Taking a business concept from greenfield to top-tier foundry
08/01/2003
Steve Della Rocchetta, Silterra Malaysia Sdn. Bhd., Kulim, Kedah Darul Aman, Malaysia
Wafer foundries have become an integral part of the semiconductor economy. Some analysts have forecast that foundries' share of wafer shipments will reach more than 40% of total volume within 10 years. This may require five to six more healthy wafer foundries to provide adequate support for the large and still growing semiconductor industry. There is a definitive outline of success factors required to turn a start-up company into a successful wafer foundry.
As the semiconductor industry matures, it continues to evolve into an industry of specialization. Semiconductor suppliers have been outsourcing increasing portions of noncore business to manufacturing contractors, so that suppliers can focus on core competence in product design and marketing. They are leaving assembly, testing, and wafer manufacturing to specialists who can provide better and cheaper solutions because of economies of scale and product specialization.
The wafer foundry industry is projected to grow faster than the semiconductor industry as a whole during the rest of the decade (Fig. 1). To support this growth, the competitive landscape is shaping up to support several mainstream wafer foundries that supply advanced CMOS technologies, which accounts for the largest overall revenue in the industry. There will also be specialty shops such as Jazz Semiconductor and Communicant that provide silicon germanium (SiGe) and gallium arsenide (GaAs) technologies.
Several promising new wafer foundries, each seeking to establish itself as a viable mainstream supplier, have started production in the last couple of years. My experience with several start-up foundries, especially as a member of the founding team at Chartered Semiconductor — taking them from small players to tier one competitors — enables me to provide a unique perspective on how foundries gain good reputations as reliable manufacturing partners, as well as how they earn high marks from customers.
The evolving wafer foundry landscape
According to Semico Research, wafer demand for the foundry industry will grow at a compound annual growth rate of 28% over the next five years, significantly higher than the 10% expected for the overall industry [1]. The opportunities are good for start-up foundries to emerge as leading suppliers against the current industry leaders. At last count, there are almost 10 pure-play wafer foundries offering CMOS process technologies and many more offer specialty process technologies for niche markets (see the table). For purposes of this article, the focus will be on the CMOS technology segment, which serves the largest application space and has the highest revenue and wafer volume in the industry.
Critical resources
There are two very important resources that must be in place before the start-up company is launched: adequate financing and a veteran management team. The senior management team must have experience in semiconductor manufacturing to understand the highly technical nature of the business. There is no substitute for the medals earned in the battlefield. Having the experience allows them to understand customers' requirements and avoid costly pitfalls. Their vision is critical in guiding the young company quickly up the steep learning curve.
 Figure 1. Wafer foundry share of capacity continues to increase. (Sources: SICAS and Silterra Malaysia) |
null
Silterra is an excellent example. Its management team consists of experienced veterans who had started semiconductor companies or had run worldwide fab operations at major companies for many years. The company became operational 18 months from ground-breaking in a greenfield site and shipped revenue wafers the quarter after that.
The geographical advantage
Semiconductor manufacturing involves many intricate steps, high-quality supplies, complex equipment, and a highly skilled labor force to support its activities. Therefore, a facility must be located in an area where a highly educated labor force and modern infrastructure are available. An ideal location is in an established high-technology hub, especially one that already has a strong packaging and testing presence, where the exceptional quality of life helps to attract valuable engineering resources.
Another important advantage is the close proximity to the backend service houses, which significantly reduces the costs and cycle time from the total supply chain. Customers could have the entire product, from chips to systems, built within this high-tech megaplex in a matter of several weeks.
 |
null
Two noteworthy technology centers, backed by their respective governments, illustrate the importance of factory location. The Science-based Industrial Park in Hsinchu, Taiwan, which houses the two largest wafer foundries, and the Kulim High-Technology Park in Malaysia, where Silterra is located, offer this geographical advantage. Both of them fit the model very well with an abundant supply of highly skilled labor and well-established chip-packaging and test houses in addition to system assembly facilities. They have become an integral part of the supply chain in the electronics industry and will be so as long as they remain competitive.
Focus on the right technologies
There are many process types and variations that serve the electronics, computer, and communications markets, including CMOS logic, DRAM, nonvolatile memory, analog, high-voltage, and rf technologies. Each technology addresses specific types of designs and applications. The start-up foundry, at least initially, must stay focused on its target market segments and required technologies. You cannot be everything to everyone, and attempting to provide more process capability than resources can accommodate leads to disastrous results.
Focus is key to success. Focusing the foundry increases its chance of success, maximizes its return-on-investment, and establishes its credibility quickly. Company executives must remain true to the strategic business plan, which can mean turning down potential business because it is inconsistent with the foundry's focus.
Wafer foundries that target leading-edge technologies must be able to keep up with technology trends. However, it is not necessary for a start-up foundry to use all of its resources to provide "bleeding-edge" technologies. Each new technology generation has challenges and pitfalls that the start-up should, in its early stages, avoid.
Let established companies with deep pockets take on the challenges first, but be prepared to ramp up compatible processes within 12–18 months. This "close follower" strategy allows the start-up to maximize its modest R&D budget and still benefit from the enormous opportunity that leading technology brings.
The sweet spot for a given technology is often two years after pilot production. Most customers are likely to shy away from adopting new technologies too early. An unstable process often requires many redesigns and re-optimizations. This strategy applies whether the foundry develops its own process technologies in-house or licenses them from a partner. A more mature technology node requires substantially smaller investment to bring to market. As long as the schedule delay is less than 12–18 months, the opportunity cost is relatively small.
Since new process technologies are generally more expensive, only applications that really need to push the speed and power envelope are likely early adopters (Fig. 2). High-end graphics chips, where chips must manipulate millions of polygons/sec to satisfy game graphics designers, are often a driver for new technologies. Field programmable logic arrays are another example. These devices offer great flexibility, but must rely on advanced technologies to compensate for their inherent inefficiencies compared to ASIC alternatives. The majority of other applications will eventually migrate to new technologies, but only after their costs reach parity and design IPs become widely available. Historically, that crossover point is one to two years after pilot production starts.
 Figure 2. Close follower strategy allows a company to participate during a high-growth period with lower investment. |
null
Experience has shown that customers prefer to work with similar processes to minimize redesign work when taping out a new product, especially with analog circuits. By providing technologies compatible with those of the market leaders, the start-up can offer products similar in layout design rules and electrical parameters, which means less redesign work for the customer.
Information technology
Information technology plays a very important role in manufacturing. A well-integrated IT infrastructure provides access to essential databases for process parameters, electrical data, lot status, and priorities. One advantage for a start-up is that it has no legacy system to maintain. A flexible, state-of-the-art system that allows engineers and logistics personnel to gain instant access to all relevant data is critical for the operations staff to maintain high yield and efficiency. Logistics personnel also need to have a clear picture of WIP to forecast inventory levels and shipment schedules.
Operational excellence
Management vision, engineering talent, physical location, IT investment, and market research all lead to one factor that really matters — the company's ability to execute the plan. Customers demand that their manufacturing partner maintain a world-class operation and deliver high-quality wafers with predictable yield and short cycle time. An emerging foundry has to demonstrate its capabilities by matching or even exceeding these parameters compared to those of the leading foundries. The foundry's credibility is established through consistently meeting commitments and superior customer relations. This is the basis for being market-focused. It allows a young foundry to concentrate on perfecting a smaller set of process technologies more quickly. Overextending its resources to properly execute its business plan is the most common cause of failure for an overly ambitious start-up.
Final words
Building a competitive wafer foundry is challenging, especially during one of the longest and most difficult downturns the industry has ever experienced. Is it even more challenging now that the industry has still not turned on; the longer the recession, the more difficult it gets. Fortunately, challenges bring opportunities. The down cycle accelerates the outsourcing trend as IDMs further delay or reduce their capital investments. New foundries will do well if they stay focused on their customers and have the discipline to learn how to execute before jumping into too many projects and overextending their resources.
Reference
- Foundry Suppliers: The Competition Heats Up, Semico Research, April 2003.
Steve Della Rocchetta is executive VP of sales and marketing at Silterra Malaysia Sdn. Bhd., 555 E. California Ave., Suite #2, Sunnyvale, CA 94086; ph 408/530-0888, fax 408/530-0877.