If there's still any doubt as to IBM's commitment to topple Seagate from the top of the disk drive heap, this aggressive rollout of its GMR technology should dispel it.
To review:
1.0 - 5.0 Gbits/in2, longitudinal or horizontal recording = IBM has demonstrated that it can extend conventional MR via single and dual stripe designs all the way up to 5.0 Gbits/in2. IBM Microelectronics makes most if not all of the read channel electronics for the disk drive unit so it is reasonable to assume that they will be able to develop the silicon to allow IBM to extend its density lead
2.7 - 10.0 Gbits/in2, longitudinal or horizontal recording = This seems to be the roadmap that IBM is developing for its GMR heads. Note how IBM is using its low-volume, high-margin server business to introduce its first GMR heads. This is the most lucrative part of IBM's disk drive biz because all its production goes into IBM's own line of workstations, servers and mainframe DASD products. Not coincidentally, these are not pure hardware sales for IBM which sells these products as part of a package that includes software, leasing and servicing deals.
10.0 Gbits/in2 and above, perpendicular or vertical recording = ????
Historically, IBM has been an unmitigated disaster in segments of the market where the money is made in the pure sale of hardware; e.g., disk drives, sub-$1,000 PCs. This go-around, however, they have developed the following model.
1) sub-$1,000 PCs - IBM is waiting for ACER, its main contract manufacturer, to increase its manufacturing capacity so ACER can make more of the low-margin, pure-hardware-sale type of PCs, laptops and servers for IBM. With disk drives accounting for up to 25% of the cost of materials of a PC, I am sure that IBM is going to give ACER the flexibility to take advantage of the fact that other manufacturers can make disk drives more cheaply than it ever can.
2) OEM Heads business - IBM spent $400 million in 1996 to develop the capacity to sell industry-leading MR heads to other OEMs. The premise is simply that this is another way for IBM to extend the useful lives of its MR heads as it moves from being cutting edge (and used only in IBM laptops and servers) to trailing edge (when yields are the highest).
3) NEC - IBM has lined up NEC to be the second manufacturing source for specified models of its MR heads and MR disk drives. Again, this is a way for IBM to extend the useful lives of its MR heads and disk drives. IBM gets to participate via royalties and it also puts itself in a position to get the business of Packard Bell/Zenith in much the same way that IBM's deal with Samsung lets it participate, via head sales and royalties, in the growth of Samsung's own PC business (2.5 to 3.0 million units this year).
IBM is a formidable player in disk drives now, but it still has major weaknesses that Seagate and to some extent Fujitsu are exploiting to the hilt. Succinctly, IBM doesn't have Seagate's capacity or Fujitsu's currency-advantaged cost structure and thus, it can not compete effectively in the fastest growing segment of the market, in unit terms.
Some people are making the mistake in thinking that Seagate and Fujitsu are both dumping in the low-end. I will argue, however, that Seagate and Fujitsu are executing an essential part of their necessary response to the IBM challenge by grabbing as much of the segments of the market that IBM is unable to participate. Is it really a coincidence that SEG and Fujitsu are "dumping" in the fastest growing areas of the PC business?
Think for a moment of the CGS (cost of goods sold) of each disk drive as having fixed cost and variable cost components. Because Seagate (80-20, heads and platters, ASICs and motors outsourced) and Fujitsu (50-50, heads and platters, most of the ASICs and motors are outsourced) are heavily integrated, they both have higher fixed costs than QNTM, WDC, Maxtor, etc. The more units they sell, the lower the fixed cost component of CGS. The same applies to variable costs. The more units they sell, the faster the economies of scale kick in and lower the blended costs of platters, heads, ASICs, etc. In other words, the higher the volume, the lower the fixed cost and variable cost components of CGS, the higher the gross margins.
Note that because the high-end segment is more competitive (read: lower margins), the premiums now only reside in segments of the market that require the cutting edge. For example, Seagate has raised the performance bar in capacity (Elite 47 - 47 GB disk drive) and speed (second generation Cheetahs are faster by 1 ms than IBM and Fujitsu's own 10,000 rpm disk drives). Seagate is also transitioning to a new type of fluid-bearing motors over the next 3 years that will insure that it can continue to raise the performance bar while it attempts to narrows IBM's increasingly formidable lead in density. As a point of reference, I am willing to wager to both IBM and Fujitsu will continue to buy some of Seagate's high-performance drives this year to satisfy the performance requirements of its customers.
So far, Fujitsu has not show any demonstrable technology lead other than to translate its world-class manufacturing process skills towards making more competitive disk drives. The fact that the yen continues to slide vs the dollar must give Fujitsu some kind of cost advantage because they use heads, platters and ASICs made in its own factories in Japan AND they buy heads, platters, ASICs and motors from the part of the disk drive supply line in Japan.
Given that backdrop of the integrated players' moves, what are the virtually integrated players like QNTM, WDC, etc going to do? Both QNTM and WDC have indicated higher R&D spending to catch up. WDC is headed by Haggerty who used to work at IBM. I believe that WDC has a licensing agreement with IBM that requires it to pay for the use of a multitude of IBM disk drive-related patents so I think we can expect WDC to pay IBM's price for its heads and technology to keep up. QNTM has broad cross-licensing agreements with Seagate and IBM so I am assuming that it can, for a price of course, keep up by paying the requisite licensing fees.
The bigger question for the virtually integrated players, however, is how are they going to get the volume and the turnover required to make their models work if the realignment results in a situation where Seagate and Fujitsu control much of the 3.0 GB and under category (the fastest growing category) and the 3.0 GB and up category turns viciously competitive (much to the delight of the PC makers) with Seagate, Fujitsu, Maxtor, WDC, Quantum and IBM all fielding competitive commodity (read: interchangeable) desktop products?
Bottomline: the uncertainty practically insures that we can forget about multiple expansion until the realignment of the desktop and the high-end become clearer. The uncertainty also insures that the entire disk drive supply chain will be extremely volatile until the smoke clears and the bloodied winners stumble into clear view. Strong stomachs are required. |
