Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
The Almighty Buck

Toshiba Latest Casualty of DRAM Price Wars 117

Tsar writes: "ITWorld.com tells the story: Toshiba is getting out of the DRAM business. They had 6.2% of the world market last year, but soon their Manassas, VA facilities will belong to Micron, the Yokkaichi plant's DRAM production will be reduced to a trickle, and Toshiba will be out of the commodity memory market. Guess you can sell DRAM for a hundred bucks a gigabyte, but you can't make a living at it yet."
This discussion has been archived. No new comments can be posted.

Toshiba Latest Casualty of DRAM Price Wars

Comments Filter:
  • Since hitting a high of around US$9 per 64M-bit chip in 2000, the price has fallen to below $1 and newer, more advanced chips are also under price pressure

    Prepare for flood of "Why in my day, RAM cost ... I can still remember my first 8086.... blah blah" posts...
    • by gatesh8r ( 182908 ) on Wednesday December 19, 2001 @05:56AM (#2725127)
      We had to pay TEN THOUSAND DOLLARS for one MESELY 32k of RAM! We had IRON CORES to actually use. And we didn't have these flimsy PCBs to work with we were lucky to have VACCUM TUBES!

      Why I remember when we used punch cards too and we programmed on the BARE METAL! Why we were lucky to have an ASSEMBLER or a punch card!

      Why you kids NEVER had it rough!

      • Try "informative". He's not exaggerating; this is within living memory.

        One of the great breakthroughs was when memory prices dropped below "a buck a byte"--for the monthly rental . . .


        • Well, it was meant to be funny in a way, but you're absolutely right. This cheap RAM is the only way we could possibly have the speedy CPUs we do today -- instead of having to worry about weither to do a CISC or RISC archatecutre to save four bytes or so and allow the OS and apps to fit within say, 640k RAM, we can now focus upon heavy optimizations of the CPU (and now the CISC and RISC ideas are a matter of preference; RISC though is cleaner in design the majority of the time). If anything, the ability to make cheap RAM was a "killer app" of its time, and we must not forget that.

          Now off to my studies for my final in CPU design...

      • Well, in my PDP-8 days 4K 12-bit words (6 KB or, more realistically, 8K 6-bit characters) cost a bit over $4,000. That was in 1969 dollars. What's that, about $12K today?
        • <Dusts you off> well well... honestly, being a college student, you hear about how the computers sold for hundreds of thousands of dollars for some of the mainframes (though I certain can believe that on a PDP-8) it's so hard to comprehend the costs these days! It's such a wonder what you could fit into those small areas...
    • Commodity RAM...Flibbildy Floo! In my day we had to put together our own memory cards! We didn't have any of those fancy 30 pin SIMM cards with their flashy gold or tin pins! We had to pay some guy working in his parent's basement to etch a board that we new wasn't going to fit in any of the S-100 slot of the backplane then we had to drill our own holes and solder our own chips. Those little bastard chips cost us $25.00 each. Then when we'd solder them in the iron would slip and we burn ourselves and we'd say "Oh look at me! I've got third degree burns from all this sloppy soldering!" and thats the way it was AND WE LIKED IT!
  • RAM prices are at obscene low levels right now (not that I mind)... will it stay this way or is there really a manufacturing glut right now? Another question is why would manufacturers price their memory to lose money and go out of business? Yeah, sure, cheaper is cheaper, but couldn't the companies that charge higher prices tout some feature? Like, higher reliability or better failure rate or some shit? Mostly PR, but can boost your sales.

    Another question is now that a semi-major chip supplier is going out of business, will prices jack up overnight?

    Ok, enough deepish thinking it.
    • I know, it's hard to click on that link, we're all guilty of it sometimes.
      They're selling the plant, intact, to Micron who will continue to produce chips out of the plant, they'll just have a different name stamped on them.
      The supply isn't going anywhere, and I doubt seriously that this sale will have any noticible impact on RAM sales.
    • by Anonymous Coward
      Another question is why would manufacturers price their
      memory to lose money and go out of business?

      The memory market is very homogenous - i.e. there's not much to differentiate between memory from one company and another. This means that the price is wholly dictated by the market. For example, say you own a chip factory. Lets say the market price of RAM has fallen below what it costs you to manufacture. What are you going to do? Your suggestion is to refuse to sell below your cost price. But now, nobody will buy your chips! They can get exactly the same elsewhere for less money. If you don't sell them, they will eventually become obsolete in which case you lose all your money, so you might as well sell 'em for whatever you can in order to minimize your losses.

      At the same time you would cut production hoping that this will increase the market price.

      The alternative is to enter into a price fixing cartel with other manufacturers. If everyone agrees not to sell below some price, you can hold the market price artificially high. This is illegal in many countries, and is also a fairly unstable situation. (Eventually someone decides they can make more money outside of the cartel so it breaks down.)

      (Take all this with a large pinch of salt. I am not an economist, although I did economics 101 at Uni).

  • It looks like companies really are finding out the hard way, that just because a market is really expensive to enter
    (A 'chip' factory is definitely one of the most expensive things you can build) that hasn't prevented savage competition when it was profitable.
    Then the market saturates, and people stop buying. Prices crash, and that really expensive factory you bought is a white elephant.

    Course, that doesn't stop me feeling all enthusiastic about 4Gb of RAM in a desktop :)
  • by Andorion ( 526481 ) on Wednesday December 19, 2001 @05:33AM (#2725096)
    Toshiba didn't have much going on in the ram market for a while now...

    Check out the toshiba pieces up on pricewatch [pricewatch.com] to see what I mean.

    Wonder when we're going to see the DRAM market bottom out.... soon as enough people drop out of it I guess.

  • by Anonymous Coward
    Well.. its not like the fabs are closing. They are just being sold off. Big deal. And if anyone knew what was good for them, they would buy a shitload of ram while its cheap. The price WILL go back up. It always does. Its kinda like gold, but only more useful.
    • Always a dangerous option. I offered exactly the same advice to friends when the price of ram had dropped to an all time low. Traditionally, it rockets very shortly afterwards.

      Unfortunately, it just *continued* falling to the price it is now. The trick is, as with the stock market, knowing just how low it'll go before it rockets.

      I almost fell into the same trap after observing Nasdaq:TMTA stock - for a few weeks it kept alternating between $25 and $35 or so (I can't be bothered to check the charts) and I was so close to buying when it hit a low. Good job I didn't, as it continued to plummet to its present level.
      • What would make you think RAM prices would come back up again? RAM prices have been falling continuously since the beginnings of Moore's Law decades ago. Next thing you know, you'll be telling us to buy pentium IIIs before "the price rockets back up again."

        Not that there aren't temporary fluctuations in the price of RAM, but the overall trend is down, down, down.
        • What would make you think RAM prices would come back up again? RAM prices have been falling continuously since the beginnings of Moore's Law decades ago
          As a general trend, true. And also note that it wasn't me who suggested buying lots of ram this time, it was the parent post.

          However, what made me suggest buying ram before the price rose (the first time around) was that in the not too distant past the price of ram had suddenly *doubled* and this lasted for several months before the price dropped to its previous low (and then continued dropping, of course) The cycle was acknowledged to be caused by a whole bunch of companies getting in on ram manufacturing, saturating the market (causing the low price) and then going bust because it is no longer a viable product, causing a demand, and thus pushing prices high. Repeat ad nauseum.

          Although I agree that in the long term the price of ram will continue to drop, in the shortish term it is possible to wait too long to purchase an upgrade, and then the price shoots up when you actually need to buy it.
        • Hmmm...we had a temporary, but fairly significant, rise in memory prices what, about two years ago? Whenever I was building my personal SMP webserver, anyway. I remember being annoyed that I could've saved a few dozen dollars if I'd bought the memory a few months earlier.
    • RAM may be more usefull, but my big ole block of gold will not be made worthless by DDR GOLD, GOLD 2700, or the ever overpriced RD-GOLD.

      The problem with RAM price flux is that the price always does go back up, but often it is the result of a new RAM product entering the market (pc100, pc133 ect) rather than the old stuff shooting up in price (although this does happen some). This makes it very hard to arbitrage (buy low, sell high) the ram market, because you buy the old stuff low, and when the market jumps it is for the new stuff (meaning you get stuck holding the old RAM).
  • DRAM Schme-RAM (Score:5, Interesting)

    by CaptainAlbert ( 162776 ) on Wednesday December 19, 2001 @06:01AM (#2725135) Homepage
    DRAM is a funny product. Market analysis is harder because the demand for it is not governed by your "ordinary" factors. For a start, it doesn't wear out half as fast as you might expect (particularly if you compare its failure rate to the average lifetime of a home PC). Demand in recent years has been similar to that for mobile handsets - the technology matures and people want it... until they *have* it, then they don't need any more.

    If you imagine that every household didn't have an electric jug kettle, and someone suddenly invented it and sold it at an affordable price... well you get the idea.

    DRAM is in over-supply because it's one of the only silicon products which has a huge domestic market. My company makes telecoms base-station hardware, and we have next to no DRAM on our boards - it's all SRAM and fast DPRAM embedded in ASICs. The demand for *those* is easier to analyse because the market is steadier and less "faddy".

    Shame they couldn't put the plants they're closing to better use though... it's not like there aren't more exciting designs around the corner waiting to be spun...
    • Re:DRAM Schme-RAM (Score:4, Interesting)

      by seanadams.com ( 463190 ) on Wednesday December 19, 2001 @06:34AM (#2725170) Homepage
      we have next to no DRAM on our boards - it's all SRAM and fast DPRAM

      I'm very curious - do you have any idea what you're paying for SRAM? My company [slimdevices.com] is paying $3.00 a pop for 128KB standard 60ns SRAMs, and it seems incredibly high in light of the plummeting DRAM prices. In fact SRAM prices don't seem to have budged in the last year.
      • >do you have any idea what you're paying for SRAM?

        Personally I don't, although I'm sure I could find out if I tried. As you can imagine, we don't have much choice in the matter; the data rates we deal with are pretty high and the arbitration is already a nightmare, without throwing dynamic refresh into the equation. Hell, sometimes our DSPs' internal data RAM isn't fast enough for us! :)

        Also, the quantities we use are pretty low (maybe 32MBytes per board, if that). And our teams are split between the UK and Japan, so I don't know where the purchasing actually happens.
      • SRAM costs more because it uses a multi-transistor architecture where DRAM requires only one transistor and a capacitor (you can build a capacitor on silicon with much less real-estate/layers). That is why they cost more. If it was as cheap to build, nobody would use DRAM because SRAM is way faster. But it isn't.

        As to you thinking that DRAMs are unprofitable and SRAMs are profitable... that is bogus as well. The price we (Yup, work at a memory company) get for these is in the toilet as well. Part of the big problem is all the communications companies (like you claim to work for) had huge amounts of inventory and so they quit buying (just look at the writedowns of Cisco, Nortel, etc. if you don't believe that bit). Of course, 128kb srams prices probably haven't budged because nobody makes such small devices anymore. Go price 2GB SCSI drives vs. 16GB for a similar scenario...

        DDR demand will take off though, so get your memory cheap while you can!
        • SRAM costs more because it uses a multi-transistor architecture where DRAM requires only one transistor and a capacitor (you can build a capacitor on silicon with much less real-estate/layers).

          Thank you, I know what SRAM is.

          As to you thinking that DRAMs are unprofitable and SRAMs are profitable... that is bogus as well.

          Did I say that? In case you haven't noticed, the whole semiconductor industry has slowed down. It's easy to see what's going on with DRAM, because it's sold direct to consumers. However, as CaptainAlbert said, the prices of other ther semis are much less hysterical. When you're buying this stuff in small (less than 10K pcs) quantities, you don't have much leverage in getting good pricing as the market moves.

          Of course, 128kb srams prices probably haven't budged because nobody makes such small devices anymore.

          Where are you getting your information??? Also do you mean kb or kB?

          Have you counted the number of devices in your home that have 4 or 8-bit cpus? I'll bet for every PC in your home, you have twenty or thirty smaller CPUs, embedded into everything from your mouse to your walkman to your coffee maker.

          Another thing you might be interested in knowing about SRAM is that it has other advantages besides speed. The interface is *much* simpler, and a block of SRAM can be easily integrated into any microcontoller or ASIC, because it's all CMOS. So when you're talking about *mega* bytes, you only use SRAM if you need the speed (and deterministic, immediate access - there's no refresh). If you need 128KB or less, you're probably not going to use DRAM because it would have to be on a separate chip, and you'd also need a controller (another 1K gates) to attach it to your CPU.
    • Re:DRAM Schme-RAM (Score:2, Insightful)

      by vanillicat ( 258354 )
      I think the plants are being sold, not closed...
    • Re:DRAM Schme-RAM (Score:3, Insightful)

      by adubey ( 82183 )

      I'm having a hard time understanding what you've written! You're saying that the DRAM market is hard to understand because once you have it, you don't want it anymore? But isn't this true of most goods? Also, the "wearing out" bit doesn't make any sense: I have a Maytag washing machine that hasn't worn out since the days when DECs had 64K of memory. This doesn't make the market for washing machines any harder to understand.

      How does having a huge domestic market imply you'll be in constant over-supply?

      Rather than some pseudo-economic reasons as to why the DRAM market is bottoming out, how about some real ones. First it is a commodity, so there is no chance of making an "economic profit" on DRAM. Second, the main market for DRAM, personal computers, it itself declining, causing a decline in the DRAM market. Third, better technology has made DRAM cheaper and cheaper. While in the early days of computing, the declining cost of RAM was matched by increasing volumes (ie every computer coming with more RAM). Recently, however, the cost to produce a megabyte has been decreasing faster than the avg. megabyte/computer has been increasing.
      • Re:DRAM Schme-RAM (Score:4, Interesting)

        by CaptainAlbert ( 162776 ) on Wednesday December 19, 2001 @07:32AM (#2725258) Homepage
        > I'm having a hard time understanding what you've written!

        <re-reads own post> hmm... so am I :-) Lack of coffee, I think.

        I don't think I meant it was "hard to understand", I think I meant it was more fickle and quite unlike a lot of other silicon markets. Most chips and circuit boards are never sold directly to the general public; they're designed, manufactured and purchased by the various companies who manufacture hardware for end-users. Thus, applying the same analysis that you might use when considering, say, high speed digital-to-analogue converters, might yield some phoney results.

        The PC (and related components) market is different from the washing machine market, because people expect the product they buy to be obsolete within a year, and worthless within three. That's not because it will break down, but because of the speed at which the technological developments are progressing. I guess that was true when the washing machine was first invented, too.

        All I was saying was that for some reason, a lot of investors and analysts have assumed without any solid justification that the growth curve in sales of DRAM/mobile phones/PCs/etc. can be extrapolated skywards without any consideration of external factors.

        Because the market trends are not quite the same as that for any other products, I think there were lots of mistakes ready to be made. If only 64KB *had* been enough for everyone :-)
        • >Most chips and circuit boards are never sold
          >directly to the general public; they're designed,
          >manufactured and purchased by the various
          >companies who manufacture hardware for end-users.
          >Thus, applying the same analysis that you might
          >use when considering, say, high speed
          >digital-to-analogue converters, might yield some
          >phoney results.

          yeah. bought three of those last week, instead of my usual two, do to the great sale . . .



    • You compare the demand for DRAM with that for cellphones. But software grows with its functionality and demands more RAM, unlike humans who after all make just phone calls.
      As long as software companies bring out new programs that grab it as if it's free users will have to stock it up. This, of course, does not apply to users who stick with their software - but these do not need other new components either.
  • memory prices (Score:1, Interesting)

    by Anonymous Coward
    Sorry about the AC bit I am a long time reader but don't comment much...

    I used to work for one of the biggest wafer companies in the world (MEMC) I got laid off in March when everything went to hell. I'm suprised it took this long for the chip companies to start feeling the pressure. We kept hearing it was a temporary setback and they has to clear out some old inventory. Well since our company is just about cleared out of 'old inventoy' I'm not suprised the upper steps of the ladder aren't feeling some pain,
  • by nyquist_theorem ( 262542 ) <.moc.liamg. .ta. .mehgellebm.> on Wednesday December 19, 2001 @06:44AM (#2725183) Homepage
    According to this story [thestar.com] Micron's not exactly picking the money trees either.

    One interesting tidbit from that article to give you an idea just how much the DRAM market has turned - Micron recorded a record $625-million loss for its entire 2001 fiscal year, which came after a record $1.5 billion profit in fiscal 2000.

    So I wonder if it's not that Toshiba wasn't doing ok (relatively speaking) but rather that they didn't have the pockets (or desire) to hang in there to watch it turn back (as one would expect it to in time).
    • I really wanted to buy Micron DDR 512MB chips for the computer I built two months ago, because Micron seems to be among the very few companies that are really Good Guys. The problem is that the chips weren't available yet, so I ended up with some chips from a unspecified manufacturer. They were really cheap and they work ok, but I would be happy to pay more for Micron chips if I could get them.

      So, Micron really needs to get their products out to the market faster. I hope they've got it out before I build a new computer for my parents.

      • Uhh, Crucial PC2100 512MB DDR modules have been out for a few weeks now.

        Then again, theres next to no reason to buy 512MB modules for your parent's computer.

        I have 4x256MB Crucial modules in my dual AthlonMP box. What could you possibly need more than 768-1024MB of ram in a box for your parents?

        Nevermind that 512MB modules are bloody expensive compared to two 256MB modules.
        • I've got only two slots on my ASUS mobo, so I didn't want to fill them both, it would make a future upgrade more expensive since I would throw away one chip. Also, when I build this computer, I'm going to have the Micron chip for myself, and let them have the chip I now have... :-)
    • Compared to a lot of technology companies, going from a 1.5 billion profit to a .625 billion loss isn't bad. There was officially a recession in 2001 after all, in addition to a lot of market volitility. Micron also has an edge over other manufacturers. They have a crucial.com, a popular place to aquire good memory. Since they're selling directly that cuts out middle-men meaning a higher margin of profit. Also, the best time to aquire competitors is during a recession.
      Keep in mind that building a new fab can cost Billions, acquiring one at a firesale could give micron a huge edge over competitors. Considering that Windows keeps driving up RAM requirements, there should eventually be a stable point in the market again sometime soon.
      All those people buying XP boxes with only 128MB of RAM are going to find out how painful XP is with less than 320 megs. Still, it is only a small chance that enough people will upgrade to make a dent in the oversupply of ram.
      Then again there could be an earthquake tomorrow and RAM prices could triple overnight if just one multi-billion dollar plant was destroyed.
  • I tried to get a job at that plant....they hardly hire any entry-level techs and stuff.

    Yeah, I live near it...only about 5 or 6 miles...10 miles to the new AOL building....my god..Manassas is turning into Herndon/Stearling/Reston, VA

    But, about RAM prices....haven't they been up lately because of christmas? I don't follow this as much as some people, I'm not that much of a nerd.
  • Simple Economics (Score:4, Interesting)

    by thebigwaffle ( 323430 ) on Wednesday December 19, 2001 @07:14AM (#2725238)
    "Toshiba is not alone in wanting to exit the **commodity** DRAM market. Earlier this year NEC Corp. said it would end DRAM production by 2004 and transfer all operations to Elpida Memory Inc., a joint venture formed with Hitachi Ltd. to insulate the two companies from the cyclical swings of the market."
    //emphasis mine

    Well, there you have it; a perfect example of classical economics. Whenever you have a particular market where profits are above normal (excess earnings after you pay your overhead), competitors will enter the market place. As more competitors make more product, the supply of the product increases (shifts the supply curve right, for all you Econ buffs) and pushes prices down. That's all and dandy but it doesn't last forever. Because prices are pushed down (and profits back to normal or below), competitors/entrepreneurs/Toshiba will exit. It makes sense. Why stick around?

    Be careful folks. As this model suggests, as supply decreases, the price will rise. Exiting an industry preludes price increase; it's inevitable.
  • How much then (Score:2, Insightful)

    by cloudturtle ( 260857 )
    The submitter of the story makes the observation that you can sell RAM for 100$ a Gig, but you can't make a living doing so. I think this important because it really sets up what we as the consumers need to know, how much do you have to sell RAM for to make a living. From the looks of it - with Toshiba and Micron both losing money - the market is selling below cost. Does anyone know enough about the manufacturing process to offer an educated estimate as to how much these chips cost to create. We really need to know this because market consolidation will only go on for so long before the major players (Micron, Samsung) decide that losing money isn't cool, stop buying up floundering competitors, and start making money again. A little price war isn't so bad because it allows the market to patrol it's self. Eventually though, this will wear off, and we as consumers should be aware of where the market is going so as to not be completely floored when the price bounces off the roof.
    • I dunno how much you have to sell it for. But before everybody gets all het up about individual fab closures (too late) it'd be a good idea to keep in mind that it's the direction the whole industry is taking [semi.org].
    • The profitability of DRAM chips depends highly on the cost and availability of
      Low ESR Tantalum. [google.com] While not the only factor in profitability, this is quite often more than 75% of the cost of materials in a DRAM chip. Any shift in the price of low ESR tantalum will shift the cost of building each DRAM chip.
      the easiest way to 'save' money is a die shrink since it dramatically reduces the amount of materials used to build the same quantity of RAM.
      If you're using 64-mbit chips and the competitor can fit 256-mbit per chip, (because of a die shrink) they have the advantage in materials. The downside is the cost of developing such high densities and then deploying them.
      Also, I beleive that micron is loosing money mostly on the development of new density chips etc. The pricing on crucial.com seems to compensate adequately for manfacturing costs, other than research.
  • Also on the BBC (Score:2, Informative)

    BBC [bbc.co.uk] also have this. They say the reason is "DRAM memory chips have become commoditised in recent years, with massive oversupply as Asian manufacturers moved into the market after a drought in the mid-90s."

    I sure hope this doesn't leave to increases! Crucial had 128MB for 30 bucks last I checked!

    • There seem to have been some increases lately,

      A couple of months ago i was looking at upgrading and it was ~£28(ukp) ( $40) for a Crucial 256mb 2100 DDR sim, I checked again yesterday and it was now £52(ukp) ($75) thats almost a doubling in price! Its put me off upgrading until the new year.

      I'm not sure what the prices have done in the USA
    • 128mb for $30 is a ripoff in the current market. It was $6.00 (SIX dollars, you read that right) at Star Components last time I checked. Couple months ago I got a GIG of PC133 DIMMs from them for $72. A week later the price for the same sticks was down to $55 !! Mind you this is namebrand stuff, usually Micron but they also carry others.

      Star is a Los Angeles area dealer, very reliable and honest. I've been buying all my memory, CPUs, and hard disks from them for years. www.star-components.com

  • I live in Manassas, Virginia; it has a history:
    First, IBM sold out to some unknown company, then they sold out to Lockheed Martin, then they sold out to BAE. Now we have the same problem with Dominion Superconductor. And shoot, I wanted to get a job there.
    • I live there too. . . Lots of Tech pretensions, but the new Data Center off 66 and Sudley is standing empty, and most of the projects at Innovation@Prince William are delayed or pulling out. Yet the Dulles corridor is still buzzing, amidst the crash of the Dot-bombs (I know, I worked for one that recently died, just moved back to Government Contracting. . . ).

      Why don't companies stay in PW County ? I suspect, because we're a bit too far from fashionable locations on the Power Track (Crystal City-Rosslyn-Ballston, Downtown, and the Dulles Corridor): we're ok for manufacturing, but Manassas still has the rep of being on the howling edge of the wilderness, despite the real howling edge being somewhere around Winchester.

      Heck, you can even get several different broadband providers here (I use Speakeasy on the Covad network . . . ), which is as proof-positive of non-hick-town status as I can think of. . .

      • Lots of pretensions but, until recently, a hellish trip to the nearest highway. As far as the howling edge goes, shoot, there's even a place in town that does lan gaming.....how's that for tech savvy...:)
  • by standards ( 461431 ) on Wednesday December 19, 2001 @08:49AM (#2725358)
    "DRAM Manufactureres are losing money on every chip they sell".

    Well THAT'S a load of bullshit. Then why are they selling?

    "We'll sell you a million of them. That'll cost us a $1,000,000, but let's do it anyway."

    It's NOT true.

    Chip manufacturers MAKE money each time they sell a chip. And if they sold enough chips, they'd walk away with a tidy profit. And - SURPRISE - they do!

    The problem is that they've expended lots of money building new fabrication facilities, and then, whoops!, the PC economy takes a hit. It isn't that DRAM prices got more expensive - it's that they over-invested in fabrication. And let me tell you - the plants themselves aren't the most expensive part - it's the people and managers that run them.

    What's the answer? To sell off the facilities to those who won't be competitors. Heck, why sell something good to a competetor? Therefore, the entry into "vertical markets" by the likes of Toshiba, selling off the facilities to Micron. As long as there is no competition between partners, they're happy.

    So what does this mean to DRAM prices? They might fluctuate a little, but the trend will continue downwards as real manufacturing costs are lowered and the technology improves.
    • by Anonymous Coward
      The only reason why prices are so low and are kept so low is to flood the smaller players out of the market. I.E. Toshiba.

      Also prices will fluctuate a lot because once every thing is consolidated the flooding will stop and prices will rise. Rise until they start getting back on track and put 300mm fabs online. Then the cycle will just start all over. This is nothing new.
    • by Anonymous Coward
      They are actually loosing a LOT of money. But it is not easy to throw away all the (expensive) machines needed in a fab.
      If they sell chips at a bigger price than the competitors (some of whom were supported by the local government), of course they go out of business, and, in the long term, is better to loose money now than close all the expensive facilities they have and loose money forever. Obvious.
      And people (even managers) are a very little cost, compared to everything else is needed in a wafer FAB...
    • Some companies have very high fixed costs and very low variable costs (like the DRAM market). Sometimes when they enter a temporary period where they can only sell their product at a loss, they will continue to do so. Why? Imagine the capital costs of owning the factory are $100mm a year, but that you can produce up to 100mm chips at an incremental cost of $0 each. If you can sell the chips for $.50 each, you might as well do it (for a limited period of time) because losing $50mm is better than losing $100mm. Granted, the real world DRAM industry is not as purely fixed cost as this, but its a good example of why companies sometimes do sell products ar a loss.

      Still don't believe that companies will ever sell at a loss? Go look at the earings cycle for Micron over the past 15 years. Some years they lose money, some years they rake it in hand over fist.
    • Go back and take microeconomics. If, by chance, you took it and somehow passed, do your professor a a favor and leave his name out of it . . .

      Firms sell at a loss when they lose less by selling than by sitting idle. If my choices are to sit still and lose two billion from my idle plant that I'm stuck with, or to pay another half a billion to produce a billion and a half in revenue, I choose the latter, losing one billion instead of two.

      hawk, economist

    • I *was* going to post explaining why this guy is wrong. But then I noticed 6 responses below my read-threshhold. They all already said what I was going to say.

      So why am I posting? To gripe about the idiots who modded parent up to +4 interesting, when apparently most people can see it's wrong.

      Don't bother modding this up, or any of the other responses. MOD PARENT DOWN! Or metamod.

      Declining to use my +1 bonus. I'd post this at zero if I didn't have to go AC to do it.

    • Then why are they selling?

      Well, it's quite simple. Manufacturers have fixed costs (eg rent) and variable costs (raw materials).

      In the short term, if the price they sell their chips at is greater than their variable costs, they will continue production.

      Each chip they sell will cover all variable costs, and a part of their fixed costs. This is better than letting it all sit there and just having fixed costs and no variable costs.

      In the long term, since they are making a loss, they will exit the industry. This is exactly what they are doing here.

      ECON100 pays off. :)

  • DRAM Pricing (Score:2, Insightful)

    by PoiBoy ( 525770 )
    From what I gather, DRAM prices actually bottomed late in the summer and then have had a rather sharp run-up in the wholesale market over the past few months. In fact, MU has doubled from its post-9/11 low of around $16 to $32 now.

    In terms of market structure, prior to Toshiba's divesture I think 4 companies produced about 80% of the world DRAM supply. Is this level of concentration sufficient to maintain collusive behavior? It all depends.

    In one commonly used model of price wars, the memory chip market is in fact collusive. A key assumption is that firms cannot perfectly predict demand conditions ahead of time. Suppose that you and I are the only two producers, and we agree to the following arrangement. In period 1 we both will charge $1. If at the end of period 1 the realized price is $0.90 or greater, we charge $1 in period 2. If it is less than $0.90 cents, we begin a punishment phase in period 2 and charge our marginal cost (and earn zero profits). The realized price may differ from the ex ante declared price because of weak demand for the product or because of cheating. Finally, if the punishment phase has been in effect for 3 periods, we then revert to charging $1; in other words, the punishment lasts 3 periods.

    This may in fact characterize the DRAM market. Because of a fall-off in demand for memory late last year and into this year, prices plummeted. Hence, the four DRAM producers began a punishment-phase price war. In the past several months demand for PC's and DRAM has stabilized somewhat, and the firms (ex post) realize this. The price war is now over, and the firms are returning to a high-demand stage where the higher collusive price is charged.

    Perhaps this sounds implausible to you, but in fact it is one of the most common models of collusive behavior in the economics profession. Of course, I've left out a ton of details, but the main point is clear: firms in an industry cannot perfectly observe demand, and so falling prices (whether due to low demand or actual cheating) must result in punishment. Once the punishment phase is over and demand returns to its normal level, the higher cartel price can again be charged.

  • "...but soon their Manassas, VA facilities will belong to Micron..."

    I have quite a few friends who work there, and the japanese never really seemed all that great at managing Americans in a factory environment. Hopefully Micron can do better...
    • I don't know about Micron's other facilities, but I've heard from several sources that working at their Boise ID plant is like being in a prison camp.

      • ...japanese never really seemed all that great at managing Americans in a factory environment.

      Can you give me some examples? The Japanese seem to do a good job at managing Americans in all the Auto plants that have sprung up in the ten or twenty years. At least, I've heard good things from the workers at Honda and Toyota plants and the products that come from those plants seem to be of comparable quality to what is produced in Japan.

    • Dominion had one of the most liberal sick policies I have ever seen. Unlimited from day 1 with medical coverage!!! Its like Manassas's own welfare system. I've heard stories of at least two woman that got jobs being 7-8 months pregnant, went out on sick leave, never came back and collected a paycheck with full medical for months before finally being let go.
  • I find it fascinating that DRAM can be had for $100 per GB these days. I remember buying 1 MB SIMMs for $100 apiece 10 years ago.

    The really interesting part is that magnetic drive storage capabilities have risen and the costs decreased over the past 10 years, too.

    Otherwise, if we had 1990 vintage hard drives and costs, the DRAM would look real tempting to use as an artificial "hard drive", albeit with doodads to keep it powered up full-time.

    So, 10 years from now, when 1 TB DRAM sells for $100, will disk drive technology have kept pace?

  • I Work in a independent OEM computer store that sells custom boxes and the word on the street for a long time is this: Micron has been intentionally selling below cost to kill off its smaller competitors. The market had become to competitive and beyond the control of the larger industry leading companies. They no longer could control pricing and the profit margins were not to their liking, so Micron responded by starting a price war to force smaller manufacturers out of the business. Prices will go back up when they can contrlol the market again and make extremely high profits.

    The price war is just a power play to regain control of a market that was priced to high for the consumer but good for the bank book.
    • And they'd better be careful of games like that- too big of a price spike and nobody will really buy memory. If nobody really buys memory, they don't make the profits they thought of, etc. I suspect that memory will maybe go up half again or double itself in price, but not too much more than that. It'll push things out of control the other way for them.
    • You are woefully misinformed. Micron doesn't want to sell below cost, they have to do that to maintain market share. Micron will definitely put smaller competitors (like Toshiba or TI) out of the business if they can, but they aren't as nefarious as you make them out to be.

      The problem is there is too much capacity. Eliminating most of Toshiba's capacity will help, but not a great deal. There is simply no demand for new PCs (which is where most DRAM goes and you can check Gateway and Dell's performance if you doubt me) and without that driving the industry (as it did in 1999 & 2000) you're going to see over production. This results in higher inventory levels which represent tied up capital. So, either you sell the inventory at what the market is paying to free up operating cash or you lay off people and try to hope that your competitors will also slow down production and not steal your market share.

      Since everyone at Micron making over $60K a year just took a 10% pay cut and the executives aren't getting paid, I'd say that they aren't doing this just because they weren't getting the profit margins they wanted.

      The real problem with the DRAM market is that Hynix (who is also in negotiations with Micron) owes 7 Billion dollars on 2 Billion worth of assets refuses to go into bankruptcy. They continue to sell at even higher losses than Micron and Samsung racking more and more debt.
  • by istartedi ( 132515 ) on Wednesday December 19, 2001 @10:48AM (#2725756) Journal

    Guess you can sell DRAM for a hundred bucks a gigabyte, but you can't make a living at it yet."

    Give the DRAM away for free, and sell support for it.

  • I've bought a pile of EDO 32 megs when it was worth near nothing, for maxing out my older PCs, now I sell them for 3 or 4 times as much on ebay :) I've bought a load of 512MB ECC DDR when it was worth next to nothing a few months ago.., it's already doubled back :) hell, f** the .COMs :)

    Seriously, it's always the same curves every year for the memory, end of the summer, prices at their lowest, a month before xmas, "something happens" and the prices jumps back (xmas sales being the non-official reason).
  • DRAM has become a commodity, that's why pricing fluctuates like other commodities markets. Commodities follow the laws of supply and demand more so than any other types of products. Oil producing nations realized this and formed OPEC to regulate supply and in turn regulate price.

    The ONLY way DRAM manufactures will stabililze prices is to band together and form a cartel like OPEC. They must collaboratively develop DRAM technologies and adhere to production caps to regulate price. This sucks for consumers, but that's how energy companies regulate oil prices and make money.

  • As players jump off the memory bandwagon, competition will decrease. When that happens, surviving manufacturers will have no reason to keep the prices low. I suppose it'd be a good idea to buy up RAM now.
  • I just got a box of RDRAM from Dell today to upgrade a bunch of my clients servers. How funny is it that not only are they using Rambus (blech), but that they are buying it from a company that is getting out of the business.
  • by John Murdoch ( 102085 ) on Wednesday December 19, 2001 @01:45PM (#2726983) Homepage Journal


    A key reason behind Toshiba's decision to exit commodity DRAM production is a long slide in DRAM prices that began in the summer of 2000, which has pushed prices to below-cost levels leaving Toshiba, and almost all DRAM makers, losing money on every chip they sell. [Emphasis mine]

    In other words, there is a price war going on. And, if ITWorld's sources are correct, there is at least one producer who is making money manufacturing DRAM at these prices. Given that Micron is buying the DRAM plant (and running TV ads selling DRAM via Crucial), it seems clear that Micron a) thinks they can make money in the DRAM business, and b) thinks that they can use that plant in Virginia to make DRAM more competitively than Toshiba did.

    Two points:

    1. This suggests strongly that we won't see a DRAM price rise anytime soon. Several posts on this topic have pointed out that when you see market players exiting, supply tightens up and prices rise--but in this case supply isn't going to decrease, only the brand names will change.
    2. This is also extremely interesting to those of us who heard of the "vaunted Japanese electronics industry" throughout the '80s and '90s. The dominance of the DRAM market by Toshiba and NEC was so overwhelming that Peters and Waterman bent over backwards to chronicle the contrarian investments of an Idaho potato farmer named J.R. Simplot. Simplot, who more-or-less invented the frozen french fry, was watching TI and Motorola bail from the commodity DRAM market, and plowing his millions into a plucky upstart...by the name of Micron. 12 or 13 years after In Search of Excellence featured Micron in a clearly David-vs-Goliath-and-we're-not-betting-on-David tone, it would seem that an American company has beaten the Japanese electronics industry at their own game, driving them from the marketplace. That is an interesting story.

Kill Ugly Processor Architectures - Karl Lehenbauer