuses on AI and computer vision, which continuously analyze high-frequency operational data such as machine cycle times, micro-stoppages and quality rejects.

According to Tarik Hammadou, director of Supply Chain Developer Relations at Nvidia, AI models can distinguish true system constraints from symptoms, predict where bottlenecks will emerge and recommend corrective actions in real time.

Computer vision further enhances digital twins by capturing real-world variability that traditional sensors often miss. In packaging and palletizing, this includes tracking product orientation, load stability, mixed-SKU stacking and collision risks.

As simulations learn from live data, virtual models become “living systems” capable of predicting failures and optimizing layouts and robot motion before deployment.

A key advantage of the approach, Siemens says, is ease of adoption in brownfield environments. Digital Twin Composer supports legacy equipment using a mix of existing CAD data, point-cloud scans and standardized simulation templates, allowing manufacturers to start without perfect datasets.

The PepsiCo collaboration mirrors a wider industry push toward AI-enabled manufacturing.

Recent initiatives include other global FMCG players piloting digital twins to cut energy use and improve overall equipment effectiveness, while major automation suppliers are embedding AI agents into packaging lines to dynamically balance throughput, labour and maintenance.

Together, these developments signal a growing shift toward fully data-driven, predictive packaging operations across the food and beverage sector.

Subscribe to our email newsletters that provide busy executives like you with the latest news insights and trends from Africa and the World. SUBSCRIBE HERE