Extracting value from outsourcing

READ OTHER DEC./JAN. ARTICLES:  Contents, Dec. 2006/Jan. 2007

Even when outsourcing production tasks, IDMs want to use their internal resources to add value by reducing product-development cycle time and by developing more cost-effective and competitive products. To do that, they need to partner with the appropriate SEMs and ATE vendors to develop cost-effective packaging and test solutions. For an outsourcing strategy to succeed, these partnerships must be “win-win.” They need to account for the market behavior as well as the needs of the SEMs and the ATE providers.

When choosing an asset-light strategy, IDMs need to remember that “asset-light” differs from “capital avoidance.” In the asset-light model, IDMs maintain some manufacturing equipment to support advanced development and strategic accounts and to monitor SEM pricing. They then work with their SEM partners to develop standards for wafer fabrication, packaging, and test of new products in the SEM factories. With this strategy, IDMs encourage SEMs to make capital-equipment purchases based on a solid business plan that acknowledges the SEMs’ profitability requirements.

Capital avoidance, on the other hand, is not really a strategy; it is merely an attempt by IDMs to avoid purchasing any manufacturing equipment at all. The IDM requires SEMs to purchase and maintain equipment independent of a business plan, forecast, or partnership. This approach tends to result in higher prices and less flexibility to respond to fluctuating markets.

IDMs also must realize that competition and benchmarking are crucial to the success of an asset-light strategy. If an IDM focuses only on price while ignoring strategic considerations such as packaging roadmaps and ATE platform alliances, the IDM is in danger of paying more than is necessary. Developing and maintaining multiple suppliers and constantly benchmarking the SEM service market for both technology and pricing can help avoid this pitfall.

For IDMs, finding value within the SEM market involves more than negotiating the lowest price. SEM services have diversified in type and have increased in quality. SEMs now offer true turnkey services, taking a customer’s design from wafer fabrication and sort through to assembly and final test, including selecting the tester platform and developing the test program.

These services are not limited to the production of digital devices—they are becoming pervasive for analog and system-on-chip (SOC) devices as well. Yet, some IDMs might be unwilling to take full advantage of these services if they consider test development to be competitive intellectual property—especially with analog, RF, and SOC devices.

In these cases, IDMs can still take advantage of the productivity, yield-enhancement, and other engineering activities that SEMs provide. An IDM may outsource product development totally to SEM partners when limited intellectual property is involved to reduce time-to-market and to leverage the legacy product learning cycles that occurred within the SEMs. Because SEMs deal with multiple IDMs and fabless companies, their ability to leverage product-development cycles from multiple sources becomes a new source of value for IDMs.

IDMs must choose production-test platforms for the SEM market carefully, because SEMs will hold them responsible for underutilized equipment. The IDM should try to choose a test platform that is part of a SEM’s installed base or its future tester roadmap. Because SEMs build their test-cost models around capital depreciation and utilization, the IDM can reduce costs by choosing a scalable platform that can serve for a long period of time (ideally, five to 10 years) and that is selected with a SEM’s installed base and other customers’ requirements in mind.

In addition, to achieve the lowest possible test cost, the IDM must develop a strategy of platform indifference. By supporting test on multiple ATE platforms, an IDM enables its SEM partners to improve the utilization of their test systems and reduce capital expenses. IDMs need to become comfortable with the fact that SEMs will want to port mature products and test programs from one ATE platform to another.

The traditional IDM/SEM test support model (Figure 1) follows a capital-avoidance strategy and exhibits these characteristics:

Figure 1. In the traditional IDM/SEM test support model, the IDM retains test engineering responsibilities. The asset-light model shifts responsibilities to the SEM.

• The IDM maintains all test engineering resources.
• The IDM develops the test program while developing and debugging the silicon.
• The IDM optimizes the test program (to provide test-time reduction).
• The IDM provides all updates and supports all sustaining activities, including yield enhancements, program modifications driven by customer feedback, and modifications driven by test-platform changes to hardware or software.

In this model, the IDM performs all the engineering activities and provides the SEM with a billing mechanism. This model does not address time to market or use the SEM’s full potential. Because SEMs do business with multiple customers—some of whom take advantage of the entire value a SEM offers—this flow could be considered a “competitor subsidy” model.

The resources and capital expense required to support a SEM’s true value proposition are in fixed costs, which are applied against all of its service charges. This makes it extremely important for an IDM to use as many of these services as possible, to maximize the value of the SEM partnership.

In the less-traditional asset-light model, the IDM relies more on the expertise of the SEMs and ATE vendors. The IDM must extract value from the elements that a SEM has to offer. If not, the IDM could pay twice—once to own and maintain its own large internal manufacturing resources and once more for the resources it uses within the SEM. An asset-light strategy could work like this:

• To protect intellectual property, the IDM may continue to develop and debug the initial release of the test program.
• Once the program has been qualified, it can be released to the SEM.
• The SEM optimizes the program.
• The SEM monitors program performance, makes yield-enhancement recommendations, and provides approved modifications.
• The SEM provides sustaining engineering work that could include approved customer-driven program changes, productivity improvements such as multisite development, and product-derivative test-program development.
• The SEM drop-ships fully tested devices to the IDM’s end customer.

Figure 2. In this value curve, notice that IDMs and SEMs agree on the value of time to market, but IDMs are more concerned with the capital cost of the ATE while SEMs are more concerned with the utilization of a chosen asset.

SEMs tend to evaluate utilization and cost down to the channel-board level of any given ATE asset. Overly configured ATE platforms (supersets) may provide some flexibility, but if the instrument resources used to provide that flexibility do not support a SEM’s cost model, the IDM can expect to pay a premium when other devices not requiring those resources are tested on it.

Equipment longevity is a sensitive issue for both the IDM and the SEM. Product life cycles tend to be short, and the integration of more and more functions onto a single chip often require the migration to a new test system every few years. In this case, neither the IDM nor the SEM can extract full value from an ATE capital investment. Therefore, platform selection and subsequent alignments must go beyond the functional capability of the ATE platform and should consider the following:

• program portability,
• platform scalability,
• software and tools,
• multisite efficiency,
• instrumentation and configuration roadmaps,
• market inputs, and
• cost of hardware.

Figure 3. The IDM and SEM can negotiate a baseline test cost threshold based on business opportunities that the SEM can expect for an ATE system. The alternative is a pay-per-use model.

If an IDM specifies a particular ATE platform without permitting mature products to be ported to another platform, the IDM may end up with a noncompetitive test solution and a non-cost-effective SEM strategy. The IDM can expect to pay a higher test cost as well as be held responsible for any unused capacity of the equipment it specified. Some platform-specific requirements may be unavoidable, but in high-volume commodity electronic markets, the IDM should work with the SEM to take advantage of any installed systems.

The two most important elements that can improve test costs are utilization and purchasing power. These will drive both the capital cost of the equipment itself and the charge out to a SEM’s customers. Volume purchases combined with higher utilization achieved by actual loadings and machine performance improve the return on investment. In turn, these benefits become the drivers for reducing the price of test that SEMs offer their customers.

Competition is also important for extracting value. Maintaining multiple ATE platforms is a healthy practice and serves as a “checks and balances” function for a SEM’s customers. A true multi-platform model produces competitive benchmarks that cannot be ignored by either the SEM quoting prices for test services or the ATE vendors competing to provide the platforms. Competition serves as a catalyst to drive continuous improvements to development tools, software and hardware, platform performance, productivity, capability, and price.

A baseline plan for loading (Figure 3) requires close coordination among the SEM, IDM, and ATE manufacturers. It involves some risk management for all parties. In practice, an IDM agrees to forecast a “constant” run rate against an annual forecast provided by its customers.

The SEM then agrees to manufacture products at that capacity independent of actual demand. The inventory might build up at times, but never to a level that is not ultimately consumed by either normal demand fluctuation or by forecasted demand. When either an unforecasted upside occurs or a late-year forecasted ramp occurs, the SEM can manage the capacity ramp with the built-up inventory. This plan also minimizes the chance that the SEM will be cycled to zero loadings, which would impact pricing.

Figure 4. The red curve represents production volume fluctuations throughout a product’s life cycle that can be addressed by establishing a buffer capacity above a baseline production capacity.

One way to manage the unforecasted upsides is to create short-term “flex capacity.” The SEM agrees to pay the ATE vendor a monthly fee and pays for test time at a negotiated rate when products are actually tested. The SEM customers—the IDMs—agree to pay a fee slightly higher than the base test cost in order to secure the uncommitted “flex” capacity they require to support their business product ramp cycles as depicted in the Figure 4 graph.

Because the success of outsourcing strategies depends on effective, well-managed partnerships between the IDM, the SEM, and ATE vendors, the ATE vendors must do their part as well. They need to adopt aggressive and flexible sales practices to support these models and drive the capital cost of their solutions to be in line with the markets they serve.

By fostering competition and multiplatform capability, IDMs can effectively drive the results they require. When IDMs enable SEMs and ATE vendors to provide and maintain test programs, the shared benefits can include higher utilization of total test assets, overall lower test cost, and more effective and efficient management of test capacity.