Advantest Moves Into Non-Memory Markets
By Mari Iwata, Electronic Business Japan -- EDN, August 6, 2004
In a bid to make itself less vulnerable to the silicon cycle, major semiconductor chip tester manufacturer Advantest’s two-fold strategy is to reduce its dependence on memory testers and rationalize parts procurement. The president and CEO, Toshio Maruyama, disclosed the strategy in an interview with Electronic Business Japan.
Electronic Business Japan: Are you booking plenty of orders for testers?
Maruyama: We have seen a surge in orders since the third quarter of 2002.
EBJ: How long do you think the current boom will last?
Maruyama: We expect order taking to go from strength to strength for the next one to two years. In fiscal 2001 Advantest experienced a massive drop in sales, down 63 percent year on year. But I don’t think we will see a repeat of that, because more and more semiconductors are being applied in various kinds of products such as mobile phones, DVD players and automobiles. Furthermore, even if sales of computers in advanced countries level off, sales volumes in China, Russia and other emerging markets will be rising.
EBJ: It has been suggested that orders for semiconductor manufacturing equipment may slow from around the end of 2004. How do you see this outlook?
Maruyama: We are optimistic because 300mm wafer process lines are being set up all over the world and, at the same time, applications are proliferating and demand is broadening geographically.
EBJ: How big is Advantest’s share of the memory tester market?
Maruyama: It’s about 60 percent for the entire memory tester market. We would like to raise that to 70 percent. Our shares of the DRAM and flash memory packaging testing markets are around 80 percent. For DRAM wafer testing it’s about 60 percent but for flash memory wafer testing it’s only 20 percent.
EBJ: In recent years, Advantest has been expanding the scope of its business from memory testers to include testers for non-memory chips. What are your objectives?
Maruyama: One of our objectives is to mitigate the impact of the silicon cycle. Although it may not work perfectly, because business often deteriorates for non-memory testers at the same time once the silicon cycle is heading downward, nevertheless it should work to some extent. Another objective is to expand sales of testers as a whole. Our share for testers for non-memory chips, such as logic ICs, is around 10 percent at present. We would like to double that by 2005 or 2006 to 20 percent.
EBJ: Why has Advantest been weak in the non-memory field?
Maruyama: In the 1980s Advantest was synonymous with logic testers, because our products offered the highest measuring speed and outstanding reliability. Then, an American company came along with a faster and better tester and gobbled up much of our market share. The new tester adopted “per-pin architecture” with which resources were allocated to each channel. Although it was an elegant concept, we didn’t embrace it because we thought it would prove too costly to be practicable. While we were biding our time, development of CMOS and other device technologies roared ahead, making it possible to manufacture testers at low cost.
EBJ: What is your strategy for raising the market share to 20 percent?
Maruyama: As of now, American companies still have an edge in technologies for logic ICs. But the large quantities of logic ICs being used in DVD recorders and mobile phones are prompting Japanese companies to vie for leadership. Because we are also a Japanese company, we can communicate with them in Japanese and work closely with them. Geography works to our advantage when it comes to developing good communication with these companies. Another strategy is to propose new concepts. In the past we lost because of the concept, but this time Advantest was early to propose a new concept, open architecture. We have established a completely independent consortium in the U.S. to promote this open architecture so that our concept will be accepted by many companies. Advantest is supporting this consortium in terms of money and human resources, but our clients are also offering resources.
EBJ: Why have you opted for open architecture?
Maruyama: The fact that hardware and software vary depending on tester manufacturers makes life difficult for our customers. For SOC devices, in particular, they often want to modify the circuit configuration slightly. In that case, if the tester can be used with a little modification, testing costs can be reduced. An open standard is the only way to achieve this. This strategy is a double-edged sword. Because any company can be a player, our core competence cannot make us a winner. On the other hand, it certainly will be advantageous that development costs can be kept low because of standardization. Competitors whose market shares are 30 to 40 percent are bound to lose quite a bit of business once this architecture is adopted. But because Advantest has only a 10 percent market share, we won’t lose much. That’s why we could decide to go ahead with the open architecture concept. Our first product based on this architecture is the T2000. To verify the architecture, we made this product with a comprehensive, high-end specification. Even so, it’s selling very well. In terms of units shipped, we are already into three digits.
EBJ: Can Advantest enjoy a good profit margin?
Maruyama: Well, logic testers are based on high-mix, small-quantity production. Also, support requires considerable time and labor. The profit margin for low-end products is slim because the barrier to market entry is low and competition focuses on price. To solve this problem, we intend to reduce the development cost by using open architecture. In addition we are rationalizing everything we can. We have thoroughly revamped the production line, cutting the lead-time to two weeks.
EBJ: Advantest has done a lot of restructuring and rationalizing over the past two years. Has employee morale suffered?
Maruyama: Clearly, morale is undermined when employees see their co-workers leaving the company. In that situation, a promise to share profits with the workforce once the business improves will meet with a skeptical response since nobody knows when the company comes back into the black. Employees simply won’t trust management. Instead, I didn’t cut R&D expenditure even in the worst time. That kept employee morale riding high. When two-thirds of Advantest’s sales evaporated, the ratio of R&D expenditure to net sales rose to nearly 30 percent. This policy resulted in commercialization of our new logic IC testers such as T2000.
