Advanced Micro Devices
Good Time to Buy

Format for Printing

Format for printing

Request Reprints


By ValueNut
July 15, 2004

Posts selected for this feature rarely stand alone. They are usually a part of an ongoing thread, and are out of context when presented here. The material should be read in that light. How are these posts selected? Click here to find out and nominate a post yourself!

I've just bought another 1000 shares at around $13.75 a few hours ago and also quite a few more Oct 15 call options (I wouldn't buy margin right now, but some near-the-money options seem like a very good bet).

Ever since Q1, it's been clear that AMD would outperform in Q3 and Q4 of this year, but that Q2 had such a confusing mix of negative and positive issues, that it made sense to look for a lower entry point than $15 for the last few months. I don't think it's reasonable to expect that we are going to get a better entry point for a sale in Q4 or Q1 than today, which is not to say I necessarily expect AMD stock to increase rapidly any time soon. I think it is simply rational that a buy for under $15 today is likely to result in an opportunity to sell in excess of $20 sometime in the next six months -- perhaps considerably north of $20.

Like many people on this board, I expect neutral to positive results for this quarter in terms of expectations, but I expect to begin to hear more and more positive projections from AMD -- if not in this conference call -- then the Q3 after-earnings call. That is likely to have a positive impact on the stock price over the next few months.

I suspect that flash earnings will be good this quarter even if Intel did manage to steal some of AMD's share (which I think is unlikely). AMD had recently projected that they could not meet demand for flash thorough at least Q3, so not selling to expectations in Q2 would be a huge change in projections that would have probably been noticed by now.

I still expect microprocessor units to be down for Q2, but I also expect ASPs to be up by about the same amount, so that microprocessor revenues will be flat. If AMD managed to take advantage of volatility caused by Intel's Grantsdale recall, they might even have a bit of a microprocessor unit revenue increase with a corresponding drop in units in inventory. This will be something to check for in today's report.

I interpret Intel's inventory overhang with guarded optimism. On the one hand, Intel may need to sell that inventory at lower prices than expected to reduce its size, which in the worst-case scenario could mean a price war with AMD. However, I interpret Intel's announcement that they are slowing production to be a communication to the market that they are NOT going to launch a price war. This means that AMD and Intel can continue to keep prices high, which makes both companies look good. Clearly, neither company is looking for a price war.

In addition, Intel seems to clearly have problems producing parts at higher speeds, whether that is because of binning or yields. This is potentially a serious problem for Intel because if they are averaging similar clock speeds with the transition to 90 nm as they were at 130nm, there are only two reasons to aggressively convert to 90nm. (1) They expect better high clock rate yields in the future, so it's worth producing parts no better than those at 130nm now to keep their learning curve relatively fast. (2) The 90nm parts are cheaper to produce and yield more good dice per wafer.

While I suspect that (1) is true, I suspect that (2) IS NOT TRUE. The 90nm Prescott is about the same die size as the older Northwood and requires an additional metal layer and strained silicon. So unless yields at this stage are outrageously good for the current point on the learning curve, the Prescott is more expensive than Northwood.

If Intel is producing fewer parts, they need to choose whether to produce fewer 130nm parts or fewer 90nm parts. One choice reduces future Prescott clock rate performance, the other choice produces lower margins in Q3 and perhaps the rest of the year. The very fact that Intel needs to make this choice (regardless of what that the outcome is), is good news for AMD.

Will Intel solve its Prescott problems? Undoubtedly! But if they reduce the speed of the 90nm ramp, they may not get there quickly enough to stay ahead of AMD. If Intel increases the 90nm ramp rate, they will have to drop 130nm production, which will cause margins, perversely, to drop. In the years I have been following Intel and AMD, this is first time I have not seen either significantly greater clock rate or greater yields from a gate size reduction within the first six months of a product launch.

Is AMD's 90nm launch going well enough? I say the answer as I write this is "Who knows". But the answer to that question may be known and fairly well understood by the end of today, or as late as the beginning of Q4. However, Intel's current problems make me less concerned about AMD's 90nm conversion. AMD appears to not really need 90nm before Q4.

When AMD does get 90nm, AMD's 256k and 512k cache 90nm processors will give them an immediate reason to switch to 90nm production even if yields or binsplits are low, simply because of the die size advantage. AMD may choose to switch lower performance parts to 90nm in high volumes just to reduce the costs. Killing the AXP and producing Semprons will likely reduce AMD's costs significantly -- if not in Q3 than at least in Q4 and lets them take a leisurely pace to high performance on 90nm while keeping the 90nm learning curve fast. When both companies successfully transition to 90nm, AMD will retain a die-size cost advantage as long as their performance remains high. I believe that the die size advantage will approximately cancel out Intel's 300mm advantage.

Couple that with the recent rumors about SSE3 support for the new 90nm parts and lower power and higher efficiency at the same clock rate, and it implies that AMD may be able to launch the 90nm parts at even higher model numbers than the clock rates would suggest. (I suspect this will be a 100 MNU improvement at the same clock rate -- based on 5% better performance.)

Overall, this looks like a very risk-free time to buy AMD with worthwhile upside potential.


Become a Complete Fool
Join the best community on the web! Becoming a full member of the Fool Community is easy, takes just a minute, and is very inexpensive.