Manufacturers aren’t just testing AI anymore, and industry executives now rank it as their top investment priority. That shift makes sense given what’s happening on the ground. Over the past five years, manufacturers have been hit hard with geopolitical instability, unpredictable trade flows, and rising ESG expectations. Disruption is no longer an exception manufacturers can simply plan around; it has become an operating condition they need to design for. CGI’s 2026 Voice of Our Clients research reinforces the scale of that pressure: 68% of manufacturing executives say shifts in the world economic order are having a high impact on their organizations.

Consider a manufacturer facing simultaneous disruption across its network: A critical supplier is experiencing delays, energy costs are rising at one production site, and demand is shifting toward another region. Each signal can be managed independently, but doing so may optimize one part of the business at the expense of another. A sourcing decision could protect supply while increasing logistics costs; shifting production could preserve customer commitments while creating capacity constraints elsewhere. Resilience increasingly depends on connecting these signals and understanding the trade-offs across the entire operation before deciding where and how to act. Legacy operational and technology environments were not designed to handle this level of interconnected and sustained disruption. Volatility has become the norm.

The manufacturers best positioned for what comes next are redesigning how their operations work, based on three goals: absorb shocks without stopping operations, hold their ESG and compliance commitments while conditions shift underneath them, and move quickly when demand changes. All three goals depend on a connected digital foundation—one that brings together trusted data, integrated operations, secure platforms, and clear decision-making across the enterprise.

Connecting AI and human intelligence to deliver business-aligned outcomes

Manufacturers already generate substantial data, and the most significant gains don’t necessarily come from accumulating more of it. The issue is fragmentation. Research highlights this gap: While 54% of manufacturers report having a holistic data strategy, only 48% say they can readily locate their data assets. The challenge is therefore not simply having more data, but making the right data accessible, trusted, and usable across the enterprise. Demand patterns, supplier lead times, asset availability, labor constraints, and external signals like risk indicators typically exist in separate systems without a unified view. With customers now expecting hyper-personalized products, experiences, and shorter lead times, making sense of that scattered data matters more than ever.

Connecting data gives cross-functional teams a shared view of the operation that no single department can produce on its own. Seen through only a procurement lens, a supplier’s delay looks like a sourcing problem to escalate. Read against current demand signals and inventory positions, that same delay might be something the plant can absorb without a customer ever noticing, or it might be the first sign of a stockout six weeks out. Procurement on its own cannot tell the difference, and that gap is what a fragmented view actually costs.

Getting to that second stage requires unified data foundations and clear processes for surfacing and acting on signals. Importantly, manufacturers do not need to clean every piece of data before they can start. They need to determine which data matters for the decisions they are trying to improve, establish clear ownership and governance, and create a trusted foundation that AI can act on.

When cross-functional teams share a common operational view, they can catch early warnings before they cascade into disruptions across the value chain.

A connected picture also makes AI-enabled scenario planning useful in practice. It helps manufacturers know how different decisions play out across service levels, inventory, cost, and speed simultaneously. The question shifts from “What is the right plan?” to “Which trade-off are we willing to make, and what does the data tell us about its consequences?”

Where AI is delivering fast returns

Once manufacturers have a connected data foundation in place, the next decision is to choose which problems to tackle first. Today, some of the clearest opportunities for AI are in areas where decisions are frequent, data-intensive, and constrained by well-understood business rules. Procurement is one example. AI can automate the intake work, flag compliance gaps, and keep an eye on supplier health in real time, cutting manual effort and improving consistency.

Two capabilities stand out in the current trade environment. Continuous supplier risk and ESG monitoring allow manufacturers to move beyond periodic audits toward real-time sensing of financial health indicators, regulatory status changes, and ESG flags across their supplier base—pinpointing potential risk factors before they escalate.

AI-assisted supplier discovery handles the other half of the equation. In constrained or fast-moving categories, manufacturers can identify and qualify alternative suppliers in days rather than months. The result is a procurement function that spends less time gathering information and more time acting on it. That shift, from data gathering to judgment and decision-making, is where the real operational value shows up.

This shift also points to a design principle. AI is most effective when it structures a decision and hands it to a person, with clear boundaries around what it recommends, what it executes, and what it escalates. The objective is not to remove human judgment, but to focus it on decisions where it creates the most value. In procurement, those lines are easier to draw than most teams expect. A routine reorder against an approved supplier at an agreed price can run on its own. A shift in a supplier’s risk score should produce a recommendation and a named person to weigh it. Anything that changes the commercial relationship, such as a new supplier, a revised contract term, or a price outside the agreed band, goes to whoever is accountable for that call.

Well-designed governance can accelerate adoption by making responsibilities and decision boundaries clear. In this sense, governance is not a brake on AI adoption; it is one of the conditions for scaling it responsibly.

Scaling beyond the pilot stage

Most manufacturers have run AI pilots, but extending use cases into day-to-day operations and scaling them across functions, plants, or regions remains a significant challenge. Proofs of concept can demonstrate that a model or use case works. The harder question is whether it can operate reliably within existing processes, systems, and decision structures and whether the organization can repeat that success elsewhere. Moving from experimentation to enterprise-scale AI therefore requires not only technical scalability, but also clear measures of business value, governance, and operational accountability.

The manufacturers that have made that transition consistently share four characteristics:

• Strong data quality, governance, and orchestration
• AI integrated into core systems like ERP, planning, and sourcing platforms
• Clear definitions of which decisions AI recommends, executes, or escalates
• Feedback loops that improve model performance over time.

Taken together, these characteristics point to a broader lesson: Manufacturers cannot scale tomorrow’s AI ambitions on foundations built for yesterday’s operating models.

Governance matters here too. Teams need to know where AI’s responsibility ends and human decision-making begins; in other words, the difference between what the system can suggest and what it’s allowed to do on its own. Without that kind of clarity, people tend to go to one extreme or the other by either ignoring recommendations they don’t understand or simply accepting whatever AI tells them.

Michelin’s experience illustrates one example of what moving beyond AI tool adoption can require inside a manufacturing organization. In its software development lifecycle, Michelin found that deploying AI tools was only the beginning. Expanding adoption across the development and delivery process required changes to processes and ways of working, structured enablement across roles, and new approaches to measuring success. Its experience reinforces a broader principle: Deploying technology can happen relatively quickly; embedding it into how an organization works requires deliberate transformation.

Building what comes next

Trade volatility, geopolitical uncertainty, and ESG pressures are now structural features of the manufacturing environment, and the pace of change won’t slow. The manufacturers best positioned for what comes next will be those that can connect signals across their operations and ecosystems, understand their implications, and turn that intelligence into action. This groundwork is more important than having access to or simply investing in the most advanced AI models.

Treating AI and data as tools for renewal instead of patches on legacy systems is what separates structural resilience from quick fixes. Resilience, ultimately, is not simply the ability to absorb the next shock. It is the ability to sense change early, understand its implications, make informed trade-offs, and adapt quickly.

The factory of the future works as one connected system: globally coordinated, locally reconfigurable, and adaptable by design.

Magali Amiel, Global Manufacturing Lead, CGI

Magali Amiel, PhD, serves as CGI’s Global Lead for Manufacturing, where she guides the strategic direction of the company’s manufacturing portfolio and supports clients navigating digital transformation across complex industrial and supply chain environments. She brings more than two decades of international experience spanning manufacturing, logistics, and goods movement, with deep expertise in transportation systems, data-driven operations, and ecosystem collaboration.