At Renault, a single daily planning cycle involves 45,000 decision variables and 100,000 constraints, and the system is expected to produce a high-quality answer in under five minutes. Behind it sits an optimization solver: a mathematical engine that evaluates trade-offs at a scale no planning team could match manually.

In case you want to jump ahead

1. Inside the Solver: Different Paths to the Same Goal 

2. The Rise of Hybrid Optimization

3. Behind the Solver: Renault’s Packaging Management System

4. Evaluating Optimization Through Different Lense

5. Supply Chain Optimization Maturity Self-Assessment

6. In Conclusion

The Rise of Hybrid Optimization

Instead of relying exclusively on linear programming, heuristics, or constraint programming, modern platforms orchestrate several approaches simultaneously, allowing each method to contribute where it performs best. 

Depending on the problem, they may use linear optimization to evaluate strategic trade-offs, combinatorial optimization to manage complex allocation decisions, and heuristics to accelerate the search for high-quality solutions. Rather than relying on a single algorithm, solvers can switch between or combine approaches to balance solution quality, computation time, and model complexity. 

Behind the Solver: Renault’s Packaging Management System

At Renault Group, reusable packaging continuously circulates between approximately 1,400 suppliers, 40 plants and cross-docks, and multiple cleaning, repair, and recovery locations. Ensuring that the right packaging is available at the right place and time is essential for maintaining production continuity while controlling transportation and asset-related costs.

For each day, supplier, plant, and packaging type, Renault must determine how many packaging units should be moved throughout the network while respecting inventory balances, shipment requirements, facility capacities, and operational constraints. At the same time, the system must balance several potentially conflicting objectives, including minimizing packaging shortages, reducing the number of shipments, and limiting overall travel distances.

To support these decisions, Renault developed a Packaging Management System (PMS) powered by Hexaly.

A typical optimization model includes approximately:

The challenge is finding a high-quality solution within a timeframe that allows planners to act on the results.

According to Hexaly, the system achieves an average optimality gap below 1.5% within a five-minute solving window, enabling Renault to regularly optimize packaging flows at a scale that would be impractical through manual planning or spreadsheet-based analysis.

Optimization solvers may sit at the center of the same supply chain platform, but analysts and executives often evaluate them through entirely different lenses.

Click on the perspective you want to analyze – executives or analysts.

Supply Chain Optimization Self-Assessment

Not every organization uses supply chain optimization in the same way. The following framework can help organizations assess their supply chain optimization maturity.

In Conclusion

Optimization solvers rarely receive the same attention as artificial intelligence, digital twins, or control towers. Yet behind many of the most important supply chain decisions lies an optimization engine evaluating trade-offs that would be impossible to assess manually.

Whether supporting strategic network design, production planning, packaging management, or inventory optimization, solvers provide a structured way to navigate complexity and identify actions that align with business objectives.

At the same time, optimization itself is evolving. Modern platforms increasingly combine exact mathematical methods, heuristics, simulation, and artificial intelligence to solve problems that would have been impractical only a few years ago. The result is not just faster computation, but more realistic decision support.

The organizations that create the greatest value in the future will be those that use each tool where it performs best: human expertise for judgment and oversight, AI for pattern recognition and prediction, and optimization for navigating the complexity in between.

Demand volatility has become a structural feature of modern supply chains, not a temporary disruption to be managed. Weather swings, viral product trends, geopolitical shocks, and sudden shifts in consumer behaviour can reshape demand patterns in days rather than quarters. Traditional forecasting systems, built primarily on historical sales data and updated on weekly or monthly cycles, were not designed for this pace. By the time a conventional model detects a shift, the stockout has already happened or the overstock is already accumulating.

Artificial intelligence is changing that dynamic. By combining machine learning with generative AI, companies can now integrate real-time signals from weather feeds, social media, IoT sensors, and point-of-sale systems and simulate multiple demand scenarios before disruptions materialize. It is a fundamental change in what forecasting means: from predicting a number to understanding demand dynamics as they unfold.

In case you want to jump ahead

1. The Methodologies Behind the Shift

2. Ice Cream as the Case for Real-Time Forecasting

3. Additional Industry Cases

4. Organisational Readiness Check

5. Key Implementation Steps: Choose Your Perspective

6. Conclusion: Sense Faster, Respond Earlier

The Methologies Behind the Shift

To understand what AI is changing in demand planning, it helps to understand what the new methods actually do and how they differ from the statistical models that preceded them. Four approaches currently define the state of the art.

Ice Cream and the Case for Real-Time Forecasting

Few products expose the limits of traditional demand forecasting as clearly as ice cream. And few companies have gone further in rethinking how demand planning works at scale.

Demand for ice cream is simultaneously predictable in aggregate and deeply volatile at the local level. Seasonal patterns are well understood. But a sudden heatwave can trigger a regional surge within 48 hours, while a rainy week suppresses sales just as rapidly. For a product sold almost solely through impulse purchases at the point of a freezer cabinet, with a cold chain that cannot be improvised overnight forecasting errors translate directly into lost revenue, excess inventory, and inefficient logistics.

This is the structural challenge faced by Unilever, whose ice cream supply chain spans 60 countries, 35 factories, and roughly 3 million freezer cabinets worldwide. At that scale, even a 5% improvement in forecast accuracy is worth hundreds of millions of dollars annually. And the cost of a stockout on a hot afternoon when a customer reaches into a cabinet and finds it empty is not just a lost sale. It is a brand interaction that fails at the most critical moment of the purchase journey.

From Historical Extrapolation to Real-Time Demand Sensing

The Freezer Cabinet as a Demand Sensor

The most operationally significant change has happened at the retail edge. More than 100,000 of Unilever’s freezer cabinets are now equipped with AI-powered image recognition systems that monitor stock levels in real time. When a camera detects that a cabinet is running low, the system generates a replenishment recommendation automatically, not based on a projected depletion curve calculated last week, but based on what is actually in the cabinet right now, cross-referenced with the current weather forecast and local sales velocity.

The practical consequence is that roughly 5% of seasonal orders are now generated directly from these AI-driven recommendations rather than from traditional planning cycles. That percentage will grow as coverage expands. For operations teams, this means fewer manual ordering decisions, more efficient delivery routes, and a fundamentally different relationship between demand signal and supply response.

Why Ice Cream Case Matters Beyond Ice Cream

The ice cream case is valuable precisely because it is not a special case. The same pattern applies to beverages, fresh food, seasonal apparel, and any product where the gap between demand signal and supply response carries a real cost.

What Unilever demonstrated is that closing that gap requires three things working together: a real-time data layer (weather, POS, cabinet sensors), a forecasting model capable of processing those signals faster than the planning cycle, and an execution system that can act on the output without waiting for human review. None of those three elements alone is sufficient. The competitive advantage comes from their integration.

The freezer cabinet is not just a storage unit. For Unilever, it has become the primary demand sensor, generating signals that the planning system acts on within hours, not weeks. That is a different model of forecasting.

Additional Industry Cases

The ice cream model is not an isolated experiment. The same underlying shift is playing out across industries, each with its own version of the volatility problem.

Click on the use case you want to learn more about:

Before committing to a technology investment, supply chain leaders should assess their organisation’s readiness across four dimensions. The table below provides a structured framework. Most large organizations entering AI demand planning in 2025–2026 sit at the Developing stage on data and at the Early Stage on governance, the latter being the more consequential gap to close.

The developments described above are not sequential innovations. They are converging simultaneously.

Click on the perspective you want to analyze – executives or analysts.

Conclusion: Sense Faster, Respond Earlier

Traditional forecasting systems were designed to predict numbers. AI-driven systems are designed to understand demand dynamics as they unfold. The combination of machine learning, transformer-based probabilistic models, generative AI scenario simulation, and agentic autonomous execution is transforming demand forecasting from a periodic planning activity into a continuously learning, real-time system.

The Unilever ice cream case makes this concrete. A freezer cabinet that detects its own depletion, cross-references the local weather forecast, and generates a replenishment order without human intervention is not a futuristic scenario. It is already operating at scale across 100,000 retail locations.

For supply chain executives, the strategic implication is clear. Better forecasts improve inventory efficiency, reduce waste, and ensure product availability when demand spikes. But the deeper competitive advantage is structural: organisations that can sense demand faster and respond earlier than their competitors gain a durably superior position.

The organisations that gain structural advantage in the next five years will not be those that deployed AI first. They will be those that built the data foundations, governance frameworks, and organisational capabilities to use it reliably at scale.

A major challenge in supply chains is coordinating and consolidating transportation flows, as disruptions can create cascading effects, particularly for perishable goods requiring strict temperature and handling conditions. To address this, scientific studies have developed bi-objective Mixed-Integer Nonlinear Programming (MINLP) models that minimize both costs and time while accounting for disruption uncertainty through credibility-based possibilistic programming approaches.

Disruptions caused by facility failures, natural disasters, or other unforeseen events can severely impact perishable products such as pharmaceuticals. As a result, research increasingly focuses on designing reliable supply chain and logistics networks using consolidation hubs while considering disruption risks and product perishability.

In case you want to jump ahead

1. Mathematical Modeling and the Importance of Network Resilience

2. Addressing Disruptions and Product Perishability with MINLP Models

3. How Does the Optimization Process Work?

4. The Role of Quantitative Modeling of Supply Chain Networks

5. Implementation Fields of Quantitative Modeling

6. Trends, Characteristics & Future of Sustainable SCM

7. In Conclusion

Mathematical Modeling and the Importance of Network Resilience

Resilience has emerged as a critical factor in the performance of supply chain networks, especially in the face of global disruptions. Events such as pandemics, cyberattacks, and capacity shortages can have severe consequences for logistics operations. These statistics underscore why resilience is so important:

Mathematical models and operations research/management science (OR/MS) methods offer effective strategies to enhance the resilience of supply chains. In the following sections, we explore how these methods are applied in practice.

Addressing Disruptions and Product Perishability with MINLP Models

One innovative aspect of Mixed-Integer Non-Linear Programming (MINLP) models is their ability to manage facility disruptions in key hubs and distribution centers. When disruptions occur, operations can switch to pre-identified backup facilities, enabling quick product rerouting while minimizing delays and maintaining service levels. Sensitivity analysis further helps evaluate how disruption probabilities and perishability constraints affect network cost and reliability.

An essential factor in these models is product perishability, particularly for goods which are highly sensitive to time and environmental conditions. When disruptions occur, the risk of spoilage or degradation increases, especially for products with limited shelf lives.

The model integrates real-time data on product life cycles, temperature controls, and environmental factors into the logistics network. If a hub fails, products can be quickly rerouted to alternative facilities or directly to customers through adjusted inventory policies and optimized transportation routes, helping maintain product quality. The model also evaluates mitigation strategies such as redundant cold chain storage and emergency response protocols to protect perishable products from facility disruptions.

How Does the Optimization Process Work?

To minimize cost and transportation time, the MINLP optimization model should include the following elements:

The model’s constraints should also account for the capacity limits of hubs and distribution centers, the need to meet customer demand, and the correlation between hub failures and product perishability. Each supplier is single-allocated to one main hub and one backup hub, ensuring a streamlined and efficient network configuration.

Implementation Fields of Quantitative Modeling in Supply Chain Networks

The placement of warehouses is another crucial component of resilient supply chain network design. Quantitative location models evaluate geographic risks, such as the likelihood of natural disasters, labor strikes, or political instability, and help companies select optimal warehouse locations that minimize exposure to these risks. Additionally, businesses must manage inventory levels within warehouses, employing inventory buffering strategies to handle demand fluctuations without overstocking or causing bottlenecks.

Businesses often experience sudden demand spikes, such as during holiday seasons or unforeseen events like global pandemics. To build resilience, they use quantitative models to forecast these demand shifts and adjust warehouse and transportation capacities accordingly. Flexible labor force modeling ensures that sufficient personnel are available to manage surges, while capacity redundancy models help plan for increased storage needs or rerouting of goods during disruptions.

Companies often manage goods across multiple transportation modes (trucks, ships, planes, and rail), each with its own risk exposure. Quantitative models optimize these multi-modal systems by simulating how disruptions in one mode (e.g., port shutdowns) affect overall delivery times and costs. By analyzing this data, businesses can preemptively reallocate resources and reroute goods through alternative modes, ensuring continuity of operations.

Trends, Characteristics & Future of Sustainable SCM

Sustainable Supply Chain Management (SSCM) research mainly focuses on industries with high environmental and regulatory pressure, such as transportation, textiles, consumer goods, automotive, and electronics. Most studies rely on normative models that optimize costs and reduce carbon emissions, while descriptive models explaining real supply chain behaviors remain less developed. SSCM increasingly incorporates economic, environmental, and social metrics, with future research expected to strengthen social sustainability and develop more industry-specific approaches.

A key SSCM strategy is Closed-Loop Supply Chain Management, which integrates reverse logistics, recycling, remanufacturing, and product recovery to minimize waste and support circular economy principles. Unlike traditional linear supply chains, closed-loop systems extend product lifecycles and recover value from end-of-life products, helping companies reduce raw material use, lower disposal costs, and improve sustainability performance. Successful implementation depends on effective reverse logistics, supply chain collaboration, and technologies for tracking and processing returns.

In Conclusion

In an era defined by unpredictability and rapid change, building resilient network designs has become increasingly important. Research highlights the value of proactive approaches that anticipate disruptions and equip organizations with strategies to adapt effectively. By integrating advanced modeling techniques and fostering collaboration, companies can create flexible supply chain networks capable of withstanding high-impact, low-frequency “black swan” disruptions.

The ability to pivot during crises through alternative routing, supplier diversification, and technology adoption can significantly reduce risks and improve operational efficiency. Companies that invest in resilient network designs strengthen both their crisis response capabilities and long-term viability.

Ultimately, resilience in logistics and supply chain management requires continuous innovation, strategic partnerships, and learning from past disruptions. By embracing these principles, companies can better adapt to challenges and create opportunities for long-term growth and operational excellence.

Supply chain network designs are drawn up in a world that no longer exists. What is emerging in its place is fundamentally different, not just technologically, but structurally and philosophically.

Supply chain network design used to be a periodic strategic exercise. Now it is becoming a continuous, multi-objective, AI-assisted capability. And the organizations that understand this are restructuring their networks in ways that will be very difficult to replicate.

This article dissects what is actually changing, drawing on the most recent academic models, real-world 2026 implementations, and practitioner-level intelligence that doesn’t make it into press releases.

In case you want to jump ahead

1. The Problem With Classical Network Design

2. What the Science Is Actually Saying

3. The Closed-Loop Imperative

4. How Leading Companies Redesigned Their Networks in 2025–2026

5. Geopolitical Pressure and the “In-Region For Region” Shift

6. Agentic AI: The Emerging Backbone of Continuous Network Design

7. How to Act on This: Choose Your Perspective

8. Conclusion

1. The Problem With Classical Network Design

Traditional supply chain network design operates on a periodic, project-based logic: every three to five years, a consulting team or internal strategy group runs a facility location model, evaluates a handful of scenarios, and produces a report with a recommended network structure. The network is then locked in (more or less) until the next wave of disruption forces a revision.

This model has three structural weaknesses that are now impossible to ignore.

2. What the Science Is Actually Saying, and Why It Matters in 2026

Academic research in supply chain network design has been running ahead of industry practice for years. The gap between what is possible mathematically and what most companies actually implement is significant — and that gap is now beginning to close.

Multi-Objective Optimization Is No Longer a Niche Method

The most important methodological shift in recent research is the normalization of multi-objective optimization as the default framework for network design. The augmented epsilon-constraint method (AUGMECON2) allows decision-makers to generate the full Pareto frontier – the set of all non-dominated solutions across competing objectives – rather than collapsing the problem into a single weighted-sum formulation.

2026 RESEARCH HIGHLIGHT

A study published in Sustainability (Yildirim Arslan and Soner Kara) applied the hybrid NSGA-II algorithm to a multi-echelon sustainable supply chain network design problem — simultaneously incorporating solar energy investment, water footprint, carbon emissions, and total cost under stochastic demand. Near-optimal Pareto solutions were generated with deviations under 5% from exact solutions for large-scale instances.

The Reliability Dimension: An Undervalued Objective

Reliability-conscious network designs select facility locations with higher redundancy and shorter average distance to backup resources, even at higher fixed cost, because the model explicitly penalizes configurations where demand cannot be met during disruption events.

There is a ‘sweet spot’ range of reliability investment that delivers high value for modest additional cost — and that insight is directly applicable to any industry where service continuity carries significant revenue or reputational consequences.

3. The Closed-Loop Imperative: From Compliance to Competitive Advantage

The closed-loop supply chain has spent years being treated as a sustainability overlay on top of “real” supply chain operations. That framing is now obsolete. Three forces are converging to make it a core competitive capability.

Extended Producer Responsibility Is Reshaping Network Economics

In the United States alone, EPR legislation is now active in 12 states. Similar frameworks are accelerating across the EU under the Corporate Sustainability Reporting Directive. Companies that build closed-loop network capacity before it is mandated will have structural cost advantages over latecomers who build it reactively at higher unit economics.

Remanufacturing Is Proving Economically Dominant

ENVIRONMENTAL IMPACT

Caterpillar’s remanufacturing produces 65–87% fewer greenhouse gas process emissions compared to manufacturing new parts. This translates directly into competitive positioning as customers increasingly choose suppliers based on lifecycle emissions.

4. How Leading Companies Redesigned Their Networks in 2025–2026

The gap between academic models and industrial practice is closing, but unevenly. A handful of companies are implementing genuinely advanced approaches.

Kroger and the Loop Platform: Closed-Loop Packaging at Scale

Kroger’s 2025 pilot with Loop in Ohio stores produced results that fundamentally change the economic case for closed-loop packaging networks.

Microsoft: 25+ AI Agents Running Supply Chain Operations

Microsoft’s supply chain transformation, described publicly in March 2026, represents one of the most detailed accounts of how agentic AI is being embedded into actual supply chain operations. The company has deployed more than 25 AI agents and applications, with a target of over 100 agents by end of 2026.

THREE KEY AGENTS

Demand Planning Agent — AI-based demand simulations for non-IT rack components. CargoPilot Agent — continuously analyzes transport modes, routes, cost, carbon, and cycle times. Multi-Agent DC Spare-Part Space Solver — forecasts spare-part storage needs using computer vision and multi-agent reasoning.

Toyota Material Handling Europe: Digital Twin-Enabled Warehouse Network Testing

Working with SoftServe and Microsoft, Toyota Material Handling Europe built a digital twin for simulating autonomous forklifts in virtual warehouse environments. Training times for autonomous systems were reduced by more than 30 percent, and the ability to test alternative warehouse network configurations virtually has fundamentally changed how the company evaluates network design options.

5. Geopolitical Pressure and the “In-Region For Region” Shift

The tariff environment created by US trade policy since 2025 has not just affected supply chain costs — it has invalidated entire network architectures that were optimized for a different trade regime.

IKEA provides one of the clearest examples. Facing US tariff pressure on furniture imports with significant Chinese manufacturing content, the company began accelerating US domestic production capacity, including the Mocksville, North Carolina facility. When the tariff premium on imported goods exceeds the production cost differential for domestic manufacturing, regional production becomes not just resilience policy but pure economics.

6. Agentic AI: The Emerging Backbone of Continuous Network Design

The most consequential technological development for supply chain network design is not any specific optimization algorithm. It is the emergence of agentic AI — autonomous AI systems that can reason, plan, and act across the supply chain without waiting for human triggers.

What Agentic AI Actually Does in Network Design Contexts

GOVERNANCE WARNING

Deloitte’s 2026 agentic supply chain report: 40% of current agentic AI projects are expected to be scrapped by 2027, not because the technology doesn’t work, but because of integration drag, governance failures, and unclear business value attribution. The failure mode is predictable: organizations deploy agents before establishing clear boundaries around decision authority.

The developments described above are not sequential innovations. They are converging simultaneously.

Click on the perspective you want to analyze – executives or analysts.

8. Conclusion

Supply chain network design is being restructured by three forces that are not temporary: the mathematical maturation of multi-objective optimization methods; the economic and regulatory imperative to integrate reverse flows into the core network architecture; and the emergence of agentic AI as an operational layer that allows networks to be continuously monitored, stress-tested, and partially self-adjusting.

The companies that are ahead of this shift share a common characteristic: they treat network design as a capability, not a project. The cost of waiting — of running a cost-only network model every few years while the environment changes around it — is increasingly visible in the form of tariff exposure, resilience failures, and missed circular economy revenue.

The question is not whether to redesign your network. It is whether you will do it on your terms or under pressure. Leading companies are already partnering with advanced analytics teams to redesign their networks continuously—combining optimization, scenario testing, and AI-supported decisions to build more resilient and responsive supply chains.

Supply chains today operate under permanent volatility: demand shocks, supplier disruptions, energy price swings, and growing sustainability pressure. At the same time, manufacturing alone accounts for a major share of global energy consumption and CO₂ emissions, forcing companies to balance cost, service, resilience, and carbon, simultaneously, not sequentially.

Digital twins promise to solve this tension. They create a live mirror of operations, allowing companies to simulate disruptions and test decisions before implementing them physically. But the real shift is happening now: digital twins are moving beyond dashboards and scenario analysis toward real-time, AI-driven decision-making.

In case you want to jump ahead

1. Scientific View: The Next Phase of Digital Twins

2. PepsiCo: Building an AI-Powered Industrial Blueprint

3. Digital Twin Readiness Check: Where Do You Really Stand?

4. How to Act on This: Choose your Perspective

5. Conclusion

1. Scientific View: The Next Phase of Digital Twins

Recent research shows what this looks like in practice. A high-fidelity digital twin combined with multi-objective reinforcement learning was tested across 10,000 disruption scenarios, including machine failures, energy crises, and demand surges. The results were striking:

The demand for speed and efficiency has never been greater. Consumers now expect orders at their doorsteps within hours of placing them. This shift has ignited an intense race among businesses to enhance their last-mile delivery strategies, the final leg of the journey that directly impacts customer satisfaction.

The last mile is not just a logistical challenge, it is a critical touchpoint shaping the entire customer experience. As consumers increasingly prioritize convenience and immediacy, businesses must rethink their delivery methods and embrace technology-driven solutions to stay ahead.

1. How Changing Expectations Impact Delivery

2. Express Delivery: Current Trends Shaping the Market

3. What You Should Know About Last-Mile Delivery

4. Four Last-Mile Supply Chain Network Configuration

5. How Leading Companies are Setting the Pace

6. In Conclusion

The last-mile segment is where costs, customer satisfaction, and operational efficiency converge. Several key factors determine its effectiveness.

“Increasing time flexibility in delivery windows leads to reduced travel distances and lower costs per parcel, reception boxes cut per-parcel costs by 42% compared to time-windowed deliveries.” — Kämäräinen (2001)

Urban areas enable efficient routing and lower costs. Route optimization software, AI-powered demand forecasting, and electric vehicles collectively reduce fuel use, delivery times, and carbon footprint while improving overall customer satisfaction.

Based on Lim & Winkenbach (2018), four LMSN configurations define how companies approach responsiveness and product variety.

Last-mile delivery has become one of the most critical competitive battlegrounds in commerce. The data is clear: slow delivery leads to cart abandonment and lost loyalty, while speed drives satisfaction and repeat business. Route optimization, real-time tracking, AI forecasting, and electric vehicles are reshaping what’s possible.

Companies that master the balance between speed, efficiency, sustainability, and customer satisfaction, through flexible logistics networks and technology adoption, will define the future of the market.