Robert F., According to Jay Swent (per the Conference Call), all of the expensive May 17th Wafer Agreement chips were sold last quarter. Also remember that Cyrix recently went through a number of die shrinks, from 0.6 micron to 0.5 micron to 0.44 micron to finally 0.35 micron. Each die shrink not only lowers the cost simply based on the fact that more die are on each wafer, but also lowers the cost because each shrink improves the yield. Remember Cyrix buys wafers not functioning die. For the last three quarters, Cyrix sold 200,000 units (2Q96), 250,000 units (3Q96), and 575,000 units (4Q96). It is not inconcievable for Cyrix to have fully tested a large number of 6x86's anticipating growth in demand and therefore having a fair number of fully tested and packed 6x86's in inventory.
Here is an example of how swapping fully functional chips and more expensive (larger die chips) could result in a reduction of 24 million dollars worth of inventory. Let us assume that Cyrix had 200,000 fully tested 6x86's in inventory in Q3 and that these chips cost 80 dollars to produce (if these chips were larger die chips manufactured under the original IBM foundry agreement, this cost is possible). If current 0.35 micron die in raw form cost 30.00 each (See Carl's post), then swapping those 200,000 6x86's for raw 0.35 micron die reduces the value of this inventory by 10,000,000 dollars. If we assume that the remaining die were procured under the May 17 IBM Foundry Agreement and they were manufactured with the 0.44 micron process, so that the wafers cost 50% more and number of chips are 50% more expensive due to less yield and less chips (i.e., the chips cost 65 dollars instead of 30 dollars for the 0.35 micron). The value of replacing 375,000 expensive die with inexpensive die is 35*375,000 or 13,125,000.
The total reduction in iventory value is 23,125,000:
for 200,000 6x86's fully tested (50*200,000)= 10,000,000,
and 375,000 expensive die (35*375,000) = 13,125,000.
Of course, if the total number of units declined even a small amount lets say 50,000,000 and vendor leadtimes are more favorably controlled it looks like some smart inventory management and appropriate swapping of inventory could reduce the inventory value dramatically.
I am still waiting for Mark Lipscom's response to my inventory questions to confirm these calculations. I think that analysis indicates that we are "in the ballpark".
In regard to the Cyrix open position postings. These postings could easily be needed for M3 support and for the 5gx86 support as well as future generations of these chips. If M3 is going to be introduced in 1998, the core team will need to be increased in size to do all of the grunt work required to complete the project.
Bob |
