by Debra Vogler, Senior Technical Editor, Solid State Technology
The spirited debate about whether or not the industry should migrate to 450mm wafers continued last week at the International Symposium on Semiconductor Manufacturing (ISSM, Oct. 15-17, Santa Clara, CA). Provocative questions and commentary from some of the presentations could make one pause and ask a question: what if the industry decided to transition to 450mm wafers, but no one showed up at the party? And it’s not just the equipment suppliers, but the IC manufacturers, that may have to take a second look at a transition.
A case in point is the situation in which IC manufacturers now find themselves. “Today, the top 10 companies have 70% of 300mm capacity, and that percentage will go up every year,” observed Nick Bright, EVP at Lam Research. “Every wafer size change tends to consolidate the industry.” He pointed out that while 70% of the profit in the industry comes from companies with >$10B in revenues, about 100 companies making semiconductors today have >$100M in revenues, but they are trailing-edge in terms of where the profitability lies. Meanwhile, the number of companies and alliances investing >$1B in capex/yr going forward is coming down to about 10, and Bright predicts a further reduction in the number of players.
“There are some players at both ends doing very well in terms of profitability, but it’s the middle that is getting squeezed,” he told conference attendees. “As you go forward, only the very rich can afford to keep up the game [of Moore’s Law].”
In the past, the priorities for semiconductor manufacturers have included technology to shrink, device reliability, cost, and cycle times (CT). Bright told the audience that the new priorities for IC manufacturers are as follows: CT, operational costs, R&D costs, technology capability, and technology alliances — with device reliability now simply assumed. IDMs have been moving more and more to the foundries, enabling them to change their P&Ls by reducing R&D costs and operational costs separately, he noted. And IC manufacturers are going to need a much broader technology capability in terms of packaging and materials, especially if 3D structures take off.
Additional pressure on IC manufacturers comes from the divergence between logic and memory manufacturing in terms of the cost to produce 10,000 wafer starts. According to Bright, if the costs are normalized, the cost of logic has gone up about 42%, while that of memory has decreased about 29% — so it’s no surprise that the memory manufacturers are the ones buying most of the of new equipment, retiring 200mm equipment in the process. The discrepancy in memory manufacturers’ spending on new equipment vs. what foundries are spending is even more pronounced given that IDMs are moving more work to foundries to utilize their production capacity, he observed.
Turning his attention to equipment suppliers, Bright noted that consolidation in this sector has also occurred with every wafer size transition. There are only seven companies with revenue >$1B/yr — “a very important number,” he noted, “because below this size it’s very difficult to serve customers.” To serve customers, suppliers need a global infrastructure, investment R&D, as well as the ability to invest locally in joint programs with customers. Suppliers also need to have financial resources to get through industry cycles and invest for the future — i.e., they need a high critical mass. “Successful equipment companies have formed into two groups: multi-product companies, and dedicated product companies,” he said.
Amid the consolidation of equipment suppliers, previous wafer size transitions also drove productivity improvements based on the learning from previous generations. Cost reductions realized by IC manufacturers in going from one wafer size to the next were “dominated” by lithography, noted Bright, but in general, another benefit to the industry that came along with wafer size transitions was the opportunity to redesign equipment.
A transition to 450mm wafers, though, will come with additional baggage, and some of it may be undesirable. Would be benefits” from future lithography and wafer size changes are now being questioned — EUV’s anticipated arrival, still many years away, and its anticipated expense has the potential to slow or even stop the roadmap, Bright noted. Also, with the 450mm wafer transition comes the potential to over-consolidate the semiconductor industry, likely forcing massive R&D dilution in the equipment companies and cause a lot of support issues for existing 300mm fabs, he said. “There is very questionable ROI for the equipment companies, and even for the industry as a whole.” He also observed that the tendency now is for companies to not blindly follow the industry roadmap, but to drive their own roadmaps.
In spite of the potential for additional consolidation, however, Bright still sees a lot of opportunity for equipment suppliers. In particular, they should be able to take advantage of catering to the need for 3D devices and open up process windows at design nodes below 65nm. Also on the horizon is the need for equipment that enables the use of new materials. Amid these opportunities, Bright acknowledges that there will be further consolidation of existing companies and technology start-ups into bigger players over time.
Other opportunities for equipment suppliers include solar (especially for thin-film equipment providers) as well as FPDs, LEDs, and MEMS, areas in which “larger feature sizes will be very important going forward,” he said. And the opportunity in solar appears to have special significance. Noting that the solar panel industry is growing at a rate of 40% with only a 0.1% penetration rate, that is expected to increase over time, Bright asked conference attendees the critical question: with this vast opportunity looming for the equipment industry, why would it want to invest in 450mm manufacturing? — D.V.