Dear Wanna_bmw:
That is not what I am doing! You throw up an argument that Intel can not do this, but give no proof of it at all! You must show how my numbers are not possible or even likely. You have yet to do this. Falling back on the shadowy world of "But, I do not like yur numbers" and "Numbers out of thin air!" arguments are just that.
As to where I got $8 just divide Intel's GAAP profits for the quarter and divide by the claimed units sold in Q2. The rounded number is about $8. Now that is the change required to zero out the profit all other things being equal. Now to get to the amount that P4 ASPs would need to drop to requires the amount of the unit sales are P4s. Intel's plan for Q2 was 4 million units, but almost everyone else estimates that number to be somewhere between 1.5 and 2.5 million units. I took the middle of that and used 2 million units. NOw if all the ASP drop came from just P4s, you take $8 and multiply by total units and divide by P4 units and you get about $110 drop in P4 ASPs. Given Intel's past history of not producing much of a just released grade for a couple of months, no significant quantity of 1.9 and 2.0 P4s will be sold (the point at which Intel claims the revenue per statements). Thus, it is the prices of the P4s in current production ASPs. These prices are dropping about 40% to 60%. Take the average (usual for rough calcs) of 50% and you are close to the above required drop. Thus there is a good chance that all things being equal, Intel will experience a breakeven quarter in Q2.
But, all things are not equal. Intel wants to replace some P3s with P4s (they do not have the capacity to completely switch over at this time (P3s were sold this quarter)) and to do that typically requires the price of the desired CPU to be less than that of the old CPU (you have to have incentive to switch and P4 is not that much faster than P3 given the increased costs of P4 platforms over P3 (and yes this includes i845 because P3 MBs are fully debugged and tooled whereas i845 MBs are neither tooled or debugged (without i845 this would be even harder))). Thus P4 ASPs for those that are switched must be under P3 ASPs which loses ASP for each processor switched. In addition, a P4 costs more to make than a P3. Estimates range from $20 to $40 dollars each die packaged and all. I estimated the ASP loss to be about $10 and the cost rise to be $30 the average of the above. 0.13u production will not start selling in Q3 thus 0.18u production is required and the die is much bigger, yields are less, the package has more pins, etc. The best outcome is no switch of CPUs wrt Q3 revenue, but some will switch. If Intel gets all 6 million to switch (assumes no growth in overall units and 8 million P4 produced for Q3 sales), an additional $240 million (6 million times $40 ($10 + $30)) would be lost in profits.
To get back to $6.3 billion in revenue, more units would need to be sold or the mix would have to trend to higher ASP processors. The later is very unlikely in a down market and results from third parties show that if anything, the PC market is shrinking in Q3. Thus the best that can likely be achieved is sales of median ASP processors ($164 at last estimate for Q2). Since claimed gross margins are 48% (per Intel Q2 CC), an upper bound of average profit per CPU is $79 (48% of $164). Now divide $217 million by $79 and you get just shy of 3 million CPUs. Thus to break even in revenue (Q3 matches Q2) requires Intel to sell 30 million units in Q3. Now that is practically impossible given current conditions given that there is only 1 month to increase sales and no increase has happened to date.
And that is the best case since most increased sales would be on the low end, any increase of P4s will be offset by losses in P3s. Any unit losses (the corporate market is shrinking by most accounts) simply exacerbates the problem. Now the cuts in September will reinforce these trends and it is likely that true GAAP losses would result.
Product changeovers are easy when the new product performs much better than the old, when the market has strong demand (simply not make as much of the old CPU, Intel has done this in the past, but in these market conditions, most would simply not buy or buy competitors products) or when the new product can run some hot program that the old one couldn't. None of these are true in the P3 to P4 changeover. And having a competitor with a product better than your old one makes it even worse.
As to debating those numbers (you need to go back and look at old posts (they are all still there from previous debates on this thread and the moderated AMD thread)), we could restart some of those old debates. Pick one (or more). Saying they come out of thin air isn't good enough.
Tell me the one(s) you do not like and provide me with the number(s) you think is(are) correct for each of those you dispute. Why should I search through all 200K posts and all the news and press releases without requiring you to restrict the dispute to a far smaller fraction of them? Or is it that you are to lazy to do so and thus, simply cop out without showing the proof behind your thinking?
Pete |
