Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/are-the-major-dram-suppliers-stunting-dram-demand.10212/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021770
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Are the Major DRAM Suppliers Stunting DRAM Demand?

Daniel Nenni

Admin
Staff member
Skyrocketing DRAM prices potentially open the door for startup Chinese competitors.

Historically, the DRAM market has been the most volatile of the major IC product segments. A good example of this was displayed over the past two years when the DRAM market declined 8% in 2016 only to surge by 77% in 2017! The March Update to the 2018 McClean Report (to be released later this month) will fully detail IC Insights’ latest forecast for the 2018 DRAM and total IC markets.

In the 34-year period from 1978-2012, the DRAM price-per-bit declined by an average annual rate of 33%. However, from 2012 through 2017, the average DRAM price-per-bit decline was only 3% per year! Moreover, the 47% full-year 2017 jump in the price-per-bit of DRAM was the largest annual increase since 1978, surpassing the previous high of 45% registered 30 years ago in 1988!

In 2017, DRAM bit volume growth was 20%, half the 40% rate of increase registered in 2016. For 2018, each of the three major DRAM producers (e.g., Samsung, SK Hynix, and Micron) have stated that they expect DRAM bit volume growth to once again be about 20%. However, as shown in Figure 1, monthly year-over-year DRAM bit volume growth averaged only 13% over the nine-month period of May 2017 through January 2018.

Figure 1 also plots the monthly price-per-Gb of DRAM from January of 2017 through January of 2018. As shown, the DRAM price-per-Gb has been on a steep rise, with prices being 47% higher in January 2018 as compared to one year earlier in January 2017. There is little doubt that electronic system manufacturers are currently scrambling to adjust and adapt to the skyrocketing cost of memory.

DRAM is usually considered a commodity like oil. Like most commodities, there is elasticity of demand associated with the product. For example, when oil prices are low, many consumers purchase big SUVs, with little concern for the vehicle’s miles-per-gallon efficiency. However, when oil prices are high, consumers typically look toward smaller or alternative energy (e.g., hybrid or fully electric) options.

82b3f8c0-d056-47e7-97bc-75b9ef515ae8.png

Figure 1

While difficult to precisely measure, it is IC Insights’ opinion that DRAM bit volume usage is also affected by elasticity, whereby increased costs inhibit demand and lower costs expand usage and open up new applications. As shown in Figure 1, the correlation coefficient between the DRAM price-per-bit and the year-over-year bit volume increase from January 2017 through January 2018 was a strong -0.88 (a perfect correlation between two factors moving in the opposite direction would be -1.0). Thus, while system manufacturers are not scaling back DRAM usage in systems currently shipping, there have been numerous rumors of some smartphone producers scaling back DRAM in next-generation models (i.e., incorporating 4GB of DRAM per smartphone instead of 5GB).

In 2018, IC Insights believes that the major DRAM suppliers will be walking a fine line between making their shareholders even happier than they are right now and further alienating their customer base. If, and it is a BIG if, the startup Chinese DRAM producers can field a competitive product over the next couple of years, DRAM users could flock to these new suppliers in an attempt to get out from under the crushing price increases now being thrust upon them—with the “payback” to the current major DRAM suppliers being severe.


Report Details: The 2018 McClean Report
Additional details on other trends within the IC industry are provided in the 2018 edition of The McClean Report—A Complete Analysis and Forecast of the Integrated Circuit Industry (released in January 2018). A subscription to The McClean Report includes free monthly updates from March through November (including a 250+ page Mid-Year Update, and free access to subscriber-only webinars throughout the year. An individual-user license to the 2018 edition of The McClean Report is priced at $4,290 and includes an Internet access password. A multi-user worldwide corporate license is available for $7,290.

To review additional information about IC Insights’ new and existing market research reports and services please visit our website: IC Insights | Semiconductor Market Research.


 
I think part of what is going on here is that it is getting increasingly difficult to shrink DRAMs. This has led to a slower rate of shrinks and cost reductions. The manufacturers aren't getting as many additional bits for shrinks leading to less supply growth. The advantage of being first to a new "node" is smaller and the manufacturers aren't pushing as hard to be first.
 
The new nodes also flow longer and slower through the plant, which adds production cost offsetting the density. Some of these come on line only at the expense of shutting down and restarting lines that produced older parts.

The new Chinese plants will be plentiful capital in new workforces competing with experienced plants and partly depreciated production equipment, though capital is cheap in Korea or Japan too. China will repatriate a lot of sea turtles to lead and train the new workforces but they will not be at the most modern nodes. I suspect they will hollow out the volume market of low end mobiles. Any shift in the overall balance could bring the sellers' market down to earth.

This cyclic problem with commodities has been endemic to the electronics industry. It happened in HDDs, and has happened many times before with DRAM, and a few times already with NAND. Overall the trend is good but the feast and famine cycles often wipe out companies, both consumers and suppliers.
 
The new nodes also flow longer and slower through the plant, which adds production cost offsetting the density. Some of these come on line only at the expense of shutting down and restarting lines that produced older parts.

The new Chinese plants will be plentiful capital in new workforces competing with experienced plants and partly depreciated production equipment, though capital is cheap in Korea or Japan too. China will repatriate a lot of sea turtles to lead and train the new workforces but they will not be at the most modern nodes. I suspect they will hollow out the volume market of low end mobiles. Any shift in the overall balance could bring the sellers' market down to earth.

This cyclic problem with commodities has been endemic to the electronics industry. It happened in HDDs, and has happened many times before with DRAM, and a few times already with NAND. Overall the trend is good but the feast and famine cycles often wipe out companies, both consumers and suppliers.

If the Chinese successfully enter the DRAM market I would expect a big price plunge and everyone moving into negative margin territory as we have seen in the past. However, I am not convinced the Chinese will succeed, the processes are so much more complex and difficult than in the past and there will likely be a lot of legal battles around the IP.
 
I think the title (due to "IC Insights"?) is somewhat misleading. Although the answer is inevitably "yes", the implication is that the Suppliers are artificially limiting the demand beyond what is necessary to nake it match their supply capability. I believe that there is a genuine supply shortage, and the present pricing is not due to some abnormal cartel-type response.

Scotten Jones effectively points out that, contrary to usual perceptions, DRAM has become a mature technology in terms of cost base, but with without shedding the technical challenges of an advancing technology. Given the limited scope for progressive cost reduction, development has concentrated on providing performance; this has shown in the limited ability of manufacturers both to reduce price and to increase volumes.

All the above has frequently been discussed as part of the background for work on alternative memory structures. Their time is clearly here - as soon as the technologies are proved.

This looks be a potential harbinger of things to come in total volume of transistors produced: the latest advances appear capable of resulting in increased performance, but not reduced cost/device. Perhaps this has been more successfully trailed?
 
There'd be less need to go on increasing dRAM sizes if software was less inefficient. OK, much of it is written by people who weren't alive when Bill Gates said 640K should be enough for anyone, and buffering hi-res images takes more than that, but I suspect apps that take up several GB of RAM could use a lot less if there was an incentive to do so.
 
If the Chinese successfully enter the DRAM market I would expect a big price plunge and everyone moving into negative margin territory as we have seen in the past. However, I am not convinced the Chinese will succeed, the processes are so much more complex and difficult than in the past and there will likely be a lot of legal battles around the IP.

Legal challenges maybe. It will be a fun venue if the contracts are China producers and consumers.

I think they will get the processes running faster than doubters believe. It is not like they have zero experience to draw on. They will attract the most skilled technicians to seed and lead the workforce, and plenty of sea turtles at the higher level will be well paid to repatriate and run a modern plant. With the latest equipment, courtesy of tons of cheap capital.

Also, if ever there was a time to sell a trailing edge product, now is the time. They could take a "bath" on the price and still be profitable. Remember, two years ago DRAM was a third of the current price and still turning a profit. So, there are plenty of reasons to expect the China fabs to become a market force.
 
Last edited:
There'd be less need to go on increasing dRAM sizes if software was less inefficient. OK, much of it is written by people who weren't alive when Bill Gates said 640K should be enough for anyone, and buffering hi-res images takes more than that, but I suspect apps that take up several GB of RAM could use a lot less if there was an incentive to do so.

Software per se is not that much larger. Tricks like generating calls inline are done because they are cost effective with modern CPUs and memory. It is a rational decision, not sloppy, and varies on different CPUs if you hint the compiler correctly.

In any case, it is the data which is bigger, the code sizes are trivial compared to the data. That also is a rational choice. If you make storage systems when it costs 1 cent per GB per year to store some data, and that data on average can earn you 10 cents by answering questions later, then people store that data. At current prices there are perhaps 100 exabytes per year worth storing. And growing.

In that situation, having computers with 128GB of DRAM and tens of terabytes of storage media is quite ordinary. At any given moment maybe 0.1% of the world's data is in DRAM and that is where it gets useful work done.

We made big computers cheap. People come up with reasons to use them. Even Gates (who was taken out of context on the 640k) knew that. He bet his company on the exponential growth, right at the start. But I suspect he is surprised how long it has continued, and seems set to continue.
 
Back
Top