I must note that new 2002 has started straight away and brought us not only boring annual financial reports, but also a load of really fresh news that we had missed after a couple of New Year "exhibition" months. New technologies, processors, chipsets, memory products, a rough stream of scientific and technical conferences, exhibitions - life's bubbling over: you gape - you miss something - just like in good old times.
Now analysts still more optimistically look at events of 2002. In their opinion this year will stop staff reducing, will revive manufacture and restore the semiconductor branch. As soon as the peak of the crisis in the middle of 2001 has passed, the majority of market observers have begun to point at a direct analogy with the same unsuccessful 1998. That summer the industry has dropped to a deepest known minimum. Basing on similar economic forces of that old stagnation, the majority of marketing companies think that the remains of crisis'2001 will disappear in the nearest 0.5-1.5-year.
Analysts think that the restoration of the branch of equipment manufacture for the semiconductor industry will be the most painful. The common opinion is that the lithographic equipment market will be weak through entire 2002, and sales of lithographic tools will still worsen by approximately one fifth in comparison with 2001. If they sold about 800 plants the last year, this year buyers are expected only for 650 or even, according to some forecasts, for 500-550 plants.
But already 2003-2004 are expected to bring us the vigorous enough restoration. Morgan Stanley experts think that in 2003 there will be sold about 800 lithographic plants, and more than a thousand in 2004.
Anyway all analysts agree with one thing: the further, the stronger industry will need 193 nm equipment. What can ASML, Canon, and Nikon, leaders of this branch, offer clients in 2002?
According to the latest messages from industrial sources, all three companies have solved basic problems, that have arisen during the development of equipment for 0.10 micron and smaller chips manufacture. The main problem of the modern equipment production is known to be a constant delay with manufacture of calcium-fluoride materials, required for 193 nm lithographic optics production. It is known that Canon and Nikon have solved this problem independently, and ASML has found reliable suppliers "on the side". All three companies are ready to ship high-end 193 nm argon-fluoride plants in 2002.
Now terms of the beginning of mass deliveries of 193 nm scanners with the high numerical aperture from ASML, Canon and Nikon are estimated to 12-14 months. So there's no hope to expect mass deliveries of such tools up to the middle of 2002. According to some data, customers might expect not more than 30-50 plants from each company up to the end of this year. Certainly, Intel is the largest customer of this equipment that is extremely interested in getting it already this year, but, by the way, it's still unknown whose 193 nm scanners - from ASML, Canon, or Nikon - will be massively bought.
While (without any noise in press) 193 nm tools, actual for today, are being fine-tuned, the increasing attention is paid to the following generation of lithographic equipment - 157 nm for 70 nm and smaller technical processes. I must note that last tendencies of the entire semiconductor industry, namely, striving to integration and industry coalitions, have also densely touched manufacturers-developers of the equipment.
From the very beginning of this year there were about forming the new alliances for development of the equipment for 70 nm lithography.
One of them is between Japanese Semiconductor Leading Edge Technologies (Selete) and three competing Japanese photomask manufacturers - Dai Nippon Printing, Toppan Printing, and Hoya Corp. Companies have agreed upon conducting joint researches aimed at the development of masks for 0.07-micron technical process with use of 157 nm fluorine (F2) laser. The research will continue until March 2004 within the framework of the alliance, and then tests of the new equipment will begin.
Another alliance is a joint effort of Tokyo Electron and Canon that should engage in manufacture debugging and the elimination of possible errors that may appear at fluoride laser application. Tokyo Electron must deliver the evaporation plant ready to work with 300 mm plates to Canon's laboratory where they are constructing an experimental 70 nm lithographic plant.
Besides, enough authoritative LEEPL (low-energy electron-beam proximity projection lithography) consortium, created for development of new lithography methods with use of e-beam, was updated with Matsushita and Sharp, and also Nippon Control Systems, manufacturers of mask processing plants. Now the amount of LEEPL consortium members that had been created by Tokyo Semitsu, LEEPL Corp., and ubiquitous Sony in June, the last year, has reached nineteen (among LEEPL members are such serious companies as NEC, Rohm, and Texas Instruments).
Provisional dates of release of LEEPL's lithographic equipment prototypes have not even sounded yet. Though, Tokyo Semitsu has already announced the beginning of factory construction for producing up to 30 units of such equipment per month.
The position of Dutch ASML, that has addressed to this industry leaders with the offer about joint efforts in development of new equipment generations with the purpose of R&D works depreciation, and also reduction of risk of conducting such development alone, is indicative in this respect.
We should recollect these curious expenses: the development and launch of the first ASML product, g-line stapler, has costed the company $50 million in the beginning 1980. The company has already spent about $500 million for the present Twinscan platform. The development and manufacture of new 157 nm scanners as well as the EUV systems equipment, expected in 2007-2008, may cost each company $1 billion and more!
Meanwhile ASML believes EUV LLC is a sample of such consortium for common development of EUV equipment that also consists of AMD, IBM, Infineon, Intel, Micron, Motorola, and USA national laboratories.
However, there is more sense to speak of partnership in the field of lithography development with a reference to relations between equipment manufacturers and consumers. Companies busy with lithographic equipment manufacture, keep a keen and a jealous eye on each other's successes and failures.
As it's known, Nikon remained the leader of this market in 2001 by the amount of sold tools. On the second place, according to Morgan Stanley (on the basis of VLSI Research data), is Canon that shifted ASML to the third place. The fourth place is occupied by North American Ultratech Stepper.
To be exact, Nikon's share of the world sales of lithographic plants in 2001 has made 41%, having increased from 35% in 2000. However, actually they sold only 330 plants in 2001 - by 24% fewer, than 434 plants in 2000. But here I'll remind that the sold plants total has decreased from 1232 in 2000 down to 805 last year.
Canon's sales have increased from 23% in 2000 up to 31% in 2001. That is 284 and 250 plants accordingly (though, this relative growth has made actual 12% annual reduction of sales volume for the company).
ASML's business was not so successful: from 29% of presence in this market in 2000 the company has slid to 24% in 2001. That is the reduction of delivery volumes has made 47% - 362 plants sold in 2000 and only 192 in 2001. And all this - in spite of the fact that the present company results were complemented with Silicon Valley Group Inc. data, bought by ASML for $1.6 billion in April last year.
Ultratech's share has decreased from 6% to 4%, delivery volumes has dropped from 77 sold plants in 2000 down to 33 in 2001 accordingly.
Analysts believe the next year Dutch ASML is going to overtake Japanese competitors and to reach the first place by the amount of sold lithographic plants, including tools for manufacture of chips and exposition systems for LCD panels manufacture. Despite the present start back to the third place, in 2002 ASML is going to have 33% of the market by selling 204 lithographic plants. Such hopes have their base: the company expects to receive orders from Intel, IBM, AMD, TI, Infineon, Sony, Dongbu, and SMIC.
According to this script, Nikon rolls back to be #2 in the world ratings with 31% (or 190 plants) in 2002, Canon will receive only 29% of the market (180 plants), Ultratech - 6% if it sells 38 plants in 2002.
Certainly, such prospects do not please the present leader, Nikon. It doesn't like the situation up to such a degree that past January the Japanese company made a step, very popular in the business world... Right you are! It brought an action against ASML, accusing the Dutch competitor in infringement of five patents. Well, and you can name the claim without my participation. Certainly, the thing is about the interdiction of ASML production export to USA. This time ASML offers to join efforts in development of lithographic systems of the new generation. Well, on the background of forthcoming legal proceedings such offers look a little... optimistic.
The news of the month in this section is surely the enactment of the USA bill about the increase of the computing capacity limit of exported systems. The former limit of chip export with productivity up to 85,000 Mtops (millions theoretical operations per second) was raised up to 190.000 Mtops. By some opinions in press, president Bush had at last listened to Intel's complaint that has been demanding to provide unobstructed export of Itanium processors for a long time already.
Mainly, such decision pleased companies that dreamed to start deliveries of high-efficient systems to the fast developing China. However, not forgetting about war against terrorism, Bush has practically forbidden computer export to such countries, as Iraq, Iran, Sudan, Syria, Libya, Northern Korea, and Cuba.
It's unknown for how long this new restriction of delivered processors productivity will be active, however, according to some data, there is already a law in development, called to cancel such criterion, as Mtops productivity. Many experts believe that such parameter is not really objective nowadays.
In January Intel has announced some updates in its long-term technological roadmap. Now steps of the company for introduction of the modern lithography are placed as follows: 0.9 microns (90 nm) - by 2003, 65 nm - by 2005, 45 nm - by 2007. Due to processor clock rates, Intel has been recently specially stressing chip package, advancing its BBUL technology announced in October 2001. According to Intel's experts, BBUL package technology may theoretically support higher clock rates than present packages, and has an opportunity of situating several chips in one case without application of complex MCP technology (Multi-chip Package).
Intel's vision of "terahertz" transistor for future processors is combined, first of all, with the necessity of replacement of oxide gate layer material, that can lower leakage current by multiple of three in comparison with nitrated silicon oxide, and also with the transition to more advanced SOI technologies, such as FD (fully depleted)-SOI, for reduction of transistor working voltage.
Every month there are messages about new members of HyperTransport Consortium. LSI Logic has entered it in January. At the end of 2002 the company plans to present PHY I/O levels of HyperTransport interface for ASIC developers as a part of the 0.11 micron CoreWare Library that will contain DSP processor cores from MIPS, ARM and ZSP along with special cores for Ethernet, Fibre Channel, GigaBlaze, HyperPHY, USB, and IEEE 1394 FireWire.
Isonics's business has seriously improved from the beginning of 2002. It has sounded due to the planned bargain with AMD about delivery of Isonics's plates from isotopic pure silicon-28. In the beginning of the month they've found the manufacturer of SOI-plates under Isonics trademark - Silicon Quest company. For the access to Isonics's intellectual property partly received by Silicon Quest after the purchase of bankrupt Silicon Evolution, SQI will offer its industrial and research capacities.
At the end of the month there were messages that Isonics has received the first consignment of silicon-28, silicon-29, and silicon-30 isotopes from the supplier in Zelenogorsk, Russia by means of the Department of Energy of USA. The second consignment of isotopes from Russia is expected closer to the mid-year. In some days Isonics has literally announced about the beginning of trial thick-film SOI-plates deliveries in February.
It was interesting enough to hear in January about the continuation of development of so-called fuel batteries. Prototypes of such batteries, that transform energy of hydrogen-oxygen reaction, have been already developed by many Japanese companies. However, they are still not commercially introduced to manufacture. Toshiba has informed about positive results of methanol batteries development, and has even shown PDA working with such batteries. Sony has announced the progress in development of molecular carbon fuel batteries. NEC, conducting development together with budget research laboratories, has announced shifts in development of fuel batteries with nanotechnology application, also working on methanol.
Past month has brought at least one more message, showing the closeness of launching such batteries manufacture. Samsung (rather Samsung Advanced Institute of Technology, SAIT) and Sanyo have announced the creation of a tandem for the joint development and manufacture of fuel batteries. Companies will work on technology of creation batteries on the hydrogen, received from natural gas, methanol, and certain fractions of gasoline. Actually, this time it will not be limited only to manufacture of tiny accumulators for cellular telephones. Two companies plan to create small-sized multikilowatt sources of uninterrupted and emergency supply. Who knows, maybe in some years, on hearing his UPS squeaking, a user will just have to pour in a half-litre of something forty-volumed...
Intel was as usually the largest industrial news generator in January. In the beginning of the month, announcing activity results for 2001, the company has announced the reduction of capital construction expenses this year down to $5.5 billion that is, certainly, much less than last year's record $7.3 billion. At the same time research and new technologies expenses will be increased from $3.8 billion to $4.1 billion.
In January there were messages about planned work recommencement at 300 mm Fab 24. It's known that the construction of $2.2 billion factory in Leixlip, Ireland, was suspended last year. As a matter of fact, the exact date of construction recommencement is still unknown, everything depends on the further development of the economic situation. Nevertheless, Intel is going to start pilot manufacture at Fab 24 in the second half of 2003, and the launch of mass chip production is meanwhile planned for the first half of 2004. Fab 24 is expected to become the first Intel's enterprise focused on manufacture of 90 nm (0.09 micron) chips from the very beginning.
The business of other company's factories was more vigorous. Intel has already started the mass production of processors at the first 300 mm 0.13 micron D1C factory in Oregon, USA. The equipment of 300 mm Fab11x in Rio Rancho, New Mexico, USA will begin in the near future. The factory will launch pilot manufacture of 0.13-micron processors in the second half of 2002.
Among other Intel's plans is the equipment of 300 mm factory in Chandler, Arizona, that now works only with 200 mm plates; the construction of D1D factory to become a place for development and repetition work of 0.07-micron technical process on 300 mm plates. D1D is under construction near D1C in Hillsboro, Oregon. There's least data about this factory, and exact date of manufacture launch is unknown yet.
In January Intel has also announced additional $80-100 million investments to be spent for modernization of assembly-test factories on Philippines. The purpose of modernization is the transition of industrial lines from densely populated suburbs of Manila to areas more suitable for manufacture; equipment consolidation; and also the installation of the newest equipment for assembly and testing of Pentium 4 processors.
Unfortunately, Motorola's business was not very bright from the beginning of the year. Continuing the policy of expense reduction, Motorola has closed another factory for packing and testing microchips in Hong Kong with manufacture transited to China and Malaysia. After they've published company's payments after closing three more factories: packing in Texas, USA, and in Sendai, Japan, and also parts of factory for chip manufacture in the same Sendai. The total reduced staff of the company since August 2000 has made 39 thousand people worldwide.
Despite closure of factories and unprofitable figures for 2001, Motorola does not lose optimism. Actually, expenses for semiconductor manufacture in 2002 are reduced by 67%, down to the ridiculous sum for such company - $200 million that looks unpleasant near $610 million expenses of the last year and absolutely indecent in comparison with $2.4 billion, spent in 2000.
Nevertheless, Motorola's semiconductor branch hopes to become profitable in the fourth quarter 2002. How? Due to expenses reduction, closure of factories, staff reduction, conducting a campaign for intellectual property licensing, and also to investments into the most prospective manufactures: specialized SiGe:C (silicon-germanium technology with carbon application) and BiCMOS radio-frequency technological processes. Alas, to save production volumes the company, most likely, will have to order chips from foreign partners.
Nevertheless, everything's not so bad. By December 2002 Motorola is going to transfer one of its factories to new 0.10-micron "copper" silicon plates production.
Taiwanese SiS, being successful enough, does not hurry with the construction of 300 mm factory, denying all corresponding rumors. According to company representatives, the start of financing of this construction depends on further economic situation as well as on successes in fund raising. Presumably, new SiS's factory will not only release chipsets, but also systems-on-chip (SoC) and logics for IAs.
Taiwanese Nanya, also being successful enough during last months, has already decided to spend up to $2 billion for construction of 300 mm factory. It will begin in the March 2002, complete equipment will be finished closer by the end of the year, full capacity manufacture with use of 0.11 micron technical process will be launched right at the beginning of 2003. It will be already the third company's factory (other two produce 200 mm plate). Despite other two Taiwanese memory product manufacturers slightly overtake Nanya in launching 300 mm plates manufacture (ProMOS, Infineon's partner, already finishes factory construction and plans to produce up to 5000 plates per month this spring; Powerchip will also start 300 mm plates production this year), Nanya has all chances to make up the time lost. Now the company has serious odds - quite good profits from DDR memory sales, as Nanya was one of the first to begin making this memory, and this company was the first on Taiwan to set up repetition work of DDR333 chip version.
Continuing Taiwanese theme, I'd like to mention successes of TSMC, that has shown the first chip made with the most progressive today 0.10 micron technological process. TSMC plans to launch pilot operation of 0.10 micron line in the fourth quarter of 2002, much earlier than other industry leaders, including IBM, Intel, etc.
As far as it's known, TSMC also develops two 0.10-micron technical processes at once - for 200 mm and 300 mm plates. Not the most economic step, certainly, but it allows the company to secure itself against possible problems with not completely tested 300 mm equipment. By the way, for 0.10 micron manufacture TSMC the will use the newest 193 nm lithographic equipment from ASML - Twinscan 1100 AT (that's mysterious "large Asian customer", mentioned by ASML after it had announced the deliveries of two-flow scanners).
It is the high time to proceed to AMD news. The company surprised us in January not with financial results of 2001 and not with the promise to launch the mass production with application of 0.13-micron technical process in the first quarter. Right at the beginning of the month the first interesting announcement was that in 2002 AMD will increase capital construction expenses up to $850 million, that is 20% more in comparison with past year expenses. On the background of Intel's announcement about the reduction of expenses to $5.5 billion, such proportion change, probably, looks attractive. However, we shouldn't forget about $7.3 billion, spent by Intel in 2001, and that scales of companies are still incompatible. Analysts believe that AMD is simply compelled to increase construction and R&D expenses due to present technological lag even for the sake of the release of its first 64-bit Hammer processor before the year-end.
However, the end of the month has brought us absolutely other news from AMD. As it has became known, AMD has announced the creation of AU Pte Ltd joint venture together with Taiwan UMC. The enterprise will engage in the construction of a new factory for production 300 mm plates on the territory of Singapore that will be initially designed for 0.065-micron technical process. We shall see the first chips, manufactured at the still unnamed factory, presumably in 2005.
But companies' cooperation is not limited by this: this year 0.13 micron AMD's processors of UMC's manufacture will appear for sale. It's already surely known that Hammer processors will not be produced at UMC enterprises, however, even before the launch of the Singapore factory, AMD will get an opportunity of accessing its Asian partner's 300 mm capacities: the teamwork on launching and debugging 0.09 micron technical process will be conducted there. As it has became known, deliveries of the installation equipment for UMC (Fab 8F and Fab 8D factories) as well as deliveries of positioning systems for AMD's Fab 30 will be made by the same company - Applied Materials. At a close technological cooperation of AMD and UMC, and also due to same equipment suppliers, it will not take further additional work to debug similar pipelines at the both companies' factories.
As it has been planned before, AMD is going to launch 0.09-micron processors manufacture at Dresden Fab 30 in 2003. The factory will be using its full industrial potential closer by 2004, when it will be able to produce about 50 million processors.
Write a comment below. No registration needed!