Krones cuts simulation run times by up to 98% with AI agents and Azure High-Performance Computing

A leader in the food and beverage industry, Krones uses computer simulations to evaluate how factors such as speed and temperature impact equipment operations. However, hardware constraints contributed to lengthy run times and created a bottleneck.

Krones teamed up with Microsoft and other technology solution providers to harness the scale and security of Microsoft Azure to streamline computer simulations. The team developed AI agents that run and refine models and create digital twins.

Krones decreased simulation run times from three to four hours to five minutes or less, a reduction of up to 98%. The company expects to realize significant savings in labor and material costs once the pilot is expanded to as many as 500 engineers.

“It was an outstanding experience on how quickly things can run with the right partners.”

Dr. Johannes Bernstein, Head of Product Life Cycle Management Strategy, Krones

Krones is an industry-first mover in AI-driven manufacturing, slashing simulation run times by up to 98% using AI. To model the process of filling beverage bottles, Krones worked with experts from Microsoft, Ansys (part of Synopsys), CADFEM, NVIDIA, and SoftServe to develop AI agents and physically accurate digital twins. The company reduced simulation run times from three to four hours to five minutes or less and has ambitious plans to scale its program and design the future of manufacturing with AI.

Refining complex engineering designs with computer simulations

German company Krones has been a global leader in equipment manufacturing for the food and beverage industry for 75 years. Computer simulations are a valuable tool to model equipment operation, but hardware constraints and lengthy turnaround times made it difficult for Krones engineers to quickly run multiple scenarios that could refine equipment designs in real time.

“There was a single PC or a hardware station in a room, running a software on it, which always had to be customized for us,” says Dr. Johannes Bernstein, Head of Product Life Cycle Management Strategy at Krones. “There was no copy of the hardware station, no cloud technology, and no real scalable calculation power.”

With its on-premises infrastructure, Krones could only make incremental improvements in turnaround times and calculation power. An impromptu discussion with Microsoft in 2025 became the starting point for a whole-scale transformation of Krones’s simulation capabilities in collaboration with a group of industry-leading technology providers.

Developing AI agents and digital twins in just five months

Krones and Microsoft decided to begin refining simulations of the bottle filling process. “The filler simulation connects the sustainability and rotation of machines with fluid simulation, which is one of the most complicated simulation topics we have because of the fluid’s nonlinear behavior,” says Dr. Bernstein.

Krones’s new simulation solution uses the scalability and security of Microsoft Azure and Azure High-Performance Computing to manage compute-intensive workloads. Together with Ansys, CADFEM, NVIDIA, and SoftServe, Krones developed a physically accurate digital twin application that incorporates AI agents that test and refine model scenarios.

Within just five months, Krones moved from point of concept to a pilot program rolled out to about 30 engineers and is planning to scale the solution for as many as 500 more. “It was an outstanding experience on how quickly things can run with the right partners,” says Dr. Bernstein.

Slashing simulation run times by up to 98%

This holistic, agentic AI–powered approach reduced simulation run times by up to 98%, from three to four hours to five minutes or less. The digital twin is also a powerful new tool for the entire organization and its customers.

“Before, the simulation gave you three or four calculation numbers and a probability of the fluid spilling,” says Dr. Bernstein. “Now, you have a complete 3D animated model of how the bottles are filled. It’s a tool for a salesperson who is selling fillers, for a classical engineer, or for people who service machines for customers.”

The company anticipates significant cost savings as the pilot expands to more scenarios, such as bottle blowing for plastic bottles. With improved simulations, Krones expects to reduce the number of physical prototypes it builds, saving labor and material and decreasing waste.

“Before, the simulation gave you three or four calculation numbers and a probability of the fluid spilling. Now, you have a complete 3D animated model of how the bottles are filled. It’s a tool for a salesperson who is selling fillers, for a classical engineer, or for people who service machines for customers.”

Dr. Johannes Bernstein, Head of Product Life Cycle Management Strategy, Krones

Designing the future of engineering with AI

For Dr. Bernstein, now is the time to embrace AI: “We have to design the future and show that we have one of the best engineering situations in Western countries.” Krones is already preparing a bold effort to consolidate its data management across the engineering life cycle to inform decision-making.

Collaboration with the right partners at the right time is the heart of Krones’s success, brought together in the Microsoft ecosystem. “All these partners are necessary: Krones, Microsoft, NVIDIA, SoftServe, CADFEM, and Ansys,” says Dr. Bernstein. “We have the right partners in the boat, so this is the perfect setup for us.”

Discover more about Krones on Facebook, Instagram, LinkedIn, X/Twitter, and YouTube.

“All these partners are necessary: Krones, Microsoft, NVIDIA, SoftServe, CADFEM, and Ansys. We have the right partners in the boat, so this is the perfect setup for us.”

Dr. Johannes Bernstein, Head of Product Life Cycle Management Strategy, Krones