How Li & Fung uses AI and data platforms for supply chain orchestration

Li & Fung, the century-old Hong Kong trading house that has sourced apparel and consumer goods for global brands since 1906, is remaking itself around artificial intelligence and cloud-based platforms. The firm has spent the past three years building digital infrastructure to orchestrate production, logistics and inventory across thousands of factories in Asia, competing not only with traditional intermediaries but also with software vendors that promise to disintermediate the sourcing agent altogether.

The shift matters because Li & Fung still handles billions of dollars in annual merchandise volume for household names in fashion and retail, and its technology choices shape how thousands of suppliers adopt digital tools. As brands demand shorter lead times, smaller production runs and real-time visibility into factory conditions, the company's platform strategy offers a live case study in whether incumbents can use data and machine learning to defend their role in the supply chain.

This article examines how Li & Fung is deploying AI for demand forecasting, supplier matching and logistics optimisation, what results it claims to have achieved, and where gaps and risks remain for fashion companies evaluating similar investments.

What is Li & Fung's digital platform strategy?

Li & Fung has consolidated its technology initiatives under a cloud-based platform designed to connect brands, factories and logistics providers in a single data environment. The platform ingests order information, production schedules, shipment tracking and supplier performance metrics, then uses analytics and machine learning to recommend sourcing decisions, allocate orders and flag potential delays.

The company describes this approach as supply chain orchestration: rather than simply passing purchase orders from buyer to factory, Li & Fung positions itself as a control tower that continuously optimises the flow of materials, production capacity and finished goods. The platform relies on application programming interfaces to pull data from enterprise resource planning systems at both brand and supplier ends, creating a shared source of truth for order status, inventory levels and shipment milestones.

In investor presentations during 2025, Li & Fung highlighted investments in what it calls digital sourcing capabilities, which include tools for virtual sampling, three-dimensional product design collaboration, and automated request-for-quotation processes that match a brand's specification to supplier capabilities without manual intervention. The goal is to compress the traditional product development cycle, which in apparel can stretch to nine months or more from concept to warehouse receipt, and to reduce the need for physical samples and in-person factory visits.

How does Li & Fung use artificial intelligence in sourcing and production?

Li & Fung applies machine learning models to three core workflows: demand forecasting, supplier capacity matching and quality prediction. Demand forecasting algorithms analyse historical order patterns, sell-through data shared by retail clients, and external signals such as search trends and weather forecasts to generate short-term production recommendations. The aim is to help brands reduce excess inventory and out-of-stock situations by aligning manufacturing volumes more closely with actual consumer demand.

Supplier capacity matching uses a classification algorithm trained on years of production data to score each factory in Li & Fung's network against incoming order requirements. Variables include fabric type, garment complexity, minimum order quantity, ethical compliance audit results and on-time delivery history. When a brand submits a new purchase order, the system ranks suppliers by suitability and recommends allocations that balance cost, speed and risk.

Quality prediction relies on image recognition and anomaly detection. During production, factories upload photographs of in-process garments to the platform; computer vision models flag deviations from the approved sample, such as incorrect stitching or colour mismatch. The system also monitors historical defect rates by supplier and product category, alerting account managers when a particular factory shows rising quality variance.

Which brands and product categories are using the platform?

Li & Fung has not published a comprehensive client roster for its digital platform, and major retail customers typically do not disclose their sourcing-agent relationships in detail. Public filings and investor materials confirm that the company serves large North American and European retailers across apparel, footwear, home textiles and hard goods, but specific brand names and transaction volumes for the digital tools remain confidential.

In general terms, Li & Fung has indicated that digital sourcing adoption is higher among clients in fast-fashion and value retail segments, where speed to market and frequent product changeovers create a stronger incentive to automate order placement and inventory planning. Conversely, luxury and premium brands, which often maintain long-standing relationships with a smaller roster of specialist ateliers, have been slower to shift sourcing workflows onto shared digital platforms, preferring bespoke communication and tighter control over intellectual property.

What measurable results has Li & Fung reported?

Li & Fung has shared qualitative claims about efficiency gains and lead-time reductions but has been cautious about publishing granular performance data that would allow independent verification. In a 2025 investor update, the company stated that clients using its digital sourcing tools experienced shorter product development cycles and improved on-time delivery rates compared to clients relying on traditional manual processes, but it did not quantify the magnitude of those improvements or disclose the sample size.

Anecdotal evidence from supply chain conferences and trade press suggests that digitalisation can cut order-processing time by one to two weeks and reduce the number of physical samples required during product development, yielding cost savings in courier fees and material waste. However, these benefits accrue gradually and depend heavily on the maturity of each client's own systems and the willingness of suppliers to adopt new data-sharing protocols.

Li & Fung has also pointed to improved inventory turns for selected clients, arguing that better demand forecasts allow retailers to replenish bestselling styles more frequently and to cancel or modify orders for slower-moving products earlier in the production cycle. Again, the company has not released audited figures that isolate the contribution of its AI models from other variables such as market conditions, merchandising strategy or promotional activity.

How does the platform integrate logistics and shipment tracking?

Beyond factory sourcing, Li & Fung's platform extends into freight forwarding and logistics orchestration. The company operates its own logistics arm and also partners with third-party carriers to book ocean, air and land transport. The digital platform aggregates shipment data from multiple carriers, providing clients with a unified view of in-transit inventory, estimated arrival times and customs clearance status.

Machine learning algorithms analyse historical shipping performance, port congestion data and carrier reliability scores to recommend routing and carrier selection. For time-sensitive orders, the system can automatically escalate to faster but more expensive transport modes if a factory delay threatens to miss a retail delivery window. Similarly, when multiple orders for the same client are in production at different factories, the platform can consolidate shipments to reduce freight costs and simplify warehouse receiving.

Real-time visibility is particularly valuable during peak seasons and periods of supply chain disruption. During the 2025 calendar year, global shipping faced intermittent delays due to port labour disputes, capacity shortages and geopolitical tensions. Li & Fung's platform allowed account managers to reroute shipments, switch carriers and adjust production schedules in response to evolving conditions, reducing the risk of stock-outs or late-season markdowns.

What are the risks, limitations and open questions?

Despite the strategic logic of AI-driven supply chain orchestration, several challenges and uncertainties remain. First, data quality is uneven across Li & Fung's supplier base. While large, technologically sophisticated factories can integrate their production systems with the platform via APIs, many smaller workshops lack enterprise software or reliable internet connectivity. In those cases, data entry is manual, delayed and error-prone, which undermines the accuracy of machine learning models that depend on timely, granular input.

Second, suppliers may resist sharing detailed production and capacity data with an intermediary that also serves their competitors. Factories worry that transparency into their cost structures, margins and available capacity could weaken their negotiating position or lead Li & Fung to allocate orders to lower-cost rivals. Building trust and demonstrating mutual benefit is a slow process, and incomplete supplier adoption limits the platform's network effects.

Third, the business case for digital sourcing is not universally compelling. Brands that already have deep, long-term relationships with a handful of preferred factories may see little incremental value in algorithm-driven supplier recommendations. Luxury and premium segments, where craftsmanship, exclusivity and intellectual property protection matter more than speed or cost, are particularly sceptical of shared digital platforms and standardised workflows.

Fourth, Li & Fung faces competitive pressure from software vendors that sell supply chain visibility and production management tools directly to brands, bypassing the intermediary altogether. If a retailer can achieve comparable forecasting accuracy, supplier scorecarding and shipment tracking by licensing software and hiring internal supply chain analysts, it may choose to disintermediate the trading house and capture the margin for itself. Li & Fung's value proposition rests on the premise that its proprietary data, supplier relationships and domain expertise cannot easily be replicated, but that assumption is being tested as technology costs fall and talent becomes more widely available.

Finally, the environmental and social dimensions of supply chain orchestration remain under-developed. While Li & Fung's platform can track ethical compliance audits and flag suppliers with poor labour or environmental records, it does not yet integrate carbon footprint calculations, circularity metrics or real-time monitoring of working conditions. As regulators in the European Union and elsewhere introduce mandatory due diligence and sustainability reporting requirements, the platform will need to expand its data model and analytics to help clients meet those obligations.

How do Li & Fung's capabilities compare to in-house and software alternatives?

Fashion companies evaluating supply chain technology investments face a choice among three archetypes: relying on a full-service intermediary like Li & Fung, licensing software from a specialist vendor and building internal analytics capabilities, or pursuing a hybrid model that combines elements of both. Each approach entails different trade-offs in cost, control, speed and risk.

A full-service intermediary offers the fastest path to digital sourcing capabilities because the infrastructure, supplier integrations and machine learning models are already built. However, the brand sacrifices some control over supplier selection, data ownership and the ability to differentiate its supply chain as a competitive advantage. Licensed software and internal teams provide greater control and customisation but require upfront investment in technology, training and change management. Custom-built platforms offer maximum flexibility and intellectual property protection but are expensive, time-consuming and risky, especially for mid-sized companies without deep technology talent.

In practice, many fashion companies pursue a hybrid model: they use an intermediary for commodity categories or regions where they lack direct relationships, while building proprietary capabilities for strategic product lines or key suppliers. This portfolio approach allows them to experiment with AI and data platforms incrementally, learn what works and avoid over-committing to a single vendor or technology stack.

What is the outlook for AI in supply chain orchestration?

The trajectory of AI adoption in fashion supply chains will be shaped by three forces: regulatory mandates for transparency and sustainability, competitive pressure to reduce lead times and inventory risk, and the evolving cost and capability of machine learning technology. Li & Fung's platform strategy illustrates both the promise and the friction of digital transformation in a fragmented, relationship-driven industry.

For intermediaries, the imperative is to demonstrate that their data assets, supplier networks and domain expertise create durable advantages that cannot easily be replicated by software vendors or in-house teams. For brands, the challenge is to balance control and speed, investing enough in proprietary capabilities to differentiate their supply chains while leveraging third-party platforms where commoditisation and scale economies make sense.

Over the next two to three years, expect to see consolidation among supply chain software vendors, deeper integration of environmental and social data into orchestration platforms, and growing use of generative AI to automate product design, sampling and supplier communication. The companies that succeed will be those that treat digital sourcing not as a technology project but as a continuous process of experimentation, learning and adaptation in partnership with suppliers, logistics providers and technology partners.

What should fashion companies do now?

1. Audit your data infrastructure. Before investing in AI-driven sourcing or logistics, assess the completeness, accuracy and timeliness of your internal data on demand, inventory, supplier performance and shipment status. Identify gaps and prioritise integration projects that will create a reliable foundation for machine learning models.
2. Map supplier digital maturity. Survey your production partners to understand their current technology capabilities, willingness to share data via APIs and capacity to adopt new digital workflows. Focus initial platform pilots on suppliers that are ready and willing, rather than trying to mandate adoption across the entire base.
3. Define clear use cases. Articulate specific business problems you want to solve with supply chain orchestration, such as reducing lead times for seasonal replenishment, improving forecast accuracy for new product introductions or minimising freight costs during peak shipping periods. Measure success against these concrete objectives rather than vague efficiency gains.
4. Pilot with a limited scope. Start with a single product category, region or client segment where the benefits of real-time visibility and automated decision-making are likely to be highest. Run a controlled experiment that compares digital workflows to traditional processes, capture quantitative results and iterate based on what you learn.
5. Negotiate data ownership and portability. If you engage an intermediary or license a software platform, clarify in the contract who owns transaction data, supplier performance metrics and machine learning models. Ensure you can export your data in a standard format and migrate to a different system if the relationship ends or your needs change.
6. Integrate sustainability and compliance metrics. As regulatory pressure increases, build carbon footprint, circularity and labour-rights indicators into your supply chain data model from the outset. Choose platforms that can ingest third-party audit results, track Scope 3 emissions and generate reports aligned with emerging disclosure standards.
7. Invest in change management and training. Technology alone will not transform your supply chain; people need to understand how to interpret algorithmic recommendations, override them when business judgment requires and continuously improve the models with feedback. Allocate budget for training, process redesign and ongoing support, not just software licenses and hardware.