Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: vince doran who wrote (72037)	9/16/1999 3:55:00 PM
From: Charles R	Respond to of 1573104

Vince, <When do you think Athlon mobos supporting AGP4x will be available? I ask because I think the new Nvidia GeForce graphics chip is going to be a driver for gamer purchasing decisions this Xmas as well as a powerful avdertising point for system ads (the specs are awesome compared to the current generation.) VIAs chipset for Athlon supports 4x.> I have seen some October dates but I am not counting on it for Christmas. If it happens, I will consider it an upside. <With pretty certain CUmine 733 MHz selling in Oct. and assuming the rumoured 700 MHz Athlon is selling in quant then, Intel will be able to reclaim (by a nose) the MHz crown for the Xmas season, while the benchmarks will still probably favor Athlon by 5-15%. In the public's mind, ties go to the incumbent champ. In that situation, do you think there will be sufficient demand for Athlon 700s (for whatever reason, anti-Intel sentiment, desire for higher FPU perf, higher cost or slower perf of Intel sys due to Rambus, whatever) to allow AMD to sell 1M Aths at $300 asps?> There are multiple sources confirming 700MHz shipments at this point and I am fairly certain at this point that 700MHz Athlons will be in the channel next month (along with 650s which are hard to get so far). I have a hard time seeing Intel leading on MHz for a long time to come. The ASP part is up in the air. It is clear that Athlon Q3 ASPs aren't going to be good because of the delays. As for Q4, AMD is doing the highest speed ramp for a new CPU that I am aware of in this business. This is also the first time AMD is a serious player in this segment. How it plays out on the ASP side will be interesting to watch. (I take the collapse of PIII-600 ASPs to mean that AMD is gaining sockets albeit at lower ASPs than one would like to see for those speed grades) <Do you think the reported Rambus problems are something Intel thinks will be fixed by CUmine intro, or that they are hoping simply to bull the change through problems and all? (Blow the horn on the high mem MHz and ignore the probs?) Or is it possible they can come up with a last-minute way to avoid Rambus and still ship CUmine?> RDRAM is looking pretty iffy right now. My sources tell me PC133 is the main memory for CuMine next quarter but I have public contradictions from intel to weigh that against. I am in the wait-and-see mode. <Assuming a neck-and-neck MHz race through the end of the year, do you think AMD can sell enough Aths for enough asp to reach profitability by early 2000?> I think it is strictly a volume game, if AMD has no glitches on the way I have no doubt they will be profitable. <I am long AMD at 17 7/16s (having sold INTC at 67 to shift) because the upside/downside potential is much greater, however, I don't think AMD has a long window here. I don't assume that Intel is incapable of getting it in gear, so for my purposes, I assume Willamette will arrive in summer 2000 and will be faster than the current Ath design. Thus to expand my current bet I must feel confident that AMD can reach profitability and capture significant public mindshare as an equally good high-perf supplier by spring. Do you think they can meet that challenge? I would include the assumption that Process Boy is correct and CUmine will stay close in the MHz race.> Clearly, investment decisions are very personal and it has a lot to do with comfort factor. I did not think I would add to my AMD position till I saw at least one more large OEM announce publicly but being pretty certain about the 700MHz story, I bought some more this AM on the dip. Good luck Chuck P.S.: I gotto tell you - I am counting on a pretty strong slew of announcements in the next few weeks. I will trade out in a heartbeat if that does not materialize and if I see continued execution problems.

To: vince doran who wrote (72037)	9/16/1999 4:23:00 PM
From: kapkan4u	Read Replies (2) \| Respond to of 1573104

Vince, <When do you think Athlon mobos supporting AGP4x will be available? > My estimate is that Athlon/VIA/AGP4x systems will become available in early Q1, 2000. <With pretty certain CUmine 733 MHz selling in Oct. and assuming the rumoured 700 MHz Athlon is selling in quant then, Intel will be able to reclaim (by a nose) the MHz crown for the Xmas season, while the benchmarks will still probably favor Athlon by 5-15%. In the public's mind, ties go to the incumbent champ. In that situation, do you think there will be sufficient demand for Athlon 700s (for whatever reason, anti-Intel sentiment, desire for higher FPU perf, higher cost or slower perf of Intel sys due to Rambus, whatever) to allow AMD to sell 1M Aths at $300 asps?> I think that you outlined a conservative scenario here. Several upside surprises are possible: 1. We may see Athlon 750s for sale this Xmas season; 2. The Athlon performance advantage over CuMine/DRDRAM at the same clock speed may stay similar to today's PIII/PC100, because 820/DRDRAM133 appears to be slower than 440BX/PC100; 3. DRDRAM shortages can severely affect the availability of the Camino systems. 4. DRDRAM costs will make the Camino systems more expensive than comparably equipped (but faster) Athlon systems. <Do you think the reported Rambus problems are something Intel thinks will be fixed by CUmine intro, or that they are hoping simply to bull the change through problems and all? (Blow the horn on the high mem MHz and ignore the probs?) Or is it possible they can come up with a last-minute way to avoid Rambus and still ship CUmine?> Just wait for major PC mags publish CuMine/Rambust benchmarks under-performing cheaper Athlon systems. <Assuming a neck-and-neck MHz race through the end of the year, do you think AMD can sell enough Aths for enough asp to reach profitability by early 2000?> I think AMD will sell all Athlons they make in Q4. It could be anything between 800K and 1.5M units. I project $.80 loss in Q3, $.10 profit in Q4 and $.30 profit in Q1. <I am long AMD at 17 7/16s (having sold INTC at 67 to shift) because the upside/downside potential is much greater, however, I don't think AMD has a long window here. I don't assume that Intel is incapable of getting it in gear, so for my purposes, I assume Willamette will arrive in summer 2000 and will be faster than the current Ath design.> The "window of opportunity" may turn out to be MUCH longer than most people are predicting now. As you can see I have a very bullish view of AMD stock prospects and I put my money where my mouth is. Kap

To: vince doran who wrote (72037)	9/16/1999 4:39:00 PM
From: Tenchusatsu	Respond to of 1573104

Vince, long reply follows: <When do you think Athlon mobos supporting AGP4x will be available? I ask because I think the new Nvidia GeForce graphics chip is going to be a driver for gamer purchasing decisions this Xmas as well as a powerful avdertising point for system ads> AGP-4x will certainly be a powerful advertising point; however, it's immediate usefulness seems to be similar to the immediate usefulness of Rambus. In other words, AGP-4x is more forward-looking. I don't expect even nVidia's GeForce engine to demand much more than AGP-2x for a while. <In that situation, do you think there will be sufficient demand for Athlon 700s (for whatever reason, anti-Intel sentiment, desire for higher FPU perf, higher cost or slower perf of Intel sys due to Rambus, whatever) to allow AMD to sell 1M Aths at $300 asps?> I think the real question is whether there will be a sufficient supply of Athlons at any speed that can be sold at $300 ASP or above. Obviously there's going to be sufficient demand for Athlon 700 CPUs. The supply is initially going to be very low, so even a little demand is sufficient enough. It all depends on what speed we pin the $300 price point at. Right now, only the 500 MHz speed grades (Athlon and Pentium III) are below that point. By the middle of Q4, we'll probably see prices for 550 MHz fall below that point. (I wonder many Athlons are being sold at 500 MHz right now, even if all of the attention is on the 600 and 650 MHz speed grades.) <Do you think the reported Rambus problems are something Intel thinks will be fixed by CUmine intro> If not, I think they will be fixed shortly after the intro of Coppermine. <Assuming a neck-and-neck MHz race through the end of the year, do you think AMD can sell enough Aths for enough asp to reach profitability by early 2000?> I don't know how big AMD's costs are, and that's a big factor in determining what is necessary for profits. I think, however, that AMD can sell every Athlon they make and still come out with a very respectable margin on each sale. <I don't think AMD has a long window here. I don't assume that Intel is incapable of getting it in gear, so for my purposes, I assume Willamette will arrive in summer 2000 and will be faster than the current Ath design.> I agree about the window, but summer 2000 seems to be a stretch. Intel mentioned late 2000 for Willamette release, and the story hasn't changed as of yet. <Thus to expand my current bet I must feel confident that AMD can reach profitability and capture significant public mindshare as an equally good high-perf supplier by spring. Do you think they can meet that challenge?> It takes time for AMD to build up a reputation of being a good high-performance volume supplier. But time is not exactly on AMD's side, so AMD has to make up for it by showing excellent results in the short-term. Because of this, I don't think the shareholders will tolerate even one hiccup in execution in 2000. <There is no doubt that AMD presents a compelling risk/reward picture given its net asset value and valuation relative to INTCs, however, the reward side of the analysis depends on its ability to actually show promise of capturing some of INTCs monopoly profits> Given that even the astounding Athlon benchmarks was only good enough for a two to four point rise, it seems like shareholders want to see more, much more, before AMD truly does show promise. I'm personally considering getting out of AMD sooner or later. Why? Because it's becoming apparent to me that if AMD succeeds, the rewards will come cautiously, but if AMD fails or stumbles even slightly, the penalties will come swiftly. Wall Street isn't very kind to AMD, it seems, and I don't think AMD can be forgiven again the next time they stumble. Tenchusatsu

To: vince doran who wrote (72037)	9/16/1999 4:59:00 PM
From: kash johal	Read Replies (2) \| Respond to of 1573104

Vince, Re: "AMD as an investment" Well my view on AMD as you know is bullish. Having said that I think you are focussing on the wrong issues : Q4 should be damn good as they should easily sell the 1M K7s and perhaps eke out a profit. And the impending release of Athlon Ultra's will start to help AMDs market positioning when it blows the Xeon/Cascades out of the performance workstation sockets. But there are 2 major problems on the horizon for AMD starting in Q1 2000: 1) K6 replacement. The K6 is dead and once the coppermine with 128K cache comes out from Intel, AMD stops having a competitive low end processor IMHO. There has been talk of an AMD Athlon socketed version to compete with this. However AMD MUST have the 0.18 micron version of the AThlon in volume by Q1 2000 to compete with this IMHO. 2. This brings us to the BILLION dollar question surrounding AMD ie dresden. We know that Dresden already is up to 800 folks (or half of its full employment). They are sitting there with their thumbs up their assess right now. If dresden was an AL fab these guys would be cranking product well before NOW. If Dresden comes up by Q4 and start producing some AThlons for Q1 shipments or AMD is in a world of hurt by Q1 2000. If Dresden doesn't yield etc we will have a MAJOR disaster and I can see the stock tanking into the teens. Of course if Dresden ramps better than expected we could have volumes in Q4 helping the K7 volume ramp and get AMD dramatically profitable by Q1 2000. Personally I have written some calls to give a hedge. If the stock rallies to high twenties prior to earnings with all the Ultra hype I may even buy puts. I don't think this is a stock one can buy and put away for a period of time. But it sure will make an exciting 3-6 months. regards, Kash

To: vince doran who wrote (72037)	9/17/1999 12:49:00 AM
From: Petz	Read Replies (1) \| Respond to of 1573104

vince, some comments on Nvidia GeForce and "Can AMD sell 1M?" Have you read the comments on one of the gaming sites from NVidia's CEO, about Athlon and how the GeForce will "rock" on it? Nvidia is committed to "Athlon optimizations." A second point on Nvidia is that many CuMine motherboards will not even have an AGP slot. This is because the 667 and 733 MHz CuMines can only run on the i810e chipset or the i820 chipset. Only i810e supports PC133 RAM and PC100 RAM directly, but the i810e includes AGP graphics in the chipset, and no support for an AGP slot. To support PC100 or PC133 SDRAM, the Camino i820 requires an adaptor chip which is estimated to slow memory performance to the PC66 level! No CPU, especially one with only 256K of L2 cache can run effectively with an 11:1 CPU to memory bandwidth ratio. Therefore, its a safe bet that Camino motherboards will have abyssmal performance using SDRAM. On the i820 motherboards, you'll have to plunk down over $500 for 128M RDRAM memory (its $800 now) to get "decent" performance. VIA's chipset supports an AGP slot and PC133 memory, but I think they will be production limited. On the question as to whether AMD can SELL 1M Athlon CPU's, remember, thats only 3% of CPU production in the fourth quarter. I would guess that the top two Intel speed grades get roughly 20% of the number of CPU's sold. So, AMD only needs to sell 1 out of 6 of such "high end machines." I do think they would be demand limited at about 2 million Athlons in the fourth quarter, if they could make that many. Petz