PepsiCo achieved a 20% throughput increase at their Gatorade plant without building a single new square foot.
And they found that capacity by running the factory virtually first.
Before we get into how, quick question:
How does your plant currently validate a new line or layout change before committing capital?
👋🏻 I'm Leonardo Ubbiali. This week we're looking at why the economics of building large central plants are breaking down, and why the manufacturers replacing that model are building differently by doing it in the virtual world first.
88% of manufacturers remain moderately or very concerned about tariff impacts on their investment decisions.
Most of them are just waiting, but the market is not waiting with them.
A manufacturer I spoke to recently had been planning a significant facility expansion as well.
But the tariff exposure on imported components broke the ROI model in two separate scenarios, so the project is on hold.
What running it virtually looks like
With demand rising and new construction taking years, they needed more throughput without locking in cost assumptions that might not hold.
So they ran the factory virtually.
Using Siemens Digital Twin Composer, built on NVIDIA Omniverse libraries, PepsiCo recreated every machine, conveyor, pallet route, and operator path in their Gatorade facility with physics-level accuracy.
AI agents ran thousands of layout and flow simulations, and they identified up to 90% of process issues before any physical change was made.
The result was a 20% increase in throughput within three months, with no new construction or additional square footage.
PepsiCo estimates a 10 to 15% reduction in capital expenditure by finding and unlocking capacity that already existed in facilities they already owned.
At Hannover Messe 2026, Microsoft described Krones as turning "a traditional machinery business into a Frontier digital services company."
The agentic digital twins Krones developed cut simulation times from four hours to five minutes.
AI agents autonomously optimise configurations and transfer results directly to physical machines.
The customer gets a line that reconfigures as demand changes, and Krones gets a recurring service relationship rather than a cyclical capital dependency.
Siemens demonstrated at Hannover Messe 2026 that once you can validate any layout virtually, the fixed central plant stops being the only option.
A modular, AI-orchestrated line validated in the twin, assembled physically, then reconfigured again when demand shifts.
And it didn’t require months of retooling and hours of simulation.
The manufacturers capturing this shift are building factories that can move.
Five things you can do this quarter
The problem: You have a capital investment decision this year and need to understand whether virtual validation could reduce or replace your physical pilot process.
What you need: Current validation process, timeline pressure, and the key variables you need to test.
The Prompt (copy this):
I'm a [YOUR ROLE] at a [PRODUCT TYPE] manufacturing company. We are evaluating a [DESCRIBE CAPITAL PROJECT OR LAYOUT CHANGE] and need to understand whether virtual validation could reduce or replace our physical pilot process.
Current validation process: [DESCRIBE] Timeline pressure: [WHEN YOU NEED TO COMMIT] Key variables to test: [LIST LAYOUT, THROUGHPUT, EQUIPMENT SCENARIOS]
Tell me:
What would a digital twin validation process look like for this specific use case?
Which parts of our current physical pilot could realistically be replaced by virtual validation?
What data and infrastructure would we need to run this?
What are the risks of relying on virtual validation for this decision, and how do manufacturers mitigate them?
What you'll get back:
A step-by-step virtual validation process for your use case, an honest assessment of where physical validation is still necessary, and the infrastructure requirements to get started.
PepsiCo + Siemens + NVIDIA: Full Collaboration Case Study
The supply chain simulation section covers how PepsiCo is using AI agents to test hundreds of facility configurations before committing to physical builds. If you have a capital investment decision this year, this is the clearest public example of what virtual-first planning looks like in practice.
Time to value: 5 minutes
Hit reply: PepsiCo found 10 to 15% hidden capacity without building anything new. Do you know what that number looks like in your facilities?
I read every email.
Leo