Collaboration Lessens Supply Chain Swings

Any industry will benefit from better balancing of supply and demand, but the need is intensified for semiconductor suppliers and their trading partners. Many products have long manufacturing lead times that complicate quick reaction to demand swings. New fabrication capacity is expensive and results in capacity increasing in large increments. The high fixed cost and rapid depreciation involved motivate semiconductor manufacturers to "push" products through the channel instead of letting customer demand "pull" needed quantities (Fig. 1). Using inventory to buffer demand swings is unappealing because rapid price reductions for semiconductor products can leave suppliers or customers holding overvalued or obsolete inventory.

These challenges are driving semiconductor suppliers to investigate better ways to manage production and distribution commitments to customers.

Today, demand forecasting/management and order-book reconciliation are largely manual processes. The use of e-mail, fax and electronic data interchange (EDI) delivers modest improvements, but serious problems remain, including data inaccuracy and unacceptably long communication lead times. Many semiconductor suppliers use information from internal production systems to optimize their operations, but take little account of data from trading partners. Most will share capacity availability information with customers only when pressured to do so. A major reason for discounting customer input during capacity planning is that customers often exaggerate demand forecasts when supplies are tight to improve their chances of getting required quantities.

 
1. Various factors encourage semiconductor manufacturers to “push” products through the supply chain channel rather than let customer demand “pull” needed quantities.

To overcome the current constraints, trading partners must establish collaborative interactions that rationalize and integrate their demand forecasting/management and order-book reconciliation processes. Semiconductor suppliers must make their supply streams visible to customers and, in return, customers must provide realistic demand forecasts. Collaboration will strengthen customer relationships and increase trust as processes are linked and realistic operating information is shared.

Many semiconductor suppliers use real-time capable-to-promise (CTP) systems internally to commit orders against inventory and available production capacity. Exposing CTP information to customers — enabling them to see actual supply capacity — will reduce the uncertainty that leads customers to exaggerate demand forecasts. Establishing collaborative demand-planning initiatives with key customers, including procedures for calibrating the accuracy of the forecast, will reduce overproduction that leads to excess inventories. (The collaborative interactions between suppliers and customers are depicted in Figure 2.) Real-time CTP information also improves order-book reconciliation by helping suppliers understand the consequences of accepting an order — such as increased delays or expenses caused elsewhere.

 
2. Collaboration links processes and promotes information sharing, leading to increased trust and stronger customer relationships.

Customers will demand collaboration sooner rather than later, so suppliers should begin to prepare now. Applications supporting collaboration are immature and often require integration and augmentation to provide full solutions. Greater obstacles to collaboration lie elsewhere, though — they include ensuring that data is complete and accessible, business processes are rationalized and re-engineered for a collaborative environment, and strategic trading partners are engaged and committed. Suppliers should take the following actions:

• Assess internal data completeness and quality, making necessary improvements. Important data to review include production capacity, cycle times, inventory, lot characteristic quality and yields.
• Jointly establish common measurement criteria with key trading partners. Internal metrics are often meaningless to trading partners. It is essential that each trading partner have the same "version of truth," which requires agreement on the definition of such metrics as forecast accuracy and commitment performance.
• Once metrics have been agreed on and internal data is in shape, begin pilots to provide interactive CTP information to key trading partners. Begin with just two or three strategic customers before attempting a broader rollout.
• Engage these customers in collaborative demand-forecasting pilots to improve the timeliness and realism of their forecasts, but temper forecast data with other assessments. For example, consider the financial condition of trading partners—and their customers—to assess the firmness of trading-partner forecasts.

In the cyclical semiconductor industry, even small improvements to the capacity-planning and order-book reconciliation processes will reduce inventory obsolescence risks, generating significant bottom-line savings. Suppliers can also improve customer service without increasing inventory risk. Better visibility of customers' changing demand patterns will not only reduce overproduction when demand falls off but also improve allocation accuracy in times of constrained supply. The financial penalty from shortcomings in these processes can be substantial. Cisco Systems, a large buyer of semiconductors, illustrated this point recently when it was forced to write off $2.5B in inventory purchased because of overly optimistic demand forecasts. Cisco is now using software from Manugistics Group to implement a private marketplace that will enable it to collaborate on demand and supply with its suppliers.