Gen AI Is Coming for Packaging Pricing -- And Most Companies Aren't Ready for What That Means

The numbers look great on paper. A McKinsey survey of 110 senior packaging leaders says AI adoption for commercial optimization jumped dramatically between 2024 and 2025. By late 2025, a majority of respondents reported they were considering, developing, or had launched gen AI efforts for sales, pricing, and procurement. McKinsey Partner Abhinav Goel told Packaging Dive that "2026 will be all about at-scale deployment."

I read that and felt a knot in my stomach. Not because AI in packaging is bad -- it's probably inevitable and likely beneficial in the long run. But because I've spent six years as a packaging coordinator watching this industry adopt new tools, and the gap between "we're deploying a solution" and "we actually understand what it's doing to our pricing" is where the expensive mistakes live.

The Surface Problem: Everyone's Rushing In

McKinsey's survey, conducted in August and September 2025, covered all major substrates -- flexible and rigid plastics, glass, metal, and paper -- and end markets including cosmetics, e-commerce, food and beverage, industrial products, pharma, and retail. The firm noted that when they ran a larger version of the same survey in 2024, a majority of respondents hadn't taken action on gen AI. One year later, the landscape had flipped.

"It is truly becoming a big focus area," Goel said. "They are walking the talk in many ways."

That's the optimistic read. Here's the read from someone who sits downstream of these pricing decisions: when packaging suppliers simultaneously adopt AI-driven pricing tools, the competitive dynamics change for everyone in the chain -- converters, coordinators, buyers. And most of us on the buying side aren't tooled up to understand what just happened to our quotes.

The Deeper Issue: AI-Optimized Pricing Means AI-Optimized Margin Extraction

Let me be specific about where I think the pitfall lies, because the McKinsey framing -- "commercial optimization" -- sounds neutral. But optimization for whom?

According to the report, AI is enabling "more efficient and granular analysis of customer segments across different product types and time horizons." In practice, that means your supplier can now model exactly how price-sensitive you are, how much switching cost you'd incur, and what your likely alternatives look like -- all before they send you a quote. McKinsey Senior Partner Gregory Vainberg noted that competitive intelligence that "used to be desktop analysis and through word of mouth" can now happen in a fraction of the time with gen AI.

I made the mistake of thinking vendor pricing was mostly formula-driven -- material cost plus conversion plus margin. In my first two years coordinating packaging orders for our 150-person CPG company, I took quotes at face value. Then in 2022, during a resin price spike, I noticed something odd: two of our converters raised prices by different amounts on the same substrate, within the same week, citing the same raw material increase. The difference was about 7% on a $45,000 quarterly order. That's when I learned that pricing is as much about what the supplier thinks you'll accept as it is about their actual costs.

Now imagine that intuition-based price discrimination powered by machine learning models trained on thousands of contract negotiations, competitive bids, and customer behavior patterns. That's what "AI-enabled granular customer segmentation" means in plain English.

The Unstructured Data Problem Is Worse Than It Sounds

One detail from the McKinsey report that deserves more attention: Vainberg noted that packaging companies are realizing "you could actually train your models on unstructured forms of data, contracts, websites, things like that." This is being framed as a breakthrough -- companies don't need perfectly organized databases to start using AI.

But think about what "unstructured data" includes. Your RFP responses. Your publicly listed pricing. Your website's customer case studies that mention order volumes. Your trade show presentations. Your LinkedIn posts about supply chain challenges. All of that is training data for a supplier's AI model to build a more accurate picture of your willingness to pay.

After I realized how much information we were inadvertently broadcasting, I started auditing what our company was putting into the public domain during vendor negotiations. It sounds paranoid, but when the other side has AI reading your published case studies while you're still using spreadsheets to compare quotes, the information asymmetry is real.

The Barriers McKinsey Identified Are Telling

The survey found that the most commonly cited barriers to AI adoption in packaging were intellectual property and privacy concerns, plus "limited understanding of which use cases drive value." That second one is the one that should worry buyers.

If packaging suppliers themselves don't fully understand which AI use cases drive value, it means they're experimenting. And when suppliers experiment with AI-driven pricing, the experiments land on your purchase orders. I've documented three instances in the past 18 months where quotes from AI-adopting suppliers had pricing structures that didn't match any historical pattern -- weird volume-break thresholds, new "complexity surcharges," tiered pricing that seemed designed to nudge us toward specific SKU configurations. Each time, when I called to discuss, the sales rep couldn't fully explain the logic. "The system generated the quote" isn't a great answer when you're trying to negotiate a $120,000 annual label contract.

What I Wish I'd Known Earlier

Goel's prediction that 2026 will be about "at-scale deployment" -- companies "rewiring their functions end to end" -- means this isn't a pilot-stage issue anymore. It's hitting production.

The things I've learned the hard way, and that I now share with anyone on our procurement side who'll listen:

Keep your own data house in order. If your supplier has AI analyzing your order patterns, you need to understand those patterns at least as well as they do. I built a simple tracking sheet after the 2022 pricing discrepancy -- every PO, every quote, every price change, every justification given. It's caught four pricing anomalies in three years. That's not nothing on a $280K annual packaging budget.

Ask direct questions about how quotes are generated. "Is this price algorithmically generated or did a human price analyst review it?" is a legitimate question. You'd be surprised how many suppliers will tell you if you ask.

Don't ignore the EPR and sustainability data angle. McKinsey also noted that AI is expected to play a bigger role in managing scattered data for EPR compliance and sustainability tracking. Goel said they're "seeing some early evidence of it, but we see a huge potential for it to go even more." That data -- your material compositions, your packaging weights, your recycling rates -- has commercial value. If a supplier's AI model knows your sustainability compliance position, that's another input into your willingness-to-pay model.

I should be clear: I'm a packaging coordinator, not a data scientist. I can't speak to the technical architecture of these AI models. What I can speak to is the downstream effect on people like me who receive AI-generated quotes and have to decide whether the numbers make sense. And right now, the tools for buyers to evaluate AI-optimized pricing are lagging far behind the tools suppliers are using to generate it.

McKinsey's data shows the packaging industry is moving fast on AI adoption. The question nobody in that survey seems to be asking is: moving fast toward what, exactly? If it's genuine efficiency gains that benefit the whole value chain, that's progress. If it's more sophisticated margin extraction dressed up as "commercial excellence" -- well, I've documented enough pricing surprises to know which direction my skepticism leans.