They still have roles to play, but economies of scale may force further consolidation in the long run.
For many years the trend in the semiconductor industry with regard to photomasks and chipmakers was to shed captive mask operations in favor of merchant photomask suppliers. This reflected a larger trend all along the supply chain with many companies moving away from vertical integration as, consequently, the foundry model grew.
“This was mainly driven by cost considerations,” said Franklin Kalk, CTO of merchant photomask supplier Toppan Photomasks. “The cost of R&D to stay current made it difficult to justify maintaining an internal (photomask) business—that really was the impetus for consolidation into large mask merchants.” It was only natural for merchant photomask suppliers to flourish.
Over the course of the last several industry cycles, however, a converse trend has emerged as leading-edge production technology has become ever more complex and costly. Captive mask shops have become a competitive necessity among first-tier device makers, which tend to be more vertically integrated.
“Now there are so few semiconductor companies at the leading edge, and they have scale,” Kalk said. “Those companies have enough scale to justify having a captive mask shop.” These large companies—IDMs like Samsung and Intel or foundries like Taiwan Semiconductor Manufacturing Co. (TSMC) and GlobalFoundries—all use huge numbers of masks. They have the economies of scale to maintain what has become a differentiator between first- and second-tier chipmakers.
So what are the implications for the merchant photomask industry? More consolidation? As the industry closes in on the 10nm node at the end of the decade will it still need merchant mask suppliers?
The answer is, as it always is in the semiconductor industry, an educated guess at best. But it’s clear that the big three photomask suppliers, Toppan, Photronics Inc. and Dai Nippon Printing Co. Ltd., aren’t going away anytime soon.
Leading edge drives merchant photomasks
Semiconductor Equipment and Materials International (SEMI) has forecast the worldwide photomask market to reach $3.5 billion in 2014. After reaching a market peak in 2011, the photomask market contracted 4% in 2012 to $3.2 billion. SEMI anticipates the market to grow again by 3% this year and another 3% next year, driven by advanced technology (sub-45nm manufacturing) and the growth of manufacturing in Asia-Pacific, namely Taiwan.
Meanwhile, photomask suppliers are seeing the uptick in the second half of 2013 that is being reported elsewhere in the supply chain. In reporting its quarterly earnings for the fiscal quarter ended April 28, Photronics CFO Sean Smith said the company was accelerating the installment of advanced photomask production tools in North America and Asia as the result of projected demand for advanced photomasks in Q4 of this year and into 2014.
“We are very optimistic as we get into Q4 about our growth prospects as a result of the high-end capacity, the leading-edge IC products coming online and with new opportunities for node migration and increased share,” Smith said. “And we expect that to continue into 2014.”
Toppan’s Kalk said his company was observing similar trends. “There appears to be traction out of the recent semi industry downturn,” he said. “I would say we’re optimistic about the second half of the year. It seems like what people thought was an upturn at the beginning of the year has pushed out a little bit.”
Kalk noted that today the merchant photomask business tends to mirror the chip industry in general much more closely than it did before. Driven by design starts, in the past photomask industry cycles tended to lag the chip industry by six months or more to as much as a year. “Now what we see is that the supply chain has become much tighter; people don’t let their inventories stretch as much as they did in the past.” Consequently, mask business cycles are both more moderate and much closer in terms of timing to those of chip manufacturing, due partly to the capital-intensive nature of industry in general and leading-edge photomasks in particular.
It is this capital-intensive nature of manufacturing at the leading edge, perhaps more than anything other factor, that has caused the reversal of captive and merchant mask trends. But coupled with this is the changing nature of end markets, namely an increasing reliance on consumer devices, which puts pressure on development and manufacturing cycle times.
Over the last two years, both leading foundry and logic companies have invested heavily in and expanded their captive operations quite rapidly, said Amitabh Sabharwal, general manager for mask etch products at Applied Materials. With the cost of a 45nm mask manufacturing line costing anywhere from $200 million to half a billion dollars, these are mammoth investments. “It’s not something you do on a whim,” he said.
Therefore having the economies of scale that Kalk mentioned, not to mention the deep pockets of companies such as Intel or TSMC, makes it easier for them to invest in mask production.
Then there are the pressures on turnaround times in a tight supply chain. Even as market pressures demand the time from design to tapeout to be as short as possible, mask production times are a continuing headache for the industry. Extending optical lithography has required increased resolution enhancement techniques for more and more layers in devices, which consequently impacts mask write times, and not for the better.
Thus it further behooves leading-edge device makers to have a captive shop to provide masks for critical layers. EDA vendor Synopsys devotes a lot of time and effort in optimizing and reducing mask write times at their customers, according to Tom Ferry, the group’s senior director of marketing. When it comes to working with a captive mask shop at a chipmaker, given the complexity and the enormity of the data involved, the turnaround time in mask production is quicker when everything is handled in house; with an outside merchant mask shop more challenges arise.
“Clearly the captive model…is a more beneficial model for them,” Sabharwal said of the large IDMs and foundries that maintain captive mask operations. He further noted with regard to cycle times that chipmakers with captive mask shops gain an edge when it comes to testing new mask sets. “A merchant can produce it but can’t test it. That’s a huge capability that captives have,” he said. “The cycle time is rapid with a captive.”
Merchants still have roles to play
At first glance it might seem there is a dwindling place for merchant photomask makers at the leading edge, given the cost involved and the competitive benefits a captive mask operation provides. But this isn’t really the case, and the big three, Toppan, Photronics and Dai Nippon, aren’t likely to consolidate or otherwise leave the market anytime soon.
“What role do the big mask merchants have? I would say we have two roles,” said Toppan’s Kalk. One role is to provide standard photomask sets, he explained. The bulk of chip production, even on leading-edge devices, still involves non-critical layers with features above 45nm. Device makers consequently still find it cost-effective to rely on merchant shops for these standard mask sets; this part of the market remains solidly in the province of merchant suppliers.
The second role is that of partner. It’s a matter of strategy—the so-called earthquake strategy, as Kalk put it. Or to use a much older metaphor, chipmakers don’t want to have all their eggs in one basket. Thus the industry has seen a number of close strategic partnerships in recent years among merchant photomask suppliers and chipmakers—Photronics and Micron and Toppan and IBM, for example—spreading the development of leading-edge photomasks economically as well as geographically.
Having a merchant supplier capable of supplying photomasks for critical as well as non-critical layers—a second source—provides two primary benefits for chipmakers with captive mask shops. First, it provides the ability to handle capacity overflow. Second, it provides a source of photomasks should something catastrophic happen to their captive operations. Given the global nature of the chip industry, having a second source clearly makes sense. Moreover, it’s arguably essential in terms of business strategy.
But production won’t get any easier…or cheaper
So merchant photomask suppliers still clearly have roles to play. But with each technology node the number of leading-edge chipmakers dwindles, even as the remaining ones grow larger still. As the industry closes in on the 10nm node at the end of the decade, will there still be room for three major mask merchants? Or will the industry see further consolidation?
Toppan’s Kalk characterized the answer as coming down to who can afford to play at the leading edge. He noted that for the last decade or so, with every succeeding technology node, one or more companies drops out, choosing to go fabless. Assuming this trend continues, at 10nm there may be as few as six chipmakers with their own leading-edge production fabs, split between between foundries and IDMs.
How many will remain at the 7nm node? “Who knows, but the number is dwindling rapidly,” Kalk said.
This begs a further question: Will there still be a need for three merchant makers at 10nm and beyond? The last major consolidation in the merchant photomask space occurred in 2005, when Toppan merged with DuPont Photomasks. Then, as now, the question came down to economic scale.
“Is the scale enough right now, and does each company have the scale required to supply the market? That’s the question,” Kalk said. He further noted that while leading-edge mask technology is already expensive, if extreme ultraviolet technology proves commercially viable, the related mask technology will be more expensive still.
“If you don’t have scale, you’re less attractive to a semiconductor company as a potential partner.” If that proves to be the case for one of today’s big mask merchant companies, then just as we see the number of companies involved at leading-edge production dwindle, further consolidation could be coming for merchant photomask companies.
Editor’s Note: As explained at length elsewhere on this site this is a news story written by me for another publication. This originally appeared on Semiconductor Engineering; it holds the copyright, of course.