Top 10 predictions for semis in 2008 Mark LaPedus (01/04/2008 11:35 H EST) URL: http://www.eetimes.com/showArticle.jhtml?articleID=205208561 Happy New Year! 2008 is just beginning to unfold in the electronics industry and there is already uncertainty in the air based on recent industry data. Industry forecasters seem to have different opinions about the overall outlook for the semiconductor and IC-equipment markets in 2008 and beyond. To help sort out the confusion in the market, I have released my own chip forecasts--and other predictions--for 2008. 1. Pardon my pessimism, but I see zero (that's 0!) growth for semiconductors in 2008 over 2007. Going into 2008, there are some bad signs: the subprime fiasco, soaring oil prices and a U.S. presidential election that could bring more uncertainty to the party. Regarding the microeconomic issues, this is a real concern to me: the DRAM market remains lousy. I don't see any good signs for DRAM makers in 2008. There's too much capacity coming online, especially from Taiwan. Micron and Qimonda are losing money. Hynix is expected to go into the red. And Samsung is holding on. Here's the problem: Microsoft's Vista has never been the DRAM driver that it was supposed to be. Plus, specialty memories, which were supposed to boost the DRAM crowd, never really happened. Meanwhile, NAND flash remains volatile. I see continued ASP woes. The PC market, which drives microprocessors, remains boring and will show lackluster growth in 2008. Visualization remains interesting, however. Handsets, which drive DSPs, NOR flash, RF and other devices, remain boring. How many people are jazzed about the iPhone these days? Consumer products are also boring. I'm still waiting for some new and interesting killer apps. There are some bright spots: Analog is resurgent. So is WiMax and 4G. However, we need more killer apps. --------------------------------------------- 2. Pardon my pessimism again, but the semiconductor equipment market is expected to see another downturn in '08. Fab expansion and capital spending, in my opinion, are projected to fall off a cliff. My projection: Try a 20 percent decline for IC equipment in 2008. I hope I'm wrong. But here's the reality: The foundries are not expanding or buying that much gear. The DRAM-ers are suffering. The NAND guys are expanding but the market will slow. Intel may buy more gear, but the company has learned to use equipment more efficiently. ---------------------------------------------- 3. Here's a no-brainer: More consolidation is expected in semiconductor equipment. We've seen plenty of consolidation over the years. This year, I think we'll see some real blockbusters. For the last two to three years, I have predicted that Applied and Canon will form a joint venture in lithography. This year, my predication may come true. Canon is late in the immersion lithography game. Applied wants to get into lithography. A match made in heaven (or hell)? Applied is also paying a lot of attention to solar these days. It's no wonder, given the growth in the arena. In 2007, Applied acquired some solar equipment companies. In 2008, perhaps it might make a play for Amtech, BTU, Spire, or, for that matter, Oerlikon. I think Novellus is ripe for a takeover. Look for Tokyo Electron Ltd. to make a play on Novellus. Forget the long-awaited Lam-Novellus merger. Won't happen. Lam just bought SEZ. So that tells me the wafer cleaning crowd is in play. Akrion, FSI and others could be takeover targets. I see could KLA-Tencor buying Nanometrics. Rudolph may also make a bid for Nanometrics. A bidding war in the works? In ATE, Credence is a goner. I'm sure Teradyne will gobble them up in order to gain logic ATE share. Verigy may make a bid for LTX. ---------------------------------------- 4. Speaking of chip manufacturing, extreme ultraviolet (EUV) lithography will die a slow and painful death in 2008. EUV was supposed to process wafers at the 65-nm node. Then, it got pushed out to 45-nm. Then, 32-nm. Now, it's targeted for the latter stages of 22-nm. I know ASML and Nikon are working hard on EUV, but it appears that EUV is making no real progress. Now the industry faces a huge question: Do we pour more money down the EUV drain? Another one: What will chip makers do for production at the 32-nm node and beyond? Sadly, there are no clear answers. I have no answers as well. EUV is a dog. Nano-imprint is not ready. Multi-beam e-beam is a science project. Look for 193-nm immersion and double patterning to save the day -- at least for now. Earth to the industry: We need a breakthrough for high-index fluids in 193-nm immersion. ----------------------------------- 5. Here's something I, as well as many others, have been wondering about for a long time: When will IBM spin out its semiconductor unit? 2008 could be the big year. Over the last several years, IBM has transformed itself from being a hardware company into a software and services provider. Do semis fit this strategy? The answer: Yes and no. On one hand, IBM is still a major provider of ASICs, processors and foundry services. Big Blue's secret sauce: The chip division is also a major IP provider. In fact, some call IBM the "fabless foundry.'' But on the other hand, I could see IBM spinning out its chip unit and forming a venture with private equity firms. I know the financial markets are bad, but how about IBM and Francisco Partners owning a 50-50 stake in IBM Micro? 6. Speaking of IBM, the sneakiest press release issued during the holiday season occurred when Big Blue licensed its 45-nm process technology to Chinese foundry player Semiconductor Manufacturing International Corp. (SMIC). What does this tell me? SMIC can't devise its own 45-nm process. It doesn't want to get sued by TSMC again for alleged IP theft. The real answer: It also means that SMIC is positioning itself as a takeover target for someone. Loss-ridden SMIC has already been dropping hints about going private. In 2008, I could see Chartered and SMIC forming a joint foundry company, which could threaten the big boys in Taiwan: TSMC and UMC. Chartered is part of IBM's "fab club." Now, SMIC is tiptoeing into the club. ---------------------------------------- 7. I can't see how Freescale has benefited from going the private-equity route. Since it went private, Freescale is spilling red ink. And a lot of it. Freescale is looking to go IPO again. My prediction: I look for a breakup of Freescale to salvage the company. Infineon could buy the wireless and/or wireline and/or auto chip segments of Freescale. Or, for that matter, Infineon may buy the entire company. Maybe NXP or ST may buy some of the pieces. ------------------------------------ 8. I also can't see how AMD can continue on its downward spiral. It is losing vast sums of money and market share. The company recently obtained some cash from the Mubadala Development Co. of Abu Dhabi, but that's a stopgap measure. So who will rescue AMD in 2008? One Web site recently suggested that Apple may be the likely suitor. That's BS! What a lousy fit. Why would Apple need a loss-ridden chip unit? I see three possibilities for AMD: A) IBM buys a stake in AMD. IBM is a technology partner with AMD and IBM may want to compete with Intel again. Several years ago, IBM tried to compete in the x86 processor market, but Big Blue ended up getting black and blue in the process. B) AMD goes the private equity route. Private equity is pass and doesn't guarantee happiness. Ask NXP or Freescale. C) Samsung may want to re-enter the processor business and buy AMD. Once upon a time, Samsung attempted to compete in the processor business with the Alpha chip. Remember that flop? I doubt even mighty Samsung wants to compete with Intel in the processor market. Samsung is a memory house. Logic seems illogical for Samsung. My prediction: A or B. --------------------------------------- 9. I see something happening with two memory houses: Micron and Qimonda. First, let's talk about Micron. Amid losses, Micron has already dropped hints about looking for some kind of partnership. Like AMD, Micron may go the private equity route. Perhaps Micron may combine its memory operations with Germany's Qimonda. Strange combination, but the DRAM market is horrible and the two companies cannot afford to spill red ink throughout 2008. If that is not an option, Qimonda may end up merging with DRAM partner Nanya of Taiwan. It may take these types of ventures to compete against Samsung. ------------------------------------------ 10. Finally, something has also got to happen in Japan. There are too many slow, lumbering IDMs in a rough business. Toshiba is the exception to the rule. They are doing well. Sony is more or less a goner in semis. Who's next? Epson, NEC Electronics, Oki, Renesas and Sanyo cannot take another bad year in semis. I look for Sanyo to jettison its semi unit after a previous effort to sell the unit failed. Oki and Epson will remain in the business, but only as tiny players. That leaves me to Renesas and NEC Electronics. In the distant future (beyond 2008), I see Toshiba absorbing NEC Electronics. Renesas will survive in the long term -- as a fabless company. Within the next 2 years, DRAM specialist Elpida and its Taiwan partner, Powerchip, will merge. http://www.eetimes.com/news/design/showArticle.jhtml?articleID=205208561 --

※ 發信站: 批踢踢實業坊(ptt.cc) ◆ From: 140.112.48.60