The Messy Reality of Supply Chain Automation

A Gartner survey of executives found that eight in 10 expect autonomous business to become dominant by 2030. That’s less than four years away.

During the Gartner Supply Chain Symposiums in Orlando and Barcelona, I had dozens of conversations with supply chain executives that echoed this prediction. The sentiment was similar, but the path and execution to this end were anything but clear.

CSCOs came to a supply chain conference, but AI dominated most of the topics and conversations. This led to further reflection on definitions and the impact of technology on the enterprise.

All agreed that AI has been embedded in their supply chain processes for at least a decade. Machine Learning is a widely used technique in the AI toolbox and can automate forecasting. We don’t buy ML, but we do buy forecast and demand management platforms with robust ML capabilities.

But when it comes to using the output of the demand management platforms to drive autonomous activity, that was a different story. Some companies do, but most don’t have autonomous processes on the backend of their demand plan.

Many recognize that their current processes can’t be automated, let alone autonomous. They need a complete redesign to achieve high levels of automation.

Many of CSCOs I spoke with told me that a large majority of the current workforce lacks the skills to help the enterprise achieve high levels of automation and autonomous work. But they also said this is nothing new.

Every new technology implemented in supply chain has required upskilling and retraining, resulting in turnover.

The question this time is whether a new type or class of talent eliminates other roles entirely, resulting in breakthrough cost reductions.

Gartner research isn’t showing this yet. In the second half of 2025, about 10% of layoffs were attributable to AI.

Is there a bottom line to all of this? The consensus is: not yet.

In May of this year, while testing a new internal AI model at OpenAI, they asked for a solution to the Erdős Unit Distance problem.

The problem had remained unsolved since 1946. It was solved in about 32 hours and $1,000 in tokens. OpenAI said that the result was unthinkable a year ago.

So, the answer to the question of “How fast?” is “Fast,” and the 2030 date in the prediction is reasonable? Well, not so fast …

CSCOs agree that change leading to improvements will come faster than ever before. The challenge is not necessarily technology. It is more about adoption.

And no one failed to mention data quality.

Many companies find that even when a data field is populated correctly, the values may differ across business units or regions. And many companies are finding that the field isn’t populated correctly at all. These simple issues result in slower change.

No doubt AI will help solve these problems, but how fast remains a moving target.

Headline news about cost reductions driven by AI implies that automation or autonomous processes are rapidly expanding across companies. The C suite can see this as an opportunity in their company. This results in pressure across all functions to make gains equally.  

CSCOs are all looking for the best use case, but supply chain operates in the messy physical world.

CSCOs tell us that some areas are more ripe for automation and autonomous operations than others. Customer service and order management/configuration rise to the top of the list.  

Attention has shifted from identifying use cases to solving problems with the best tools.

Like forecasts, predictions are usually wrong. Whether autonomous business becomes the dominant model for some or all business in 2030 remains to be seen.

Supply chains have proven to be incredibly resilient and agile. War, pandemics, significant material shortages, and deglobalization have hit hard in the past 48 months, but tremendous progress has been made during that period.

How much supply chain work is automated now? That isn’t clear. What is clear is that it’s a lot more than four years ago.

How much work is currently autonomous? Very little from what we heard from CSCOs at the Orlando and Barcelona Symposiums.

In four years, we might be making a similar prediction about 2034, but however it ends up it won’t look or feel the same as it does now.