The inevitable shakeout in the CMP equipment market
05/01/1998
The inevitable shakeout in the CMP equipment market
William O`Mara, O`Mara and Associates, Palo Alto, California
As more and more manufacturers discover the benefits of chemical mechanical polishing (CMP) for smaller and smaller design rules, an increasing number of CMP equipment suppliers are entering the field. As these new companies compete with early, more entrenched entrants, a shakeout is looming ahead.
CMP is taking the IC manufacturing industry by storm. CMP creates a flat surface across the entire wafer, making subsequent steps easier. For logic devices, such as microprocessors and application specific ICs, (ASICs), CMP makes it possible to achieve high manufacturing yields and high device operating speeds with multilevel interconnects, up to eight levels of metal. Memory devices such as DRAMs benefit from CMP even though they have only two levels of metal. For memories, the flat wafer surface after CMP increases the depth of focus budget available for lithography, allowing the designer to shrink the critical dimension and reduce chip size without reducing yield. A smaller chip size provides a critical cost advantage in the competitive world of DRAM manufacturing.
When devices reach a design rule of 0.35 ?m or less, CMP becomes a necessary part of manufacturing. For this reason, the CMP market has exploded in the past few years as polishing equipment suppliers have expanded production to meet the demand for CMP in new 0.35-?m fabs. Market growth rates of up to 100%/year have been common in the last few years, and further growth at rates greater than other semiconductor equipment segments seems assured as CMP equipment is installed in fabs around the world. The CMP polisher equipment market reached a worldwide total of $520 million in 1997, and will easily exceed $1 billion in value by the turn of the century.
 |
The CMP equipment business is divided among several companies at present. Major suppliers include IPEC, Strasbaugh, Speedfam, and Ebara. IPEC and Strasbaugh both got their start as tool suppliers when IBM selected their equipment for production in the mid-1980s. IPEC`s predecessor company, Westech Systems, supplied a single wafer tool to IBM`s Burlington, VT, wafer fab, while Strasbaugh`s dual head equipment was selected for IBM`s East Fishkill facility. Speedfam`s polisher, which incorporates five heads, claims a throughput advantage over these early tools, and has gained market share in the last two years. Ebara, a Japanese manufacturer that maintains close relations with Toshiba, designed its tool based on the requirements for CMP for advanced DRAMs developed by Toshiba in cooperation with IBM and Siemens. Ebara polishers include cleaning stations for dry-in/dry-out processing. Several of the joint venture production sites around the world have ordered Ebara`s equipment.
In addition to these early entrants, many other companies have entered the CMP tool market. So far a total of 20 firms have announced products, and the number seems to increase each month. Table 1 shows the current list of companies that have CMP tools for sale and indicates which suppliers also manufacture equipment for polishing or grinding silicon wafers. The CMP process is similar to the preparation of the polished surface of a bare silicon wafer, and most suppliers in the wafer polishing business have adapted their equipment for CMP. Other suppliers of CMP tools have established businesses in other kinds of fab equipment, including firms like Applied Materials and Lam Research.
 |
Worldwide CMP polisher market share for 1997
The market is divided by the four major suppliers mentioned above. A market share analysis indicates that the four major suppliers account for 86% of the 1997 CMP polisher market. The other 16 suppliers are competing for approximately 14% of the market (see figure). The market is in an unsettled phase, with some shakeout ahead; only a few firms can establish a viable position. In fact, most wafer fab equipment categories are dominated by a single supplier, termed the "800-pound gorilla" of that market (Table 2). Another alternative is the possibility of market segmentation, in which the market divides into major categories, each with its own dominant supplier.
 |
It is too early to tell whether the CMP equipment market will be a single large market, or will segment into markets for oxide CMP, metal CMP, and the like. However, the outcome will produce only a few suppliers, including one dominant supplier for each segment. Other suppliers are left competing for second or third place, or searching for a niche market that they can serve. The dominant supplier gets the major share of the market, but its share of the profits are even greater than its share of sales. With these profits, the dominant supplier makes improvements in the equipment, service, training, and other critical business areas to ensure continued success. Once established, the dominant supplier is difficult to displace.
It will take a major investment simply to become a minor player in the CMP market. A rule of thumb has been provided by William Davidow for the cost of entering a new market [1]. Davidow`s Law says that the cost of market entry is equal to 70% of the projected sales of the market leader. Assuming that the CMP market will exceed $1 billion by the year 2000, and that the market leader will have 35% of the market, or $350 million in sales, the budget for market entry should be $250 million. This will be enough for about a 10% market share. Everyone recognizes the cost of equipment development, estimated in the range of $20-30 million for a new piece of fab equipment, and the expenses of market development, training, and service capability.
Not all of the 16 "Other" suppliers shown in the figure can establish a 10% market share or anything like it. Moreover, potential dominant suppliers have to spend even more than estimated by Davidow`s Law in order to have a chance for the top spot. This means that market development and other expenses will exceed any profits generated by the business for many years.
The successful company will make the transition away from technology as quickly as possible. Although the technical capability of a major item of IC manufacturing equipment such as a CMP tool is obviously of some concern, the long term success of the company depends on creating an identity in the customer`s mind with the "product," rather than the technology itself. This "product" is actually more than the equipment item; it also includes all the additional goods and services that are necessary to make the tool work successfully. For CMP equipment, this includes materials, such as pads and slurries, as well as slurry distribution systems, metrology, and other critical items. The successful vendor is the one whose tool integrates with these other elements to create a "whole product" that produces the best result for the user. The next few years will decide whose product will capture this coveted position and transform the company into the 800-pound gorilla of the CMP market.
Reference
1. W. Davidow, "Marketing High Technology," The Free Press, NY, 1985.
William O`Mara is a consultant specializing in technical market research in semiconductor materials and processes. O`Mara & Associates, 445 Cambridge Avenue, Palo Alto, CA 94306; ph 650/323-2175, fax 650/323-0864, e-mail [email protected].