Today, the business realities of the semiconductor business can make success something that is hard won but not impossible. Using the right manufacturing strategies are the key to success.
With a hybrid, flexible approach, a fabless company can effectively use both the Known Good Die and Wafer Buy strategies, switching from one to the other when business conditions require. Such flexibility enables a company to purchase material at different points in the manufacturing process to balance material costs, risk, and yield to meet demand.
Figure 1: Tradeoffs between Known Good Die and Wafer Buy Strategies
For instance, in a Known Good Die strategy, direct material costs are higher, and risk to the fabless company lower, because the foundry actively manages most of the manufacturing process. The converse is true of Wafer Buy. Because the fabless company assumes responsibility for (and the inherent risk associated with) producing sufficient yield to meet demand, it enjoys lower direct material costs and greater control over the manufacturing process.
By selectively using the strategies when it makes sense, a fabless company can maximize each strategy's strengths while minimizing its weaknesses—and, thus, make the greatest impact on profitability.
That's what Qualcomm, the world's largest fabless company, has recently done. By actively and flexibly applying both of these manufacturing sourcing strategies, Qualcomm reduced its overall manufacturing costs and enhanced its margins and competitiveness.
The path to competitiveness
Generating the greatest benefit from a hybrid, flexible approach requires implementing the business processes and systems capabilities that can support each manufacturing sourcing strategy. It also requires knowing which strategy to use when, and keeping tabs on performance to ensure the company attains its desired cost and product delivery targets.
Implement the business processes and systems capabilities to support flexibility.
In a Known Good Die strategy, fabless companies need business process and systems capabilities to support activities that occur after the receipt of known good die from the foundry, when the company assumes ownership of the material. The main manufacturing steps from this point include three key backend processes: Assembly, Final Test, and Scan-Bake-Tape (SBT).
In Wafer Buy, fabless companies are responsible for the activities that occur after the receipt of an unprobed wafer from the foundry. Because they assume ownership of the material earlier in the supply chain, they must be able to orchestrate the additional manufacturing steps run by the foundry in the Known Good Die strategy: bump, sort/probe, and wafer-to-die conversion. A company that relies on Known Good Die will need to develop new supporting capabilities unique to probe:
- Procurement of probe hardware
- Probe hardware planning process
- Storage process
- Probe hardware repair and maintenance process
- Pick map process for probed wafers
- Wafer acceptance test criteria/limit process
Determine which strategy to use when.
With the necessary business process and system capabilities in place, a company must determine which strategy to use when. That decision is primarily a function of where a product sits in its lifecycle, as illustrated by the semiconductor yield curve.
Figure 2: The Semiconductor Yield Curve
When a product is developed and introduced, Wafer Buy is the right choice. At this stage, the product design is still being finalized, so a company should have as much insight into the product as possible. Wafer Buy allows for maximum visibility into all stages of manufacturing, enabling the company to collect valuable data during the bump and probe process to optimize manufacturing, more deeply understand design issues, and begin to improve yield prior to high-volume demand. Additionally, because the complexity and risk of managing the supply chain at this stage are low given low initial sales volumes, it's easier for the company to manage the supply chain itself and minimize costs.
As the product moves into the growth stage, demand begins to ramp up and yield becomes more volatile. At this time, minimizing risk to production and mitigating any potential supply issues is critical—which is one of the biggest strengths of Known Good Die. Paying the foundry management fees for known good die may be less costly overall during growth, as it reduces supply risk and enables the company to maximize production.
As the product matures, sales level off and decline, and yield becomes more predictable, the optimal strategy once again is Wafer Buy. In the twilight of a product's lifecycle, managing cost becomes paramount to maintain as much margin as possible in the face of pressure from newer competing products. By regaining full control over the manufacturing process, a fabless company can stop paying management fees to foundries and focus on driving down direct costs by outsourcing manufacturing to low-cost suppliers. It also can quickly shift production priorities to respond to inconsistent demand that's common at this stage.
Technology also plays an important role in determining which strategy to use. When a product using new technology (such as 10nm to 14nm) is introduced, yields are not stable for initial wafer lots—thus making Known Good Die more suitable. Yields for older, stable technology (such as 28nm and higher) will stabilize relatively quickly, making Wafer Buy the optimal strategy for most of the product's lifecycle.
Monitor the performance of each strategy to ensure operations & cost optimization.
However, simply plugging in the right strategy does not guarantee results. A company must continually monitor three key factors to understand how well it's executing each strategy and whether opportunities for improvement exist.
One is the product yield curve. Because a products' yield curve changes as it moves through the product lifecycle, companies should monitor yield closely. Steady yield calls for Wafer Buy and volatile yield signals a likely switch to Known Good Die.
A second is test hardware economies of scale.In Wafer Buy, a fabless company must purchase the test hardware needed for probe manufacturing operations—an upfront cost. But it also can create economies of scale by applying Wafer Buy to additional products over time, thus reducing long-run average costs.
A third is operational efficiency metrics, including on-time delivery of material from foundries and cycle time. Such metrics enable fabless companies to not only determine if they are executing the strategies efficiently, but also to identify whether foundries are delivering on their commitments and suppliers are being appropriately managed through bump and probe activities.
Take action to boost the bottom line
The global semiconductor industry is forecast to maintain steady growth in the next few years, expanding by 2.3% to 3% between 2015 and 2017. Some analysts are even more bullish, calling for as much as 6% growth in 2016.
But top-line growth is not the biggest problem fabless companies face. Given the dynamic nature of the semiconductor industry, and the increasing pressure from investors for better returns, fabless companies need to think more creatively about how to address their increasingly unwieldy cost structure to shore up their bottom line. In particular, they should adapt their sourcing processes, yield management capabilities, costing methods, and enterprise resource planning (ERP) systems to flexibly support both Wafer Buy and KGD strategies, which can help them rein in COGS.
This kind of creative thinking leads to improved margins and competitiveness. Has your fabless company adopted this kind of strategy? Let us know in the comments section below.
James Wildenburg co-authored this piece. Wildenberg, also a managing director for Accenture Strategy, helps communications, media, and yechnology companies around the world develop and execute transformation initiatives that enable them to respond to changing customer demands and market opportunities. He focuses on the areas of customer care, operations, mergers and acquisitions, digital disruption and business strategy, and is a prolific author on topics related to supply chain and organizational performance. Wildenberg is based in Atlanta.