FOOL CONFERENCE CALL SYNOPSIS*
By Debora Tidwell (MF Debit)

Advanced Micro Devices
<% if gsSubBrand = "aolsnapshot" then Response.Write("(NYSE: AMD)") else Response.Write("(NYSE: AMD)") end if %>
One AMD Place
Sunnyvale, CA 94088-3453
(408) 732-2400
http://www.amd.com/

UNION CITY, CA (April 10, 1997)/FOOLWIRE/ --- Advanced Micro Devices (AMD) released first quarter results after the market close on Monday, April 7th. Included in the press release the company issued (which they did not read during the conference call) is both a GAAP and a non-GAAP P&L as well as a balance sheet and a breakout of sales by group that has precise numbers for each of those groups in terms of sales and some other selected financial data. That data includes depreciation and capital spending. You can access the press release from AMD's web site.

RESULTS OVERVIEW. AMD is very pleased to have returned to profitability without the benefit of the K6 in volume yet. At last quarter's call, they said it would take about $550 million in sales for them to break even. While their sales were there and their expenses were slightly lower, this allowed them to make a profit and they are very pleased with that. Sales were up 11% quarter-to-quarter and was led by flash which was up 20%. K5s were also up about 20% and their PLD division was up about 17% quarter-to-quarter. The company's book-to-bill was substantially greater than 1-to-1 for the quarter and they are looking at a marketplace today where all 3 of their growth engines are seeing good demand. With regard to the book-to-bill, all groups were greater than 1-to-1. On the cost side, gross margin improved about 8% on the GAAP basis and about 7% on the non-GAAP basis. Again, the non-GAAP statements include the FASL joint venture as if their 50% ownership were included in the cost of sales. The joint venture income was down slightly quarter-to-quarter as they lowered the transfer price from FASL to AMD. The offset to this is a reduction in the AMD cost per sales line. This is exactly why they like to combine the joint venture results with their cost per sales so that any change in the transfer price doesn't distort either the joint venture results or AMD's gross margin.

EXPENSE CATEGORIES. Also on the P&L, the combination of R&D and SG&A expenses were 36% in the quarter, which is a decline of 3% from the prior quarter. The decline as a percentage of their business is more from the fact that their revenue grew faster than their marketing and G&A expenses did. Marketing and G&A actually did increase, and there is some bonus and compensation accruals in there too, including profit sharing. Marketing cost will be higher in Q2. The rate at which they grow their revenues will determine what those percentages are and they aren't going to make any estimates there. They were asked what the advertising budget would be and responded that they are of the opinion that somewhere in the range of 1-2% of total revenues is probably the right number for an advertising budget. They continue to spend heavily in R&D both for product development and process technology. Again, however, R&D as a percentage of sales went down because their sales grew by more than 11% and they didn't grow their R&D spending by 11%. R&D was roughly flat quarter-to-quarter in dollar terms. One other noteworthy item was that the tax rate in the quarter was 29% and they currently believe that will hold for Q2. The tax rate on FASL going forward should be closer to 50%. The tax rate is going to go up because they have adjusted the transfer price, so the residual profits that are left in FASL are going to have to come more and more under the scrutiny of the Japanese tax rate which is in the 50% ball park. The number of weighted average shares was up about 10 million quarter-to-quarter. This is caused by going from a loss to a profit. Outstanding options become dilutive only when you are profitable. Therefore, you should expect the shares to continue at this level going forward.

COST OF SALES. Cost of sales and the combination of cost of sales and FASL results were only up very slightly quarter to quarter. In Q4 they had some rip costs, so actually the costs were up more than just depreciation quarter to quarter, it was offsetting the rip costs that they had in Q4. Going forward, in the last two quarters they have had significant fall-through, meaning that for every dollar that sales went up, a very high percentage of it dropped to the operating income line. That percentage will not hold in Q2 because they are going to be ramping up the K6. The percentage fall through will come down and that is because of the costs associated with ramping up the K6. The percentage that falls through will depend on the amount of the sales growth. The costs for Fab 25 are already in the P&L but what you can't assume is that just because they are utilizing Fab 25 more and more that a very high percentage of those sales will drop through. They have incremental packaging costs and other costs associated with the K6 that will hit the P&L. But, with regard to Fab 25, those costs are already in there.

BALANCE SHEET HIGHLIGHTS. On the balance sheet, cash was $629 million, which is up about $240 million quarter-to-quarter and is the net effect of three very different things. One, they took down the $250 million term loan from the bank early in Q1. Two, they received a tax refund of $101 million in the quarter. This is from the loss carryback to prior years. They were able to file the return and get the refund all within 90 days. Third, capital expenditures were $174 million in the quarter, which would have resulted in a drain of approximately $100 million in cash. The offset to the tax refund is shown in the prepaid expense category on the balance sheet. Other than the cash and the prepaid expense there really isn't much to note on the balance sheet.

CAPITAL EXPENDITURES AND DRESDEN. They break out the capital expenditures for non-Dresden. They do that because the financing for the Dresden fab they currently have under construction is financed separately. The non-Dresden capital expenditures in 1997 they anticipate will be in the $450-500 million range. On top of that, whatever they happen to spend on Dresden needs to be added, which could be another $200 million. They have yet to make any commentary on what they expect capital expenditures to be in 1998. It will be driven largely by the expenditures associated with Dresden.

CAPACITY STRATEGY. They hope to complete the Dresden facility about a year from now, start installing equipment, and have their initial production out probably in the first quarter of 1999. They have an enormous mega-fab in Austin, Texas -- Fab 25. Fab 25 when fully outfitted would have a capacity, if they were making K5s, of about 6000 wafers a week (300,000 wafers per year). That's premier metal. The K6 is a more complex process, 5 layers of metal, so a better estimate there is 250,000 wafers per year. What they expect to do is ramp the K6 as they transition to this technology such that they will reach this maximum capacity at the end of 1998, just in time for additional capacity to come on from their Dresden fab. In the recent Dataquest announcement, Dataquest expects that NIMs (non-Intel microprocessors) will have 25% of the unit volume of PC processors by the year 2000. AMD wants them all.

PLD PRODUCT LINE. Sales were up sharply, 17% in the quarter and this growth reflects the focus they have been talking about over the last nine months utilizing their separate sales and marketing force. Again, the story for the future is the CPLD, the MACH product family, where the growth is. Small PALs were up during the quarter. They are gaining market share in this marketplace. MACH shipments were higher than small PALs during the quarter for the second consecutive quarter and were up 40% year-over-year compared to Q1 1996, with record bookings, up 40% year-over-year compared to Q1 1996. About 60% of the PLD business goes to the broad category of communications in both the public and private infrastructure.

NON-VOLATILE MEMORY GROUP (EPROMs AND FLASH). Sales were up by 14% led by flash sales which were up 20%. EPROM shipments were, therefore, down. It's a very competitive market and a declining market that AMD looks upon as a target of opportunity. Flash shipments continued to grow past the $150 million level, led by the 2.7V product that was introduced a year ago for portables and cellular phones. Unit shipments were also a record and bookings were close to record levels with their 2.7V family having broad market acceptance and ramping rapidly. The unit growth was greater than 30% quarter-to-quarter and they estimate that the bit growth for flash would be another 20% higher than that because of the mix shift to higher density products. They are going to start monitoring bit growth for this category. This is not something they have traditionally done; it's sort of a DRAM thing, but they will have it for future meetings. Flash pricing is coming down somewhat faster than learning curve. FASL is leading the price down. They have a big factory and a conversion going on to double their capacity so pricing is coming down somewhat faster than the learning curve. They introduced and shipped in volume their new 2.2V flash product for cellular phones. That is the next evolution in lowering the voltage on the way to 1.8V and lower soon. Their 0.35 micron conversion in FASL is proceeding smoothly. They are leading the prices in flash, they have the smallest die, they have the capacity, so the FASL ramp is going to be determined by their success in the marketplace. They are almost at the maximum wafer rate and the increase in capacity in 1997 is going to come as a result of the conversion to 0.35 micron, which will essentially double the capacity of that facility by the end of the year. It is pretty much uniform throughout the year.

COMMUNICATIONS AREA. The local area network product sales were sharply higher, led by products for private infrastructure as demand for intranet infrastructure continues strong. They are very encouraged in Q1 by the strong demand for their hub products in the infrastructure within networking. They suggested not to misinterpret the fact that stock prices are going down for network companies with the fundamental health of the business. AMD sees a very strong demand within the networking market opportunities and they don't see that abating. Their public infrastructure shipments improved, let by their digital cordless offering for 900 MHz cordless phones in the US. The design wins that they have had for the US 900 MHz phone encompass about 70% of the market for digital cordless phones. To compare prices of digital phones versus analog phones, the analog phones are older and sell at approximately the $80 price points. The digital narrow-band phones which are the devices they are presently delivering and which is the future for digital cordless, will be at the $100 price point by Father's Day as this product category expands the market in the United States.

COMPUTER PRODUCTS GROUP. Sales were higher on a richer mix. Total x86 unit volume was 2.6 million units in Q1 versus 2.9 million in Q4. Unit volume of the K5 improved to the 1.3 million unit level from 1.1 million units in Q4 on a much improved mix, mainly on PR-133s and PR-166s, versus in Q4 where they were shipping mainly PR-133s and 100s. Actually, 60% of their unit volume in Q1 was PR-133s and above compared to 23% in Q4 of 133 and above. So their average selling price went up by about 20% during the quarter. Fourth generation and earlier unit volume dropped from 1.8 million units to 1.3 million units. During the quarter they announced their first chip set, the AM640, their first offering toward the objective of keeping the Socket 7 infrastructure competitive as they go forward. They do not expect K5 revenues to grow this quarter, but don't guarantee that assumption. Right now they are doing everything they can to shift production to the K6, so they will not have a lot of K5s available if they are successful in making that switch, which they expect to be. Since competition is good and they have seen more aggressive pricing in the marketplace, they wouldn't expect they would see an increase in K5 revenues. They are planning to be out of K5s by the end of Q2. They want to be a 100% MMX company by Q3 in terms of units shipped.

K5 AND EARLIER REVENUES UP. They had excellent execution in the quarter. On the strength of an enriched mix, their K5 and earlier generation x86 profits and revenues doubled since the trough of Q2 1996 and were up 50% year to year. In addition, they shipped a few million dollars of K6/MMX processors in the last weeks of March. K5's demonstrated compatibility has been the key to persuading customers that AMD is more than a clone supplier and paved the way for adoption of the K6. Users of K5s are highly likely to use K6s. AMD knows of no K5 user who has ruled out using the K6.

K6 OEMS. The number of announced K6 OEMs is now in the low teens. Two of the announced users, Fujitsu ICL at #7 and Vobis at #15 were in the top 20 last year. Those companies join Everex, FIC, and Leo in Asia. There are some small companies in Europe. Troton in Japan has product in the retail channel and AMD understands that the reception is very good. Cybermax, Everex, and Techmedia are among the OEMs in the US. Announcements greater in number than those already made are expected this quarter, including 3 more in the top 20. They will be very disappointed if one of the 3 isn't a top 10 company. When the deals are announced, the companies are expected to have product available for immediate delivery. AMD will continue to honor their customers' requests to not pre-announce their products. They are pleased with their progress in developing their customer base. AMD continues to expect to ship hundreds of thousands of K6/MMX processors in the current quarter and a million units in the second half of the year. While they expect to grow their non-processor business in the current quarter, overall quarterly growth will be determined largely by their success with the K6. They expect to be successful.

SELLING THE K6 IN GERMANY. Relative to Germany, where Intel was granted a trademark on MMX, AMD is marketing the product as a "K6 processor with MMX Technology," so this is having no impact on their ability to sell in Germany. It just doesn't let them use the same logo that they are using in the US. AMD is contesting the fact that Intel has the trademark in Germany. Intel was denied the trademark in the United Kingdom, so in the UK AMD is using the K6/MMX processor logo.

PROCESS STRATEGY FOR THE K6. They are in production now with the CS34-EX technology. That is the process of choice for the K6 at this time. It is a 5-layer metal process. It's 0.35 micron. It has local interconnect. It has shallow trench isolation and it has the C4 flip chip packaging technology. That is what AMD is ramping in production today. This particular technology has enabled them to build 233 MHz parts which they have introduced. They demonstrated 266 MHz parts on that technology at their launch. The next technology they will be using is the 0.25 micron technology which they call CS44-E. That technology is currently being qualified. They have already seen the output of this technology in their sub-micron development center. They expect to have it qualified by mid-year and be in production in the second half of the year in Austin, Texas. It is that technology that they expect will give them up to 300 MHz or even faster parts at year end. The reason they have not yet introduced a mobile product is simply because they are targeting the desktop to establish themselves, as part of their marketing plan. They will be able to produce mobile products from the existing process technologies. As they go to 0.25 micron, they will see a further benefit in both performance and even lower power dissipation (although the current product burns less than 10 watts).

THE PC MARKET. AMD's founder and CEO, Jerry Sanders, was asked to comment on the PC market in general given alleged reports that sales were down 8% sequentially for the industry overall in the March quarter. Sanders responded that they feel pretty good about the PC market and reminded people that some 60% of the PCs sold are sold outside the United States and, with all due respect to the analyst group that follows the industry, Sanders commented that they seem to be preoccupied with US retail sales which are somewhat softer. AMD expects a healthy market and thinks there are going to be 85 million PCs sold this year (for x86 PCs, excluding Apple). They were asked, given 20% growth on the 85 million PCs number this year for next year of 17 million additional units, what they saw as reasonable penetration targets. The company said that they would like it all. It turns out that they are shipping 10 million units now. The 10 million units they shipped last year and the last 12 months are going to be replaced -- the old 486, the x5, and the K5 are going to be replaced. In 1992 AMD got something like 80-90% of the incremental unit volume in the market. So, it would really be unprecedented for them to get a number of units not dissimilar to the incremental increase.

MAJOR COMPETITIVE DIFFERENCE BETWEEN K6 AND K5. An analyst commented that the Winstone benchmark AMD published on the K5 166 MHz, it compares fairly well to the Intel 166 MHz Pentium. The company was asked what they see as being the major difference this time that is going to allow the K6 to take market share where they K5 hasn't, given that they have comparable performance versus the target market they are addressing with the part. Sanders responded that it is the time that they are in the market. First of all, the K6 for the first time is ahead of Intel -- AMD has an MMX solution and plans to be MMX to the masses. In the retail channels they are being very well received. As a matter of fact, the K6 was named on Friday at the Retail Vision Gala Awards as the Best New Product for Hardware. AMD thinks that they are bringing a new technology, MMX, to the mass markets at very affordable prices. They think this will give them their initial fingernails into the side of the cliff that they can climb up on. They offer a complete selection of products, starting with the K5 PR-166 then going to the K6 166, 200, and 233. They offer a complete lineup and indications are that people have been looking for this for a long time. Moreover, AMD right now has the highest performance x86 machine in the world. The K5 PR-166 would have driven substantial marketshare had they had it a year ago. Today, it's just an entry-level product. So, they think there's a big difference -- it's when you have the product. That's what they see the difference being.

BEYOND MMX. The company was asked about more visual-oriented computing standards going forward, beyond MMX. The company responded that they think that there is no question that the visual computing platform is the future. They believe that the chipset that they have now, the AM640 chipset, for the K6 in the Socket 7 infrastructure is going to be succeeded quickly by the 640AGP and they will be supporting AGP with the 6th generation processor in the Socket 7. As they look at it, they think that through 1998 the opportunity for K6s and their progeny with an AGP-enabled chipset is going to be the platform of choice in the consumer market.

SOCKET STRATEGY GOING FORWARD. The company was asked about new packaging schemes and compatibility with new motherboard standards. They replied that it is an evolving situation. Their expectation is that they can serve the mass market with Socket 7. It's lower cost to implement, there's more competitors providing solutions, and they believe that's going to be an excellent solution through 1998. As they come into the end of 1998 when they expect to begin to sample the K7, they will obviously have a different infrastructure for the K7. They are playing their cards pretty close to the vest at this time and will tell people more about that platform as time evolves.

COMMENTARY. One analyst from Merrill Lynch started his question by congratulating Jerry Sanders on getting more mileage in the stock market out of $0.09 than any other high tech company this quarter. Sanders responded, "Well, let me appreciate the fact that you got more mileage out of predicting a $180 billion market cap for a company that is a good competitor, but not infallible."

* A Fool conference call synopsis represents an effort to highlight the salient points of a conference call and should not be taken as an authoritative accounting or transcription of the entire event. Note: Statements made by a company other than historical information may constitute forward-looking statements for which the company can claim protection under the Safe Harbor Act. Please consult the company's filings with the SEC for information on risk factors which might cause actual results to differ materially from the information contained in these forward-looking statements.