If you’ve been following the roller-coaster ride that constitutes the global semiconductor memory market, then you’re probably looking over the shoulders of as many memory analysts as possible, trying to find one whose crystal ball isn’t cracked, fused, melted, or blackened. Memory analyst Lane Mason recently spent some time at Denali Software and recorded a Webcast with his overview of recent events in the global memory arena, his view of today’s status, and his forecasts for what is to come. Lane’s been following the memory market for well more than a decade and his counsel is solid. See and listen to his report here.

With all the talk of DRAM growth in 2010, you might wonder what’s driving the consumption. It’s a very logical question. Sometimes, growth is simply driven by existing market momentum. Sometimes, it’s driven by a new product category such as Apple’s iPad, discussed earlier. And sometimes, there are disruptive changes in existing markets that also drive new waves of consumption. This blog entry is about one of those changes. What’s prompted this blog entry is Eurocom’s recent introduction of the D900F Panther Mobile Workstation, a laptop with some screaming hardware under the hood. The trigger for some big, mind-altering changes in memory consumption stems from the notebook’s use of Intel’s i7-980X Extreme Edition, a 6-core processor. The use of multicore processors isn’t new and is certainly driving the rise of DRAM demand in the server arena, but the installation of a 6-core engine in a notebook PC is like putting a 1001-horsepower, 16-cylinder Bugatti Veyron engine in a Ford Focus. The car’s going to go a lot faster. More important, it’s going to consume a lot more gasoline, a lot faster. The Veyron runs at a top speed of 270+ miles/hour, but can do so only for a few minutes before draining the fuel tank dry.

That’s precisely what happens when you insert a 6-core processor into a PC, server, or laptop. The thirst for memory and storage becomes fierce. Processors starve without data and code, served up in large amounts. Where does that data and code come from? From DRAM, NAND Flash memory, rotating storage, and fast network connections. So computers like Eurocom’s D900F Panther Mobile Workstation need a lot more memory, a lot more storage, and a lot faster network connections that other computers (mere mortals). Yet unquestionably, this year’s top-of-the-line machine always, always, always becomes the norm in a frighteningly short amount of time.

So if the Eurocom D900F Panther is the shape of things to come, it bears a look with respect to its built-in memory and storage capacities. The D900F can accommodate as much as 24 Gbytes of DDR3-1333 DRAM memory in a 3-channel DDR3 configuration using three 8-Gbyte DDR3 DIMMs. That’s a lot of DRAM and a sign that this year and next might truly be blockbuster years for DRAM sales. It’s also a sign of pressure on DRAM module vendors to start thinking seriously about 8-Gbyte DDR3 DIMMs. Next, the D900F accommodates as many as four SATA HDDs and SSDs in RAID 0/1/5/10 configurations. That’s as much as 2.6 Tbytes of on-board storage. Finally, the D900F sports two Gbit Ethernet ports.

The Eurocom D900F isn’t a notebook computer designed for everyone. In fact, it will suit very few this year. But that’s always the case with top-end PCs. They’re not for everyone. In the 1980s, well-configured PCs cost $5000 and that figure remained constant for a surprisingly long time even as technology drove processor, DRAM, and HDD pricing to commodity levels. Now we’re more accustomed to $500 PCs, but specific users including gamers and video content creators need more oomph from their machines. More important, these elite users are willing to spend more money to get more oomph. The disruptive power of new technology such as multicore processors upsets apple cart, making the $5000 PC once more the pinnacle. And that pinnacle will eventually be eroded as these new, desirable capabilities and memory/storage capacities undergo the inevitable price erosion. As prices fall, demand will increase driving up sales volume in a classic, neverending cycle of innovation followed by inevitable commoditization. This time, it’s all driven by the advent of the multicore processor with its insatiable appetite for data and code.

Earlier, we reported on the appearance of several good economic signs springing up for the DRAM market. Here are a few more:

Late last month, EDN’s Ann Steffora Mutschler reported on Semico’s 1Q 2010 forecast calling for greater than 70% growth in DRAM sales this year. That’s on the heels of Q3 and Q4 2009 DRAM growth rates reported at 33% and 44% respectively (although it’s after a huge drop in 2008). Semico concludes that the demand coupled with the lack of capital investment made by DRAM vendors over the last two years will ultimately lead to market shortages.

IC Insights’ March 8 Research Bulletin previewed the year for ICs and forecast a whopping 74% growth for DRAM sales (closely correlating with Semico’s prediction). If you find that number hard to believe, consider this quote from IC Insights’ bulletin: “As a quick crosscheck to this astounding growth-rate forecast, if the January 2010 DRAM market stayed flat for the next 11 months, the 2010/2009 DRAM market would register 61% growth! The strong DRAM market is expected to help support the 100%+ capital spending increases now budgeted by many of the DRAM manufacturers this year.” This bulletin is extracted from IC Insights’ comprehensive “The McLean Report”.

Two months ago, research firm iSuppli wasn’t painting quite as rosy a picture as its research competitors. On February 22, Information Week reported that iSuppli DRAM analyst Mike Howard was predicting a “greater than 40%” growth for the DRAM market in 2010. Perhaps the more bullish predictions from Semico and IC Insights reflects another two months of results from the bullish 1Q 2010 DRAM market.

Once again, that’s one of the reasons that this year’s MemCon is focusing almost exclusively on DRAM, while still covering the biggest developments in semiconductor memory. This year, MemCon is a 1-day event and will be held July 28, 2010 at the Hyatt Santa Clara. Registration is free and gives you access to exhibits, keynotes, technical sessions, and panels. Also, if you’re an industry expert and have something to say about memory, consider making a presentation (make a presentation proposal here). If you’re a DRAM analyst and would like a large, well-attended platform to strut your stuff, be sure to register for MemCon and contact the organizers for a speaking slot.

By Steve Leibson, Technology Evangelist, Denali Software

The recent, very successful launch of Apple’s iPad introduces both a new gadget and an entirely new product category to the consumer and business markets. Apple announced last Monday that it had sold more than 300,000 iPads on April 3, the first day that the iPad was available in the US. (That includes pre-orders, of course.) Note that this number tops the first-day sales of Apple’s iPhone. Analyst predictions are all over the map, projecting that Apple will sell somewhere between 4 and 7.1 million iPads in 2010 as the company starts to sell the product around the world. Each Apple iPad contains 16, 32, or 64 Gbytes of MLC NAND Flash depending on selected options. No wonder, therefore, that DRAMeXchange is predicting that the introduction of the iPad and copycat products from other vendors will have a stabilizing effect on NAND Flash pricing for the foreseeable future. There’s a built-in, assured demand for high-capacity NAND Flash chips in the middle term.

Last month, DRAMeXchange noted that mainstream MLC NAND Flash pricing had declined by a few percentage points over 1Q 2010 and that the market seemed to stabilize during the first half of March. Looking forward, DRAMeXchange expects MLC NAND Flash pricing to stabilize due to the demand by this new product category. In fact, iSuppli’s iPad sales estimate for next year is for nearly 15 million units sold and the company’s estimate for 2012 just crosses the 20 million/year threshold. Those numbers plus competing products in the new category suggest a rapidly growing and large market for high-capacity MLC NAND Flash devices.

Sources for the MLC NAND Flash used in Apple’s iPad have appeared in photos of the iPad teardowns conducted by the US FCC in February and by iFixit.com last weekend. The earlier FCC teardown concealed the vendor markings on the chips with gray rectangles but subsequent image processing of the FCC photos by Anandtech and iFixit.com using a little help from Photoshop revealed Toshiba NAND Flash ICs in the pre-production FCC teardown unit. The iFixit.com teardown on an as-sold, production iPad shows two of Samsung's K9LCG08U1M 64Gbit MLC NAND Flash memory chips soldered to the iPad’s processor board alongside the proprietary, Apple-designed A4 processor chip . iSuppli’s estimates in February placed a cost of between $29.50 and $118 on the 16 to 64 Gbytes of NAND Flash built into each iPad.

Author's Note and Errata: There were some errors in the forward NAND pricing in the version of this article as it originally went to the web on Thursday evening, July 16, which have now been corrected. We apologize to 'early readers', who may have been confused by some inconsistencies between the text and tables.

Future NAND price reductions will be much less than what we have experienced: Users of NAND Flash have become accustomed to annual price reductions of 50-60% over the past five years, built on strong GB demand growth averaging more than 150% per year. However, a conspiracy of events...financial, technical, economic and 'perceptual'...are expected to slow this rate of price decline by up to 90% for the next five years: from 100x price/GB improvement to 10x improvements...or even less. In addition, the industry is definitely facing a point where higher performance NAND will truly cost more per GB than lower performance NAND (endurance, RW speeds, or other valuable metrics), just as SLC NAND costs more than MLC today, but is used only in selected 'cost justifiable' applications. When x3 and x4 cells' share of the market grows, and SSDs gain further traction, a highly-fragmented market is almost a certainty, and the hi-low price spread for various flavors of 'flash' could be as large as 5x, solely on differential costs to produce various performance and endurance feature-sets.

In short, the NAND marketplace can be expected to take on a whole new character for the five years out to 2014, as technical progress aimed at cost reduction becomes more expensive and harder to come by, as 'performance degradation remediation' becomes more complex and 'controller-logic-and-software' intensive, as NAND vendors are forced by their managers and shareholding masters to make it a good business, and as new markets cannot develop and expand fast enough to sustain the former GB growth rates.

Most of the low-hanging (technical) fruit for NAND cost-reduction has been picked: In the past five years, lithographic shrinks were at first easy to achieve, as NAND caught and then surpassed DRAM as the 'memory process node leader'. The move from 'SLC to MLC' was low-hanging fruit waiting to happen; almost everyone had proven it out in the lab and fab years earlier, and it just took some 'seed crystals' of market competition to precipitate a wholesale transition in the market (exc. Samsung). Sustained high GB growth brought improved scale economies into play, too, as "SG&A and R&D per GB of NAND shipped" declined almost in proportion to GB growth. Also, much progress in price reduction has come at the expense of beating down manufacturers' profit margins to the point that 4Q08 was 'all red for all vendors and all products'. Indeed, 2008 ended on something of a hard stop to all this, with moves to x3 and x4 cells stalled for a variety of technical reasons, and with troubling performance hits, as well. Manufacturers' profit margins were at bare metal levels, or lower. The "One a Year" lithography shrinks, density-doublings and process node migrations are today not always affordable (tight cash and non-existent profits), but also technically more forbidding. Much new technology has to be investigated and mastered..."85-90nm to nom. 40nm" is far easier than "35-40nm to nom. 20-22nm" is expected for 2014 will be.

The look ahead, pricing progress: The modest NAND price recovery we have seen in 1Q and 2Q is just a time to rest and regroup, but the eventual resumption of price competition and reductions will be much more measured than formerly. No vendor can afford price reductions of ~50%/year any more; no one is capable of cost reductions of anything close to that. Importantly, many of the iconic personalities who bet on the sustainability of the fast growth investment and R&D strategies of the past period are moved on, in light of stubborn markets, declining profits and huge financial damage in the past year.

New NAND demand and markets for growth will also need 'cultivation'...and time: On the demand side, most NAND used today is relatively primitive, compared to its expected technical potential. A vast fraction of the GB used are lodged in applications with modest RW capabilities, tolerance for large innate silicon imperfections and failure rates (then corrected), and limited endurance.

This too, was low hanging fruit: camera chips riding the digital camera transition, USB drives to replace 'floppies', audio store for MP3 players...almost all replacing technologies that were well past their prime. These were also 'ready markets' requiring little market development by vendors, thus avoiding a time-consuming and irregular growth path. It will not be so easy going forward, and the huge price down strides made by NAND in the 2004-09 era are not likely to be matched. One has to look no further than SSDs, to get a feel for the difficulties and costs, the hit and miss, as well as the technical, standard-setting and price-performance roadmaps that have to be argued, developed, applied, rejected or modified for new and sophisticated markets to develop. It is a slow and painstaking development process. In addition, current and forecast prices always attract new applications; if one stifles the outlook, it may stifle the energy seeking new places to take advantage of "Sea Change Pricing" that has been a part of the NAND Marketing Call for many years.

Industry's perpetual efforts to cost reduce, and demand diversification, will drive broader product mix: What we see as a still rather narrow line-up of NAND 'performance' capabilities will become a sprawl, as elemental technologies are applied in novel and unique ways to serve diverse markets, having different needs and price points. More bits per cell always (so far, for all vendors and all products) means lower performance. "SLC to MLC" meant endurance dropped by about 10x-20x, from 100K cycles to 5-10K write cycles. For most applications, users would rather have the 'half-price' pricing than the 100K cycles. SanDisk was able to keep performance the same going from x2 to x3, but not without some considerable remediation: productivity went up only 20%, not the expected 50% in bits per mm sq., due to extra 'performance-management' circuitry. Fewer stored electrons per memory bit means more finely tuned sensors, and, almost invariably, more die area, reduced performance and more build cost. In addition, SanDisk seems, a year later, to be the only company with x3 in volume production, and that not even at their most advanced process node, which has deliberately lagged by one node their cutting edge SLC and 2MLC production nodes.

And though a move to x4 would give production cost benefits comparable to those gained when the industry moved lockstep from SLC to MLC in 2004-6, it is no slam dunk technology, by any means. Tight R&D budgets, resulting from huge operating losses for the past year, will not make anything easier. It is not likely, or even not possible, for all NAND vendors to have fully developed their "nom 22nm x4" technology in 2013/14. No chance.

Today, the line up looks as shown in Table 2, NAND Makers lithography, MLC status:

note: percentages refer to fraction of wafer starts at that process node

Bold guesses: NAND price outlooks: If NAND prices "today" (using 12/31/08 as 'today') are nom. $2/GB for MLC, then by the close of 2013, we expect to see a smattering of x4 NAND, with severely constrained performance, with best-NAND pricing of about 50-60 cents/GB. SLC and "SLC level performance", needed in some applications, is denied the benefits of the move to more bits per cell, since device performance is too compromised using today's design methodologies, so the $4.00/GB+ for SLC of today maybe declines to less than $1.25/GB by year end 2013, with progress coming largely from litho node moves, making some allowance for moving to sustainable profit margins and better cell constructs and continuing improvements in scale economies.

In fact what we have seen in much of the NAND flash market pricing is a thinly concealed compression of vendors' Gross Margins, coupled with a capacity of chip designers to 'engineer out costly features, performance', which today's NAND end-markets can easily do without...'defining down demand' attributes...and replacing what in retrospect can be viewed as "Cadillac NAND" with "Tinplate" or "Plastic" NAND, at far lower costs to build but fully adequate for most of today's needs. Full priced SLC was shipped in 2004, because there was no half-price MLC. Once MLC became available, it was found to be 'good enough' for 'almost all' applications. But it is not at all clear if x3 and x4 performance, without significant costly 'performance remediation' WILL be 'good enough for those same uses.

Consequences and impacts of price slowdown: This "Sea Change Pricing Model" moving to a new "Change of Heart Pricing Model" can be expected to cause major changes in the industry's view of end markets, vendor investment expectations, competition and potential market consolidation. Importantly, huge NAND price reductions in the past five years have enabled SSDs to barely make a dent in HDDs into PCs; with declines slowing, how does that change our thinking about the outlook for SSDs? This will be the subject of our next BLOG, next week.

No one can say with any certainty, but...

Recent improvements in DRAM and NAND pricing, a firming of demand, and a 'the bloom is off the rose' attitude toward the 'TMC DRAM Salvation Initiative', have all pointed to perhaps a better memory future, after a 2008 loss of $20B and another $7-8B in 1Q09 by memory makers. This will not, however, happen "...in the first hundred days...", so out of kilter the sector is today, and so deeply mired in below-cost selling we are: 'Everyone has issues', and none will be resolved today or tomorrow.

However, we are moving past some of the most troublesome features of today's memory market, though even on that point there is no overwhelmeing consensus..."always read the fine print"

Prices: Since their nadir in 4Q08, DRAM prices (DDR2, 1Gb benchmark) have risen about 35%, from 80 cents to about $1.15. That is good for DRAM makers, but hardly good enough to stanch the flow of red ink for anyone. Likewise, NAND flash makers, who seem to have recognized which side their profit bread is buttered on, are fewer in number, and more able to act in silent concert with a common interest, have both decommissioned older NAND fabs and idled a considerable amount of prime capacity in an attempt to constrict supply. It has worked: prices have come up about 25-35% since late November, and been stable for the past few months. Apple's surprise order for '100M 8Gb equivalents' added to the feast, though if that amount is accurate, would only constitute an incremental 4-5% on the existing annual NAND demand level.

Capacity overhang, sustainability of upward price pressure: Due in most part to the manner in which supplies were constricted, and a close-to-market balance was achieved, there is widespread suspicion that the price ups the industry has enjoyed are transient, and that the slightest glimmer of improved financials will bring this latent production capacity out of hiding and into production...tilting the supply up, and prices back down. Some analysts are already seeing this, even with the meager DRAM price improvements in the past 60 days. This possibility is not to be ignored, being as undisciplined as DRAM and NAND makers are wont to be, and as capable they have shown themselves to be, to shoot themselves in the foot, or worse.

DDR3 progress into market is steady (today): There IS good news on the DDR3 front, as adoptions are starting to be seen around the edges of the market. DDR3 has been a force in high-end game PCs/systems for more than 6-9 months and recently have been spotted in Dell's new laptops...not so much for their DDR3 performance as for their 'free' reduced power due to DDR3's 1.5V operating voltage compared to DDR2's 1.8V. At the crossover speeds, -800 and -1066, this is almost free power down, and a nice checkoff item for laptop sellers; it is too early yet to see if this is a true kick-off for DDR3, or a weak marketing initiative that will buy you little for Dell or the consumer.

After all, even a 20% improvement in memory power in a laptop is still only about 6-8% system power reduction when laid against the power used in the MPU, HDD and graphics subsystem; laptops have never used any kind of low power DRAM before, either.

DDR3 prices are also inching closer to those of DDR2, with some of the reduction due to DDR2's price rise, some due to new generation DDR3 being more die-area-efficient (less cost penalty to build, compared to DDR2), and some to sheer Strategic Pricing initiatives: if you are going to wrap a $2 bill around a 1Gb DRAM to sell it, it might as well be a DDR3 DRAM as a DDR2 DRAM; then you'll have a leg up on the future business into that particular socket. As always, the newer generation DDR3 DRAM is the leadership part which can give market separation between the leaders...Samsung, Elpida and Micron...and the DDR3 have-nots (mostly in Taiwan) who still are mostly dependent on their Korean or Japanese technlogy partners (the dependency problem that TMC had hoped to solve).

This also is a differentiator between the Micron-led Micron-Nanya-Inotera collective, and the situation Powerchip and ProMOS find themselves in with Elpida and Hynix...the transfer of DDR3 designs and most advanced processes, as well as jumpstarting the actual design effort within the Taiwanese companies, is less developed than that of Nanya/Micron.

All that being said, the "DDR3 value proposition" for PCs does not really gain traction until DDR3-1333 (which part is still scarce in laptops and PCs today), and price parity has been achieved. So the jury is still out as to whether the industry has truly turned the DDR3 corner.

Taiwan Memory Company: Demise and Resignation?: EE Times Mark LaPedus called for Taiwan DRAM makers to throw in the towel and abandon their plans to consolidate DRAM makers, or to create some "Grand DRAM Challenge" to Samsung. (Hynix has somehow slowly faded from the 'Korean DRAM Threat' discussion, once it's own financial problems were scrutinized more carefully). After watching developments in the press and in the market for more than six months, I would have to agree.

Of course, no one wants to walk away from their investment, though most of it is already vaporized in the 'great DRAM demise' of 2007-8-9'; overall, shareholders have probably lost 90% of their value already, and crediors have started to write down their loans, too, though some are playing hardball with their distressed-DRAM-making customers. Perhaps the surviving DRAM makers could accelerate the 'demise process', buy up the shares of the 'price spoilers' DRAM operations with their improved DRAM profits as DRAM prices rise, and then liquidate the assets. "Buy and dismantle" is clearly a non-competitive act (of survival), but is also not as nefarious as 'price fixing' or 'blackmail', and may be a legitimate business strategy.

CapEx, cash flow, dire straits: CapEx for memory is down by about 80% from 24 months ago, and is now 'critical investments only' everywhere we look...something pinch-point tools, and nothing to add more wafers. No DRAM maker is flush with cash on the eve of huge investment requirements to go to the 50nm node in volume production, though of course, some are better off than others. But ProMOS and Powerchip creditors are knocking on their doors for repayment of bonds. We may be near some resolution within a few months or even weeks, as DRAM makers are still caught between negative cash flow from operations, reluctant capital markets for fresh funding, aging process technology in need of an upgrade, and hungry creditors who want ALL their money back. Spansion passed though some milestone just this week, by finally being delisted from the NASDAQ, and saw its market cap drop to about $70M.

All the DRAM leaders have their 50nm DDR3 designs done and proven out, and also have SOME production capability on these advanced designs. But there are not enough production tools to make the industry-wide transition from nom. 70nm DDR2 to advanced process DDR3...until more time passes, and prices and profits improve. Unfortunately, as Top Tier DRAM analyst Nam Hyung Kim of iSuppli has said recently, there is still way more DRAM capacity than demand can support, and recent prices going up may not be sustainable, even if ALL of Taiwan DRAM makers disappeared! After two very strong DRAM GB shipment growth, 2009 will revert to slower rates seen many years ago: from 90 and 70%/year down to 30-40%/year are good estimates...though here, too, no one knows for sure: Industry estimates run from 2-3% over 2008 up to about 40%.

This one is definitely a "stay tuned" event, so rapidly things are changing. All memory makers like rising DRAM and NAND prices, but the close availability of additional production capacity is worrisome to all, and those who see 'demand recovery' somehow lack all conviction.

Now is the Winter of our Discontent: Five months into a memory industry recession of unknown depth and duration, we have seen as many different strategies in action as there are companies.

Some companies instantly anticipated the worst, and laid off substantial fractions of their workforce; others came to mostly the same conclusion after one or two months, but waited (considerately) until after the holidays to reduce headcount. Some companies, uncertain about how deep and how long this ‘business reversal’ would be, are trying NOT to cut staff until the picture becomes more clear. They are using forced time off, either with accrued vacation days or unpaid days, or extended holiday shutdowns to match production and output to sales, and selling from inventory at the same time, to rebalance the business. Some, those few with nice bank accounts, like Intel, are watching carefully but making no severe cutbacks past normal attrition and the human consequences of shuttering older fabs.

NAND and DRAM makers, in an unusual show of solidarity and common interest (this one allowed by law, unlike 2001’s price fixing.), have idled top-flight 300mm wafer capacity for extended periods, which, it is hoped, will act to reduce supply growth and halt or reverse the price erosion. But the meter is running and the equipment is still depreciating, so production costs will rise. NAND and DRAM makers have also taken down fully functional (but not cost competitive) 200mm fabs, and either sold them off wholesale or piecemeal, raising equipment inventory levels in the aftermarket…to the chagrin of Applied Materials, among others. Since “Cash is King”, too, equipment purchases, often involving tens of millions of dollars for a process tool suite, are being postponed, causing trouble for lithography leader ASML. Sales of semiconductor equipment are expected to be down close to 50% in 2009. Significant node migrations, famously the DRAM move to nom. 50nm designs that usually need expensive new lithographic process tools (“immersion lithography”), are being undertaken cautiously, with only enough tooling to prove out designs…and the Big Build-out being delayed until sunnier days, and more profit.

Significant, costly and diverse actions in response to the economic outlook have been taken already.

While everyone is short of cash flow, significant memory players are within a few months of totally running out of money: Qimonda and Spansion, whose 2008 sales totaled more than $6B, are now at their children’s change jars, plus all of the actions above, to make ends meet long enough to NOT have to sell the farm.

There is a lot of business pain; there are a lot of industry actions being taken to survive, and there are a lot of adjustments being made. Unlike the 2001 recession, however, things were not all that good immediately before the market fell down: DRAMs have been in pain since late 2007, and never really enjoyed a perdio of huge profits in the 2004-5- upturn; NAND was right behind, went weak in 1Q08, and never got back on its feet (as it did for 2Q and 3Q in 2007), it has been suffering greatly since mid-'08. End markets were tepid since early in 2008, were slowing as we rolled past the end of summer, then positivley tanked in October. For most companies, the business shift from 2008 into 2009 went from ”pretty tough” to “unbelievably tough.”

As always, different men can look at the same data, see different things and draw different conclusions. Thus we see the diverse actions taken. In general, significant NAND actions were delayed about six months from the first DRAM actions in early 2008...now everyone is in trouble, and we're not at all sure that the cutbacks and slowdowns are sufficient to match supplies up to demand, which is still weakening almost daily.

Another “Consensus" is building, too: This downturn will be very, very bad: There is now, however, an increasingly strong correlation between the date of all industry 2009 and 2010 forecasts, and the degree of severity they see: more recent forecasts are considerably more negative. These forecasts are also more supported by contemporaneous hard data on sale levels, new orders, still-increasing inventory levels, unrelenting price pressures and weak end markets. Recent DRAM and NAND price improvements may last for a while, but most observers are suspicious that, with capacity in place but NOT being used, production will not stay low enough...and, that there is way still too much inventory. Improved productivity is always driving costs down, and driving net die per wafer up; within the past week, Samsung, Hynix and Qimonda have all made significant technology or production announcements of ~45nm process-node DRAMs, which mean huge productivity improvements and higher output once ramped. In 2009, some vendors' output can grow 40-50% without adding a single wafer start. For the year, it does not look good, or even tolerable. Almost every day brings a darker view into focus, and that view is probably running ahead of actions the industry has taken so far.

The latest set of semiconductor industry forecasts was recently summarized in article in Electronics Weekly, precipitated by a significant reversal of viewpoint by Forward Horizons’ always well-informed and perpetual optimist, Malcolm Penn, who lowered his 2009 outlook to -28% compared to 2008’s sales levels; The WSTS/SIA has just now reported that their estimates for 2008 itself was down by 2.6% from 2007 (all the shortfall, incidentally, was in the last quarter). TSMC, the world's leading foundry, is also suffering mightily, and its CEO, Rick Tsai, added in his expectaton of -30% for the industry for 2009. The #2 leading foundry, UMC, is expected to be running only at 30% capacity in 1Q09. In many instances, foundry business has come to a complete halt, as demand for their major accounts' products...Xilinx, Altera, NVidia...is now supported by sales from excess inventories. The notion that “2H09 will see some or all markets recover” is now discredited; no one says that, or believes that, any more.

Take a deep breath. The chip industry employees more than 1.5M souls worldwide, with probably just as many ‘dependents’ in ancillary industries...equipment and materials, semiconductor packaging and test, subcontractors at all levels, and system assembly. No wonder the Taiwanese government wants to keep its DRAM makers afloat and competitive, as the island is host to tens of thousands of fab employees, and all the back-end processes of assembly and test, packaging and system assembly for PC makers, cards and all manner of electronic equipment, subcontractors for the world. Electronics are Taiwan’s #1 Export, by a wide margin, as well.

To wit, if the industry goes down anywhere near 25-30% in sales in 2009, layoffs-to-date are only the tip of the iceberg; hundreds of thousands of jobs will be shed, worldwide, before the bottom is reached.

Now, the RIF (Reduction in Force) Dilemma: The memory business is hugely capital intensive; about 70% of COGS is depreciation for state-of-the-art fabs. But after you have sold off your excess capacity (= "depreciation"), or written it down, the way to reduce overall costs is reducing headcount: SG&A, cutbacks in R&D, holidays which reduce accrued funds, or imposed ones which cost nothing. But if these jobs are NOT cut back in line with shrinking revenue streams, or the RIFs are done piecemeal and after the fact, the industry will continue to carry labor costs well in excess of what is needed to build the product that the market is taking. Corporate profits will continue to suffer, and corporate futures may be put at risk. Running too fat for too long is death.

No one wants to lay someone off, only to rehire them two months later when the industry turns back on. But the writing is increasingly and unambiguously on the wall: this is going to be a long and deep business reversal. No one can say with any authority much more than that, about how deep and how long, or about what the world will look like when demand is restored. But far lesser recessions than where this one appears headed… the inflation-busting take-down of 1981-82, the mini-recession of the early 1990s, the net.com bust in 2001..took huge tolls on the semiconductor industry, which, in the last two cases, took years of bouncing along the bottom before a real recovery took hold and lifted the industry up. By that time, market forces, product opportunities, and technology had moved new companies to the forefront, and ‘evaporated’ or ‘disabled’ those companies that merely THOUGHT they were running a tight ship, only to discover one day that they did not have what the future demanded and, then, did not have enough time or money to make amends.

Six or eight weeks ago, one could still ‘imagine’ that this might be a ‘down and back’ recession, and we would be off and running full tilt as we moved into 2010. Today, that is an impossible viewpoint, and the pundit and analyst opinion battle today is over ‘How dark is dark?’ or ‘How long is long?’

Here is my New Year's (Christmas?) List, for next year:

(1) First and foremost, with DRAM and NAND prices in the tank and the tough market, it will force restructuring in 'memory industry' over the next 1-2 years; expect 3-4 NAND makers and 3-4 DRAM makers to survive independently, and shrink more from there over time

(2) DDR3 DRAMs, now 2-3 years late, are slowly making their way into market...but the DRAM roadmap is being rewritten continuously with reduced voltage versions and higher density and GHz starting points for DDR3 to replace DDR2...for sure 2Gb and 1333MHz, maybe more before traction builds. DDR4? Well, with changes taking place, it's really too early to talk about it. Later life-cycle DDR3s will add features and get power down, without sacrificing performance. (The industry is on Rev 2 or 3 of the 6-7-8 DDR3 designs that will happen in the DDR3 lifecycle. Recall that SDR DRAMs started in ~1997 at 450-500nm 'half micron', and are today down to 120nm or better for the volume runners, 128M-512M densities...5-6 shrinks over ten years, and maybe more to come)

(3) Netbooks are the hottest 'compute trend' in years (10 y. for sure, maybe 15-20): (a) v. affordable, esp. for BRIC economies, (b) often uncoupled from WinTel, (c) 'blank slate' designs; (d) Maybe where SSD gets its first major (total) foothold. (e) Already 10-12MU+ in 2008, growing 40-50%/year? With huge advances in cost, performance and miniaturization ("high functional density at low cost")...plus years of 'consumer acceptance training', suddenly, now is the time, and small is the form factor.

(4) SSDs, the next big thing (for NAND), are likewise tackling technical and marketing issues one by one, with some success; the problem is complicated by product diversity, different (and heretofore, uncertain) applications market needs (R/W, endurance, cost, SLC-MLC-x3-x4) and associated enablement architecture-software-legacy issues (like 'drop-in-to-match HDD slots). It will be hit-and-miss for a long while, but will find the right way into many large applications over the coming several years. Today's market slaughter will bring improved prices and understanding of both the product and applications needs. There is much to be learned, decided...and opportunities abound, in getting to "The Best, First" with designs and product offerings; anyone can have tomorrow's "Standard" named after them...Xerox, Kleenex, iPod...if they are first to the right configuration/choice/implementation/vision. The controller roadmap(s) is in a state of unusual disarray and uncertainty (= opportunity), as all raw flash trends...multibit cells and process shrinks, adequate signaling...push "NAND memory subsystem remediation and error correction" problems off onto the controller, while the 'get cost/GB down' problem resides with the raw NAND array.

(5) The playing field will look v. different when we come out of the tough market: more big company dominance, way-way lower prices. The major trends apparent in 2005-6-7, paused for recession, to come back with renewed strength: BRIC growth, Low power, CE moves to displace computering as driving engine...your major end system buyer will be the householder, not Corporate America/Europe/Asia, plus maybe weaker architectural dominance by the Icons of past 20 years: Asian CE companies are the antithesis of Intel.