Re: I wouldn't even know where to begin to discuss with you
Yes, you haven't been following this game for all that long. AMD is coming up on "third time's the charm." The first run at Intel was with the K6, which had a brief moment of glory, then fell back when AMD had process problems - that run didn't last long. The second run happened last year when AMD beat Intel to copper and the Athlon core beat the P4 core to market. That run went much better, letting AMD pay off its debt, fund the move to SOI, and bankroll a price war.
This third run, which begins this winter, things are different. Intel has always had the benefit of a better, cheaper, more developed platform - partly because PIII used less power than Athlon. Now AMD has a platform of its very own, that is supported by many chipset companies, that offers SMP support, and is coupled to a processor that requires about the same, or a little less power, than Intel's. SOI may make that a lot less power.
AMD is also going into 2002 with unprecedented manufacturing capacity. With thoroughbred speced out at 80mm2, and both Austin and Dresden at full capacity, OEMs will, for the first time ever, be able to worry less about Intel's "cut off your oxygen" threats. By Q1, between Dresden and Austin, AMD will be able to supply around 15 million CPUs per quarter (theoretically, if all wafers were production, wafers starts were 6K/week, edge loss was 10%, wafer yields were 90% and yields on good wafers were 75%, AMD could ship 18.5 million CPUs each quarter from Dresden alone, but the reality is a lot of other stuff has to go on, and yields probably won't be that good, either.)
Right now, Intel is able to intimidate the OEMs into barring AMD from the "business" market - try and buy a "corporate" PC from any OEM with an AMD processor and see what you get. I have looked, and they aren't available - Intel has made it quite clear that any company that dares to offer an AMD corporate PC will be cut off. The result is that Intel is able to get 90% of the market's revenue from 75% of the unit sales (VIA, etc, get a couple percent, while AMD gets about 22.5%).
If SOI lets AMD make a sufficiently compelling case to any major OEM, and gets a faster, lower power, SOI, CPU into the corporate market, Intel could find itself matching prices with AMD in the corporate, as well as the retail, market. Intel could go from 90% of the revenue on 75% of the sales to 75% of the revenue on 65% of the sales - and the size of the pot would shrink enormously as pricing pressure extended to the corporate market. Intel stands a reasonably good chance of finding itself with 75% of a $3 Billion CPU market (per quarter) next year instead of the 90% of a $5 Billion CPU market they had this year - a 50% revenue drop. AMD would have 25% of that $3 Billion market instead of 10% of $5 Billion (a 50% revenue increase) and VIA would get crushed by the price war.
Intel could be consistently losing $1 Billion to $2 Billion per quarter next year, while AMD made unprecedented profits.
Things can happen very, very, fast in the semi business, especially if a player makes an unusually large capex gamble that doesn't pay off. I still can't believe that Intel spent all that money to wind up lacking the process technology that every other major player in the industry claims is absolutely necessary to be a performance leader. |
