E, the problems you sight are significant, but I wonder how long they will last. The thing I like about ESI is I don't have to always get it right on a day to day basis. I think there is a quality of products and management that will pull this company through as the future demands increasingly complex products at lower costs.
Take DRAM for example. They must squeeze every possible chip out of a wafer. Yield repair is highly a visible tool with good return on equity. It's going into existing fabs and new fabs. The market is not saturated. Then look at the new DRAM's coming into market. They are faced with typical new product concerns of low yields and high profit potential should they be able to improve yields. Do you think they will buy? See an attached article.
In the end, I think short term we could see lower prices, but in the long term, I think we've got a pretty safe winner. When money does a flight to quality, they should look at ESI.
Regards,
Mark
By Ron Wilson
SAN JOSE, Calif., A sudden jump in demand for very fast synchronous DRAMs has helped pull Intel Corp.'s PC-100 specification into the mainstream of the personal computer market, according to George Robillard, vice president of Fujitsu Microelectronics Inc.
"Initially, the industry had been focusing on PC-100 for the first quarter of this year," Robillard said. "But in the first quarter nothing much happened. Now we are seeing a real burst in demand for SDRAMs that can meet PC-100 requirements, both from some major OEMs and from some module vendors."
For modules to comply with the PC-100 spec, they must hold SDRAMs that are capable of operating at about 125 MHz. The chip maker supplying the SDRAMs and the module maker must judiciously allocate the modules' flight times, set-up, hold and access times to see that the overall module complies with Intel's stringent specification.
The increased demand for such chips has raised a number of debates over chip and module timing. Module vendors are reportedly discovering significant differences in timing between SDRAMs from different vendors, all of which nominally can meet the PC-100 requirements.
Another emerging issue involves a spec known as CL-2 -- the ability of the module to initiate its first transfer two clocks after the CAS signal is sent from the system controller. While there is little difference in application performance between systems using CL-2 or CL-3 modules, the difference is detectable by synthetic benchmarks, and hence has marketing significance.
Vendors are currently seeing low yields on parts that are fast enough to meet the CL-2 timing spec, so compliance has become a bragging point for DRAM vendors as well as systems vendors. Toshiba America Electronics Corp., for example, has announced that its parts comply with CL-2.
CL-2 will be more readily achievable once DRAM processes migrate from 0.30 to 0.28 microns, said Fujitsu's Robillard. |
