POST OF THE DAY
Advanced Micro Devices
Intel Gives Chip Market Six-Month Deadline

Format for Printing

Format for printing

Request Reprints

Reuse/Reprint

By ValueNut
June 22, 2001

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!


Interesting words from the leader of Intel: (NewsFactor Network Story)

I call this a positioning statement. Intel is stating a position - the semi market may not get well for the next 6 months (i.e. Intel may continue to do poorly the next 6 months). If Intel decreases production, is it because of the market prices and conditions, or is it because of production problems and low yields or new fab construction (changing to smaller die).

That seemingly arbitrary six-month deadline is very interesting. Why did Intel choose that particular point in time to cut back on future investments? My theory is that December is the projected time that Intel runs out of free cash. Intel ran a cash flow deficit of about $3 billion last quarter; they will probably need to use another $3 billion or so in cash this quarter. If they do that each quarter until December 2001, they will have used another $6 billion in the next six months.

Intel's cash and short-term investments in December of 2000 were $13.473 billion; in March they were $10.058 billion. (See the balance sheet .) Based on Intel's reported guidance, I expect Intel to have around $7 billion in cash now -- we'll know at the quarterly announcement next month. In 6 months, if things don't improve, their cash level is likely to be less than $1 billion -- less than the $1.6 billion AMD has on hand now.

I think Intel management recognizes that the cash that remains from past earnings is a large part of their operational and strategic flexibility. They could begin to take on debt or sell non-core assets at that point, to continue to try to preserve market share, but I suspect they won't take the risk of doing so at that point. I guess I can afford to be a bit smug and say that this announcement is about what I have been expecting Intel to say for the last quarter. There is obviously a limit to how low they will allow earnings to go and how fast they can afford to burn their cash.

Of course my estimates are pretty rough. If Intel has managed to control expenses drastically, stop losses in their communications chip segment and have stopped doing new acquisitions and stock buybacks, then their cash burn rate could moderate. A lot of that is under Intel's control. The problem is that controlling some of these expenses would have negative side effects:

1) Stopping stock buybacks would cause dilution of the stock.
2) Stopping new acquisitions would call into question Intel's commitment to diversifying their business model.
3) Stopping losses in their communications businesses would either require waiting for the telecom market to improve (which might take 6 months to two years depending on who you talk to) or shutting down those businesses at a high short-term cost (read one-time charges to earnings and big losses).
4) Excessive travel and other discretionary costs can be cut with little loss in efficiency, but deeper cuts would hurt morale and efficiency.

Intel's cash burn is obviously a real problem if the economy does not improve. Since AMD is apparently going to be earnings positive and cash flow positive for this quarter, they could play the "price war" game with Intel indefinitely and not lose if the situation remains the same. They could apparently even save up enough cash to buy a new fab in a few years out of current cash flow. By contrast, Intel is bleeding and can only do so for about six months before a change in strategy is required.

Of course, Intel has the Tualitan and Northwood chips coming soon and they are competitive with the Athlon and Athlon 4, but I am fairly sure that will not be enough to halt Intel's drop in net earnings and their negative free cash flow unless the economy does improve soon. The problem is that Intel has to sell enough Tualitans and Northwoods to make up for the loss in high-margin Pentium IIIs. AMD is eating into Pentium III revenues about as fast as Intel is planning to ramp up Tualitan and Northwood. Intel will certainly increase sales of Pentium 4/Northwood and Pentium III/Tualitan, but it will be at the expense of the Pentium III/0.18 micron chips.

Just because Intel is ramping up new chips in the laptop, server and desktop markets does not mean enough of their capacity will be switched to those new parts for AMD's market share to be limited. When you put a 1.4 GHz Athlon system up against a 1.7 GHz Pentium IV, Intel's brand power might make the sale, but if you put a 1.4 GHz Athlon against a more expensive 1.0 GHz Pentium III, the Athlon's better price-performance will more often than not trump Intel's brand power. For that reason, AMD is likely to grow even more in the consumer desktop market than they already have and will even take some of the laptop market. They should have a pretty easy time making dual-processor server sales too. Even with the Tualitan taking part of the mobile market, the Athlon 4 is going to have a comparatively easy time selling against the 800 and 900 MHz Pentium III mobiles that Intel still has to try to sell against the Athlon 4 and FrankenDurons.

Intel will have less than 30% of their unit production on 0.13 micron by year-end. The lower costs and better performance of that 30% may outsell what AMD has in many circumstances, but the other 70% of what Intel makes will be vulnerable to AMD's better price-performance, just as it has been for some time.

If Intel does not find some way to slow down AMD by the end of the year, then AMD is just going to pass Intel by in performance in 2002 when AMD begins shipping their own 2.0 GHz and faster K7 parts. Just after AMD switches to 0.13 micron, the situation reverts back to AMD's advantage.

Of course, AMD could be late in switching to 0.13 micron, but the same could be said for Intel. Even though Intel has begun shipping some 0.13 micron parts about six months before AMD plans to begin shipping, there is no absolute guarantee the Intel ramp will remain smooth. I would say the risks of slippage are equal for both companies. I project both companies to ramp up on schedule unless I hear otherwise, which will make 2002 very interesting.

Even if AMD is six months later than planned in switching to 0.13 micron, AMD will still be able to gain market share, because Intel would have to be able to supply over 70%-80% of the market with 0.13 micron parts to have a chance of taking market share from AMD. That kind of process dominance by Intel isn't likely to happen until late 2003. AMD is taking market share from Intel now because AMD is dropping prices enough to overcome the brand advantage Intel has. For Intel to take market share from AMD, they would have to drop virtually to price-performance parity with AMD. At that kind of price point, the superiority of Intel's brand would let them take sales from AMD with impunity. Intel, IMO, can't afford to do that for at least a year. If they are running breakeven now with negative cash flow, how can they afford 50% price cuts with only part of their product line on 0.13 micron? They need to have the cost advantages of their 300 mm fabs and 0.13 micron running across almost their whole product line to be able to take market share from AMD.

And by the time Intel's 300mm fabs are running in significant quantities (mid- to late- 2002, assuming they are on schedule), AMD will only suffer about a 30% manufacturing cost disadvantage to Intel. That won't be enough to offset the almost two-times area disadvantage the Pentium IV will have versus the K7. Of course, by the end of 2002, the Duron at Austin will be uncompetitive and AMD will have to switch Austin to flash, but that's OK because Dresden will finish the switch to 0.13 micron at the same time.

But aren't I taking a risk in assuming the K7 will reach similar clock speeds as the Pentium IV? Not really. There is no greater risk in assuming the Athlon will run at around 2.0-2.5 GHz next year than assuming the Pentium IV will also reach 3.0 GHz. The engineering difficulties of scaling each architecture are similar.

But won't the Pentium IV Northwood have significantly better instruction-per-clock performance than the Athlon at that point in time? The Northwood probably will have 10%-30% better IPC performance than the current Pentium IV, but the Palomino has 10% better IPC performance than the Thunderbird and the 0.13 micron K7 shrink will allow future Athlon's larger caches and probably more branch-related enhancements as well. I am assuming that AMD will basically preserve the lead it already has by roughly matching improvements in performance that Intel has made. Intel and AMD have improved their technology at roughly equal rates in the last year. I have not heard any arguments to convince me this will change.

But what if the Hammer doesn't arrive on time? So what. The Hammer doesn't help AMD's earnings nearly as much as the 0.13 micron SOI K7 will. I think there is a good chance that the K7 design will be performance-competitive with the Pentium IV family at a much lower die cost at least until the end of 2002 and probably a year or two longer. Just because AMD's roadmap ends at 2002 does not mean the K7 ends as well. If AMD can keep the Athlon scaling efficiently they will, because it gives them the ability to make a profit with better price-performance than the Pentium IV even if all of Intel's processors are manufactured on 300mm and AMD's are not. (If the K7 die remains significantly smaller than the Pentium IV, it will remain cheaper to produce.)

Basically, I don't see much long-term risk for AMD unless they stumble badly. They can easily stay in business against Intel and grow market share gradually to 30% or more until they get their 300mm fab running in 2004 or 2005.

Furthermore, AMD has the potential to seriously disrupt Intel in the meantime. There is some chance that AMD can outsource their high-end Athlon processors to an external fab like Motorola or UMC. If this happens, AMD can increase their market share more quickly than planned. Rather than just take 30% of the processor market, AMD might be able to get 50% and have substantially better earnings as well, and perhaps even within the next 6-12 months. If that were to happen, it would have disastrous effects on Intel.

Since Intel is in the process of upgrading their fabs and that is costing them $7 billion a year, they can't afford to have revenues drop for very long. If AMD manages to displace a great deal of Intel's Pentium III sales with Athlons -- enough to take 50% of the market -- then Intel would have to spend a lot more than $3 billion per quarter out of their cash savings, they might have to spend all the cash they have left just to last another quarter. That's because Intel has to either keep their old fabs running or shut them down while they are upgrading. If their Pentium III revenue dried up before they had switched all their production to 0.13 micron, then they would be forced to shut down fabs and lay people off -- which would cost billions in the short term. The other problem is that Intel is losing almost as much money (about $1.5 billion) in the non-processor part of their business right now as they are making in processors. If processor revenue drops and earnings go to zero, Intel is suddenly losing at least $1-$2 billion per quarter (in GAAP terms).

If that happens, how are they going to finish upgrading their fabs and continue R&D spending to out-develop AMD?

Of course, many of you will point out that AMD would have higher costs out-sourcing processors to an external fab. I don't think that is true. I think AMD can now buy fab space at cost or even less -- even at 300mm 0.13 micron, because of the semiconductor slowdown. UMC is very desperate to get semi manufacturer's to sign on to produce processors at their massive unused 300mm 0.13 micron fabs they are now ramping up. They even have a special program to attract processor manufacturers: Sun is one of their recent wins. Because much of their fab space is unused and uncommitted -- over 50% -- they would certainly be willing IMO to make a long-term deal at near long-term wafer processing costs in return for AMD making a 3-4 year outsourcing agreement. In fact, I think there is a significant chance the deal has already been made based on things that UMC and AMD have both said recently.

If this happens, AMD could have virtually the same manufacturing costs as Intel on 300mm without having to spend a dime to build a Fab and at about the same timeframe as Intel's own 300mm fabs. Look at UMC's recent earnings announcement and judge for yourself whether UMC is desperate enough to make such a deal with AMD. According to this press release, UMC currently has 55% of their total fab space unused and sees no sign of an upturn this year. Last year they were over 100% committed.

Granted, this outsourcing possibility is based on considerable speculation, but even if the likelihood of this much outsourcing occurring is small, the potential improvement in AMD's fortunes could be considerable. AMD could plausibly eliminate Intel as a strong competitor and at the same time greatly improve earnings. If AMD even announces such a deal the stock price could appreciate considerably in a short period of time.

I'm going to keep a certain percentage of my portfolio (at least 20%) in AMD even if it goes back to $40-$50 per share.

Even if this extensive outsourcing doesn't occur, I expect AMD's revenues and earnings to expand at over 30% annually for at least the next two years, assuming the economy improves modestly by the beginning of next year. I just don't see anything stopping AMD from selling more and more Athlons and Athlon 4s each quarter.

I find it very strange that so many people are gloomy about AMD's prospects right now when it seems clear to me the stock price is likely to appreciate considerably in the next year or two, even by conservative measures. I guess I got carried away in my monolog. I should really shut up because I am still waiting for my condo-refinance money to come in. In two weeks I can buy more AMD stock. I sure hope the stock price stays low in the meantime. I think it will; the market will probably remain spooked by all the "doom and gloom" until the next earnings announcement next month.

-ValueNut