What the multi-enterprise supply chain is, and how to power it with agentic AI

https://delivery-p141552-e1488202.adobeaemcloud.com/adobe/assets/urn:aaid:aem:58f3c308-408d-4020-9d5a-a0f56273b153/original/as/AI_multi-enterprise_ft_Stephanie_Krishnan_Hero.jpg

Summary

  • We recently sat down with IDC’s Stephanie Krishnan in our most popular supply chain webinar yet.
  • The linear supply chain model is dead, she argues. Seven years of supply chain disruptions have exposed the three fatal weaknesses of linear supply chains: the tier-n blind spot, the digital tower of Babel, and the expanded attack surface.
  • The multi-enterprise supply chain model is replacing it. Think suppliers, logistics partners, manufacturers, and customers connected in real time — sharing not just data, but workflows, decision logic, and contextual intelligence. IDC predicts 50% of enterprise-scale supply chains will operate this way by 2028.
  • The hardest part isn't the technology — it's change management. Two of the top three barriers to AI adoption are organizational. Success means transparent decision-making, genuine role redesign, and choosing high-stakes use cases where the value is visible from day one.

What is happening to supply chains?

Watch the full discussion with Stephanie Krishnan on-demand →

“They see disruptions weeks before competitors can. And they secure alternate sources while others are still diagnosing the problem." Stephanie Krishnan, Associate Vice President at IDC is describing one of many organizations that recently shifted their supply chain to a multi-enterprise model.

It was through hearing more and more stories like these that Krishnan realised that supply chains had reached a turning point. After years of disruptions, including global events that no one, in their wildest contingency planning, might have imagined – the supply chain paradigm was finally shifting.

Krishnan leads IDC's supply chain execution research practice and has interviewed hundreds of operations leaders: she’s been uniquely positioned to see this shift.

The end of the linear model

For decades, supply chains have been built on a linear model: sequential handoffs, from one manufacturer to another, assuming a baseline of stability while optimizing for costs.

Yet as disruptions and shocks staggered over the last seven years (ranging from yes, COVID, but also tariffs and geopolitical conflicts, and climate events among many others), stability could no longer be counted on. It became an exception. This meant that the linear model no longer worked.

“It’s efficient when it goes to plan,” Krishnan said. “When it doesn’t, there’s nowhere for the shock to go, except straight through. What we’ve seen [in the last two years] was a convergence of volatility that supply chains just simply couldn’t absorb.”

During her research surveying a wide range of supply chain leaders, Krishnan noticed that linear supply chains were failing at three identifiable points. Each of these was, really, a failure of connectivity–a failure to acknowledge, perhaps, a supply chain as what it really is. Not a series of neat, sequential handoffs, but an interconnected, complex, network.

  1. The first failure point she named the tier-n blind spot. “It’s a visibility problem. For example, planning teams at food companies who don’t have visibility beyond their tier-n ingredient suppliers won’t find out about issues until after it’s too late to act. That’s the tier-n blindspot; it doesn’t announce itself, and it happens a lot later,” she explained.
  2. The digital tower of Babel is an interoperability issue. Krishnan gives an example: “During recent port congestions, organizations [...] running on disconnected or siloed systems couldn't adapt. Manual handoffs between disparate systems cause cascading delays that more agile competitors could avoid entirely.”
  3. The expanded attack surface is the third. More connectivity implies more security risks. New technology, including and especially AI, means that supply chains see themselves as ever more vulnerable to cyber attacks. “We're now seeing ransomware, cyber-physical attacks on IoT equipment, and AI-enabled attack vectors. IDC survey data is actually showing that 45% of global organizations rank cyber security as a top two supply chain risk, with only economic uncertainty ranking above it.”

It was clear that things were no longer working. But if the linear supply chain is declared well and truly dead, what should come to replace it? Luckily, the failure points also were signposts for what needed to change, according to Krishnan: “Together [they] define exactly what the next architecture has to replace: single-tier visibility with n-tier intelligence, bilateral data exchange with interoperability and shared workflows, and, finally, perimeter security with distributed, continuous risk monitoring across the entire ecosystem.”

Enter the multi-enterprise supply chain

Imagine a world in which suppliers, logistics partners, manufacturers and customers are all connected in real time. Not just exchanging data—although that is part of it—but also sharing workflows, coordinated decision logic, and contextual intelligence.

  • Why does context matter for AI? AI models are powerful generalists trained on patterns from the internet, but they inherently lack an understanding of how a specific business operates. This lack of understanding creates operational blind spots. To turn AI from a generalist into a specialist that can be trusted with business-critical tasks, it must be provided with operational clarity—a grounded, real-time understanding of a company's past, present, and future. Find out more.

What you’re picturing is the future of the supply chain, or what Stephanie Krishnan, and IDC, are calling the multi-enterprise.

“Organizations are starting to recognise that supply chains are a network. They require real time or near-time visibility beyond the immediate tier. Now we have the technology that makes it possible to address that and pull in the orchestration required to make it happen,” Krishnan said.

It is not a distant future, either. As we mentioned at the very start, Krishnan has spoken to several organizations who’ve shifted to that model.

“[Unilever] has invested in AI-driven supply chain intelligence. It handled the tariff shocks and geopolitical volatility of 2025. They built that speed using AI tools that continuously scan supplier networks, validate alternatives in real time, and then surface options before a disruption becomes a bottleneck. They built a digital twin of their global supply chain using satellite imagery and AI to simulate disruptions, assess risk, and inform proactive logistics decisions ahead of time,” Krishnan explained.

And in IDC’s FutureScape: Worldwide Supply Chain and Industry Ecosystems 2026 Predictions it is predicted that by 2028, 50% of all enterprise-scale supply chains will use business networks to enable n-tier visibility with the express goal of improving disruption response speeds by 25%.

The multi-enterprise, however, does not stop at connectivity, with everyone seeing the same data, but making decisions in isolation from everyone else. It also encompasses orchestration, with the whole network working together on “what happens next, who acts on it, how fast, and across which nodes of the network,” said Krishnan.

It is, put another way, a complete mindset shift: a change in the way that supply chains have been run, up until now, enabled by AI.

The unexpected adoption bottleneck

Supply chains will be shifting to a multi-enterprise model, powered by AI. The question is: how fast?. At the moment, most of them have one important obstacle standing in their way.

One of the problems is the same as with all technological innovations: people adjusting to change.

Krishnan has done a lot of research on AI adoption barriers. “The top three obstacles that CEOs highlight are: a lack of specialized AI expertise, lack of data readiness, and lack of AI awareness and usage skills in the workforce. Now, two of these three are people problems. That’s an organizational readiness failure.” (Source: IDC CEO Survey 2026).

What this might look like in real life is having an AI tool that generates a more accurate demand forecast, that is then overridden because the planner doesn’t trust it. Or building an agentic workflow that rebalances production autonomously, slowed by an operations team that insists on approving every output manually.

What successful adopters do differently

And staff training alone won’t solve the problem, Krishnan argues. Shifting to a multi-enterprise model—and so, achieving successful AI adoption–means redesigning both processes and jobs.

Put another way, agentic AI shouldn’t be conceived of as replacing a supply chain planner, but rather, taking repetitive work off of their plate so that they can focus on decisions that need human judgement.

“Leaders that can make that case credibly and back it up with genuine role redesign are the ones who are going to scale,” Krishnan says.

Watch supply chain influencer Lora Cecere discuss how rethinking roles is key to successful AI adoption, in our webinar ‘Say no to ‘AI Stupid’.’

So while the technology is new, one of the elements key to its flourishing is familiar. It rests on successful change management.

Krishnan has two pieces of advice on how to do it right.

The first is to be transparent about how AI decisions are made, what data it is acting on and where humans remain in the loop. As part of this, she recommends involving your HR department from the start on any AI adoption project.

The second is that value builds trust. Proving an outcome visibly, means that everyone sees the benefits from an AI implementation. This might mean choosing a use case “where the cost of being slow, inaccurate, or incomplete is the highest. [And] design your roadmap so each use case opens the door to the next.”

If you do get it right? It means that the next time you’re facing a tariff shift, a climate event, or a demand spike, you’ll be detecting it earlier. Early enough to act, because you’ve done simulations, modeled requirements, and set up processes where agents can step into action.

These actions will then cascade across your partner network in hours, rather than days or weeks. It also means that you’re getting ahead of your competitors. If you take action before they can, you’ll be able to secure alternate sources while everyone else is still diagnosing the problem.

Keep in mind that it all compounds over time. Supply chain organizations that are participating in multi-enterprise networks report better collaboration, faster access to partners, and lower costs, says IDC.

Krishnan summarizes it like this: “That's the multi-enterprise model in practice, right? Not reacting to what your tier-one tells you, but acting on signals from across the network before the impact reaches you.“

Watch the full session

Krishnan covered a lot more in what was our most popular supply chain webinar to date, including: the three stages of AI maturity, data readiness, and supply chain cybersecurity. Watch the full recording on-demand now.

Other resources

Discover the difference between agents, assistants and copilots with our latest e-book Destination AI: Unpacking assistants, co-pilots, and agents.

Watch McKinsey experts discuss the future of supply chains in three short interviews.

Get the key highlights from our global survey of 400 supply chain leaders, including top AI use cases and key obstacles to AI ROI.

FAQs

What is a multi-enterprise supply chain?
A multi-enterprise supply chain connects suppliers, logistics partners, manufacturers, and customers as a single coordinated network rather than a sequence of handoffs. It combines real-time visibility across multiple tiers with shared workflows and orchestrated decision-making, so disruptions detected anywhere in the network can trigger coordinated action everywhere in the network, within hours, not days.
Why are linear supply chains failing?
Linear supply chains were built for stability and optimized for cost, with each tier handing off to the next. That model breaks down at three points: visibility stopping at your immediate suppliers (the tier-n blind spot), disconnected systems can't talk to each other during disruptions (the digital tower of Babel), and increased connectivity creates new cyber vulnerabilities (the expanded attack surface). IDC research found 45% of global organizations now rank cyber security as a top-two supply chain risk.
What's the biggest barrier to AI adoption in supply chains?
According to IDC research, the top three barriers CEOs cite are a lack of specialized AI expertise, a lack of data readiness, and a lack of AI awareness and skills in the workforce — two of which are people problems, not technology problems. Training alone won't solve this; organizations need to redesign roles so AI handles repetitive work while humans focus on judgment-based decisions, and they need to involve HR in AI projects from the start.

Sources and references