Seagate tells flash bigots to get real
No way flash will boot disks out of notebooks
Comment Disk drive vendor Seagate has issued a document that tells flash evangelists to get real – there is no way flash will kick disk drives out of notebook storage for many years to come.
The pdf document is called "NAND Flash: Can It Meet the Growing StorageCapacity Demands of the LaptopPC Market?" and it's short enough to reproduce in its entirety. Read it yourself and see if Seagate is in denial or merely pointing out simple common sense:
Laptop PC storage capacity growth Continues unabated
One exabyte of storage capacity equals 1 million terabytes, or 1 billion gigabytes. The total available market (TAM) for laptop PC hard disk drives worldwide in 2010 was 69 exabytes and is forecast to grow to 95 exabytes in 2011.1 That’s a lot of storage capacity. What’s more, the average capacity of a notebook hard drive is forecast to grow from nearly 300GB in 2010 to more than 350GB in 2011.1 Laptop users want more capacity, not less.
NAND flash memory is the storage component of solid state drives. Conventional wisdom in some circles of the storage marketplace is that in the coming years solid state drives will begin to replace hard disk drives in significant numbers in laptop PCs. However, keep in mind that in 2010 the entire NAND flash memory industry had enough installed capacity to produce just over 11 exabytes of storage. More than 10 exabytes (93 per cent) of that went to consumer devices such as smartphones, tablets and SD cards.1 Just 0.86 exabytes (7 per cent) of that NAND was used in solid state drives.1
NAND flash memory production capacity is forecast to grow to 21 exabytes in 2011, with about 2 exabytes (9 per cent) going to solid state drives and the rest (91 per cent) to smartphones and other consumer devices.1 The cost to build a megafab capable of producing 3.75 exabytes of nonvolatile NAND flash memory is $10 billion.2 What’s more, a megafab – the minimum commitment for any significant increase in NAND production – would take two to three years to ramp to full production. Smaller fabs would contribute little to meeting the enormous demand for laptop storage.
Can it meet the growing storage capacity demands of the laptop PC market?
Assuming all of that additional NAND is used for solid state drives, that $10 billion2 investment would produce enough flash memory to serve just 4 per cent of the 95-exabyte laptop storage market projected for 2011. Spending $10 billion2 to buy 4 per cent of notebook storage market share, or $2 billion2 in revenue, is not viable. To serve the entire laptop PC storage market in 2010, a $170 billion2 investment in NAND flash memory fabs would have been required. In 2011, a $250 billion2 fab investment would be needed to meet projected hard disk drive capacity demand for all laptops. But $10 billion2 is just for the cost of the fab. It doesn’t include the NAND, operations, fab depreciation and other significant costs.
Worldwide installed fab capacity is expected to grow from 11.5 exabytes in 2010 to 21 exabytes in 2011, a staggering 82 per cent.1 But remember, just 9 per cent, or about 2 exabytes, of that NAND will go to solid state drives. Even at that impressive 80+ per cent growth rate, with the vast majority of the NAND going to consumer devices, the yawning gulf between NAND flash memory production capacity for solid state drives and demand for laptop storage will continue to widen.
Whatever portion of megafab production capacity is devoted to NAND flash for SSDs, the return on investment would be difficult to justify given the relatively small available market for laptop SSDs. Any additional capacity would be better justified to serve the market for smartphones, tablets and other consumer products for one chief reason: NAND makers can maintain much higher yields and lower prices for consumer-grade NAND because its performance and reliability specifications are much less stringent than the requirements for laptop PCs.
The upshot: Hard disk drives will serve the bulk of the laptop market for many years to come as makers of solid state drives remain overstretched to meet ever-growing demand for laptop storage. While the bulk of worldwide demand for NAND flash is for consumer products such as MP3 players, cellphones and cameras, Seagate believes there is ample flash to support the opportunities Seagate sees for enterprise and hybrid solid state storage.
1 Gartner, Forecast: NAND Flash Supply and Demand, Worldwide, 1Q09-4Q11, 4Q10 Update, page 2, Table 15-3, December 2010)
2 All figures in $US.
Is Seagate telling us the obvious truth here or is this document self-serving denial? With Seagate actually being in the flash business you would think the numbers would be good. It's probably the case that high-end notebooks may well go flash – as some are already – and hybrid flash-disk drives like Seagate's Momentum XT could take up some performance slack.
It could still be the case thought that the total addressable market for notebook hard drives is at or near its peak and could start to steadily decline over the next five to 10 years. ®
As somebody who sees a lot of laptops for repair etc, the vast majority of them only use a tiny amount of their storage. Most have 20-40gb of data and the rest is wasted space. Just that adjustment alone makes a huge difference to their maths, if only consumers didn't see capacity figures as `bigger=better`.
I'll bet if PC World sold identical laptops, one with 500gb 2.5" drive and one with a 60gb SSD, that most people would see the numbers and still go with the 500gb. In reality if they wanted a laptop for the most common uses (net, email, facebook) the flash version would be much more useful in real-world usage.
Not sure about the analysis
I can quite easily see that for large areas of the market, SSDs will displace hard drives. The statement about how much installed capacity there is on laptops rather implies that all that space is required. That might well be so for personal users who want to keep lots of bulky multi-media files(video and, to a lesser extent, photos being the obvious ones. Also, some games are incredibly bulky. However, these requirements are far from common in the business field. In general, the storage requirments for business documents, spreadsheets and the like are far lower than for what you find in domestic usage (with a few exceptions of course). In business, the important issues are often about productivity, speed to boot, battery life, robustness, reliability and weight. Time spent in booting machines, starting programs, applying updates not to mention support staff all costs money. In the case of my company, data is synchronised with central storage, and there is considerably impetus to reduce duplication with single-instancing and keep distributed data volumes to reasonable levels. The combined system and data partition on my worktop PC is just 64GB and is only half full (albeit that there is a hidden rebuild partition).
So, to equate installed capacity with market penetration is surely not the right metric here. That the majority of installed capacity will be on HDD is surely going to be true for a long time. The same might not be the case when measured by value or installed units.
From a personal perspective, I gave up using a laptop for my photos and video storage and processing long ago. I have a laptop, but it's not got much data on it. The bulk of data is serviced centrally, where I can properly protect it. The laptop is turning into not much more than an access device and 128GB is ample, and I could probably cope with 64GB compared to the 2.5TB I have on my central storage. What I need from the laptop is that it starts fast and runs on battery for a long time. I do not want it for bulk storage. In many ways domestic households will echo business practice in this respect with centralised storage and, for heavyweight work, processing.
During the 1920s the LNER was investigating high speed trains, and the germans had a high speed diesel railcar and the swiss had some very clever electric technology.
Because the infrastructure was there and they were familiar with the manufacturing they rejected both alternatives and built the A4 streamliners. Ten years later Mallard went on to break the steam speed record, and steam traction remained king until the middle 1950s.
Now, in the 21st century, no-one would consider even diesel for high speed operations, and electric is everywhere. Steam is for hobbyists and museums. And yet, Mallard remains magnificent.
The comparison with hard disks? Well, given that the life of a railway locomotive is between 20 and 60 years, we are around 3-4 generations on from the decision not to buy the flying hamburger. What will storage look like in 3-4 generations? How long is a generation? 3 years?
Hard disks are cruising into their Mallard era, but the future won't stop there.
Predicting the future is easy. Getting it right is nearly impossible.