Merrill Lynch Comment United States Semiconductors 17 June 2004 Joseph Osha Advanced Micro Devices We chat with Hector NEUTRAL Volatility Risk: HIGH Reason for Report: Company Update
Estimates (Dec) 2003A 2004E 2005E EPS: -$0.79 $0.64 $0.69 P/E: NM 23.8x 22.1x GAAP EPS: -$0.79 $0.64 $0.69 GAAP P/E: NM 23.8x 22.1x EPS Change (YoY): NM 7.8% Consensus EPS: $0.60 $0.73 (First Call: 21-May-2004) Q2 EPS (Jun): -$0.40 $0.11 Cash Flow/Share: $0.66 $3.13 $4.27 Price/Cash Flow: 23.1x 4.9x 3.6x Dividend Rate: Nil Nil Nil Dividend Yield: Nil Nil Nil Opinion & Financial Data Investment Opinion: C-2-9 Mkt. Value / Shares Outstanding (mn): $6,370.3 / 418 Book Value/Share (Mar-2004): $5.86 Price/Book Ratio: 2.6x ROE 2004E Average: 10.2% Total Debt / Capital: 0.5% Est. 5 Year EPS Growth: 15.0% Est. 5 Year Dividend Growth: NA Stock Data 52-Week Range: $18.50-$5.80 Symbol / Exchange: AMD / New York Institutional Ownership-Vickers: 67.3% Brokers Covering (First Call): 29
Highlights: • We spoke with Hector Ruiz, AMD’s President and CEO, and discussed AMD’s product roadmap, focusing on Athlon64’s prospects as well as the transition to the 90nm manufacturing process.
• AMD has hopes for 64-bit computing on the desktop, but doesn’t seem to have become more aggressive in that space despite its success in the server market. We believe that AMD understands how difficult it has become to get paid for performance in the desktop MPU market.
• In the notebook market, AMD has in fact become more aggressive, pulling a number of products in to the first half of 2005. Comments from Hector indicate that AMD is going to focus hard on using 90nm manufacturing to become competitive in the notebook market again.
• Interestingly, AMD acknowledged that there’s a possibility that we could see the company sign a foundry deal for microprocessor manufacturing. AMD’s own 300mm fab is scheduled to come on line in 2006, but it appears that AMD is looking for some 300mm capacity prior to that. Given Intel’s 300mm plans, AMD needs a deal.
• Dual-core processors have appeared on AMD’s roadmap, scheduled for release in the second half of 2005. That’s not a big surprise, but it does confirm that AMD is feeling the same pressures that forced a change to Intel’s roadmap.
• We’re still Neutral on AMD’s stock, although a successful outsourcing deal would be a major positive in our opinion. AMD’s manufacturing operations are not sufficient to allow the company to capitalize on important opportunities.
We spend some time with AMD . . . We spoke with Hector Ruiz, AMD’s President and CEO, and discussed AMD’s product roadmap, focusing on Athlon64’s prospects as well as the transition to the 90nm manufacturing process. We didn’t talk much about the Opteron server business, simply because we regard AMD’s story in that business as compelling and obvious.
Taking 64-bit technology to the desktop, cautiously AMD hopes to see its 64-bit product line achieve the same success in the desktop as the Opteron processor has enjoyed in the server market. Hector claims that customers are interested in performance, not process geometry or clock speeds, and Intel’s recent move away from quoting clock speeds to an AMD-style performance number validates Hector’s comments. It’s more difficult to independently verify AMD’s claims that all of AMD’s desktop parts outperform Intel parts at comparable price points. The bigger issue of whether customers are interested in 64- bit computing simply because it’s 64-bit remains tough for AMD to address. We don’t think that 64-bit computing becomes a significant differentiator on the desktop until Microsoft ships the Longhorn version of Windows in late 2005 or early 2006. Until then we believe that the Athlon64 will compete with Intel on the basis of its merits as a 32-bit processor, nothing more. We note that AMD doesn’t appear to be getting more aggressive with its 90nm desktop plans – if anything, the tone of the comments from Hector suggests that AMD will try to keep the older Athlon XP and 0.13 Athlon 64 parts alive as long as possible. Given the difficulties that both AMD and Intel have had getting paid for performance in the desktop market, we think that Hector’s thought process makes sense.
90nm parts still on target for Q3 shipment AMD started 90nm production in May 2004, and still expects to ship products in Q3, with a volume ramp in Q4 of this year and Q1 next year. Hector expects the transition to 90nm to complete within by mid-2005, and insists that there will be no disruption to production at Fab 30. Given the number of transitions AMD is attempting to execute at the same time, Hector’s claims might raise eyebrows, but we do note that execution at Fab 30 has been extremely consistent to date. AMD could realize real financial benefits when it ramps 90nm parts in volume, although we expect some gross margin hit initially.
More aggressive notebook roadmap While AMD has yet to show much success in the notebook market with its 64-bit products, the company’s plans are becoming more aggressive. We keep track of the changes to AMD’s product plans, and according to the most recent 90nm have been pulled in by six months to the first half of 2005. AMD seems to have realized the importance of attacking Intel in the profitable notebook market. AMD has some interesting potential advantages in the notebook PC market. We believe that the Athlon64 architecture will not encounter the same level of power dissipation problems that Intel’s Netburst core did when Intel moved to 90nm with the Prescott (desktop) version of P4. That may make it easier to move the Athlon64 into the mobile market on 90nm. AMD’s silicon-on-insulator manufacturing may help as well. It is important to note, though, that Intel has invested heavily in making Banias and Dothan mobile processors as power-efficient as possible, while AMD is only just beginning the process of re-entering the notebook market. We cannot yet tell whether AMD will be able to match Intel’s Dothan at 90nm.
Dual-core processors become official Another interesting change to AMD’s roadmap was the appearance of a dual core designation alongside several desktop and server products scheduled for release in the second half of 2005. It’s not a huge surprise – we had known that AMD planned to do multiple cores – but the fact that AMD is now acknowledging the timing is interesting. Intel’s hot Netburst core forced Intel to multiple cores more quickly than Intel has planned, but it’s clear that AMD does not plan to stick with scaling single processor cores either.
Fab 36 on schedule; AMD raises the possibility of an outsourcing deal The construction of AMD’s Fab 36, the company’s first 300mm wafer fab, continues on schedule. Fab equipment should begin to install in Q4 this year, and commercial wafer shipments should begin in early 2006. AMD plans to provide self-financing of more than $700 million with cash flow from operations. Interestingly, Hector suggested that there is a reasonable possibility that AMD will try to sign a deal for outsourced microprocessor manufacturing prior to the ramp of Fab 36. We had assumed that an outsourcing deal was off the table following AMD’s decision to build Fab 36, and the company had not said much about outsourcing recently. We don’t know who AMD is talking to, although IBM is the most likely and most logical candidate. The fact that AMD is talking about outsourcing again would seem to reflect AMD’s understanding that it can’t wait until 2006 for 300mm manufacturing.
Flash market still nasty As was the case with other recent checks we’ve done, AMD continues to acknowledge that pricing in the NOR flash market is tough, primarily due to Intel but also to Samsung and STMicro. Hector insists that AMD has not lost market share to Intel, although a check of the relative rates of growth for the two company’s flash businesses in Q2 suggests otherwise. The key for AMD at this point is simply to try in stay in front of the price declines. AMD did confirm that half of its flash business is now on 110nm, and expects to begin ramping 90nm by the end of the year. Chart 1: AMD’s Product Roadmap (June 2004) Source: Company information
Recommendation & Investment Thesis AMD has shown significant success with Opteron, and now it’s important for AMD to have a competitive product with Athlon64 once the part moves to 90nm manufacturing process. We think AMD can realize real financial benefits by moving to 90nm, but it’s unlikely that will be achieved before Q4 2004. Given the flash memory and manufacturing issues, we think investors would do best to stay on the sidelines until AMD finishes the next two quarters. Our recommendation remains Neutral.