Five years ago this month, Advanced Micro Devices Inc. {AMD, News, Boards} paid through the nose to acquire NexGen, a plucky but profitless little microprocessor design house that had been a hot initial public offering just months before. This year, the technology community is abuzz over whether the Sunnyvale, Calif.-based chip maker might do it again, this time with Transmeta Corp. {TMTA, News, Boards}, the stealthy but self-promotional 364-employee designer of low-power microprocessors for notebook computers.
It is a stretch. AMD, which saw its best earnings in history last year, is nowhere near as visibly desperate as it was when it put up $840 million in stock for tiny NexGen in 1995. Then, the microprocessor market's smaller players were under pressure to come up with a chip that could compete with Intel Corp.'s {INTC, News, Boards
Advanced Micro Devices 52-week stock performance
} new, high-performance Pentium line. AMD didn't quite have one yet; NexGen did but it didn't have the manufacturing facilities to meet demand. NexGen was losing money, AMD was losing market share. The merged duo continued to lose both together for years following the marriage, but when AMD rebounded spectacularly in 2000, analysts were quick to point out that the company couldn't have done it without NexGen.
AMD doesn't, by anyone's estimation, need highflying, cash-burning $3.2 billion market capitalization Transmeta right now, and the only confirmed deal the two have made simply involves AMD using the upstart's code-morphing software to simulate its not-yet-manufactured server chips for software development. Code-morphing is a process that shifts much of the work that a microprocessor does from hardware to software code.
That is a far cry from the M-word. But Needham & Co. senior analyst Dan Scovel thinks there is one thing investors can learn from the company's NexGen deal: "AMD is not afraid to buy."
"Everyone told [AMD] they were wasting their money on NexGen," Scovel says. "The joke around the Valley at the time was, if you're a small microprocessor design house in need of an exit strategy, just get bought by AMD. But AMD wouldn't be where it is today if it hadn't bought NexGen. It took a few years for the two companies to really converge, but it worked as a long-term strategy."
Transmeta stock performance since IPO
If the relationship between AMD and Transmeta gets any further than flirtation, it will also be part of a long-term strategy. Transmeta's strength, low-power software-based microprocessors, are tailored to a segment of the market that is still incredibly immature: mobile mini-computers and Internet appliances.
And high-margin server chips, the product that brought the companies together in the first place, are still in drawing-board stages at both companies. AMD will simulate its Sledgehammer server chips, due out in 2002, with Transmeta's code-morphing software and Transmeta, which is working on its first low-power server chips for three server start-ups including Houston-based RXL Technologies, will get a license to use the Sledgehammer design in other chips. Intel, too, has been trying to break into the server market with its beleaguered Itanium chip project, and the AMD-Transmeta tie-up may be aimed at beating the behemoth in the specialized server chip market.
But forays into the server business, successful or not, won't be significant to the three companies' top lines for a few years. What will be significant, for AMD, is whether it can keep doing what it has been doing: chipping away at Intel's microprocessor hegemony. The company has increased its market share to 17% from a low of less than 6% in 1997, and its earnings last quarter more than doubled from the year before, thanks to big gains in the high-end corporate market. No one expects the company to pull off that type of growth during this year's PC slump; in fact, Scovel's forecast of 15% to 16% growth in operating profits is "bullish," he says. But if the company doesn't stumble and continues meeting its short-term goals, that could be enough to send value hunters toward the stock, he says. The stock has rallied about 65% so far this year, but AMD still trades at a measly 10 times expected 2001 earnings, a deep discount to its peers, thanks to its five years as the chip sector's most-notorious underachiever. Scovel thinks that multiple should reach 20 this year, and he has a $50 12-month price target on the stock.
"As history gets older, it counts less," Scovel says. "AMD's growth prospects are a lot better than much of the chip sector, it has a one-and-a-half year record of not making mistakes, and the task at hand for the next few quarters is to just keep its head down, keep meeting its short-term goals, keep its eye on the ball."
AMD doesn't, obviously, need Transmeta to keep its eye on the ball.
"At this point I think any cooperation is more of a symbolic, enemy-of-my-enemy kind of thing," Scovel says.
Mark Feibus, a principal analyst at Scottsdale, Ariz.-based Mercury Research, says Transmeta is overhyping the deal but that cooperation between AMD and Transmeta could be good for AMD's margins.
"The reason [PC makers and Transmeta stakeholders] like Sony {SNE, News, Boards}, Compaq and Quanta {PWR, News, Boards} will back microprocessor start-ups like Transmeta is because it's good for them when Intel and AMD have competition and start waging price wars. It probably makes sense for them to cooperate," Feibus says. "But, I don't really see how AMD would be getting any value by buying Transmeta."
At these levels, AMD shareholders would be better served if the company stayed away from that thought. Transmeta, which guarded its Crusoe project like a state secret until 1999, lost $17 million last quarter on $12.4 million in sales last quarter. Its share price of is about 50% off the post-IPO highs it hit in November but with a market capitalization of $3.2 billion -- compared with AMD's $7.1 billion -- it is still a pricey conquest.
Of course, many analysts doubt Transmeta's ability to support those valuations in a market like this. The company made a big splash last week when it announced its low-power server ventures, but the move left many industry watchers scratching their heads as to why anyone would want to power a server on a watered-down chip like the Crusoe. Similarly, the company's ability to take off even in its original target mini-computer market is hardly a given. While its Crusoe chips only use a fraction of a plain-vanilla microprocessor's power, it comes at a major performance sacrifice. And because the average notebook computer uses only about a quarter of its power on the microprocessor -- the display screen uses up most of the rest -- the chips don't even improve battery life all that much yet.
Quality control, too, is a concern. NEC Corp. {NEC, News, Boards} had to recall some of the first batch of notebooks it shipped using the processors, Sony issued a warning to some customers who bought them, and last month IBM cancelled plans to introduce a Crusoe ThinkPad notebook. Finally, competition will be fierce; AMD, Intel, National Semiconductor Corp. {NSM, News, Boards} and VIA Cyrix have all introduced power-saving "mobile" microprocessors.
"Transmeta's got some intriguing technology, but the market its chips are tailored toward really doesn't exist yet, and it's definitely got an uphill battle ahead," Scovel says. "If the shares fell pretty low AMD might be interested in buying it. But at this point, I think they're more savvy than that."
It is a stretch. AMD, which saw its best earnings in history last year, is nowhere near as visibly desperate as it was when it put up $840 million in stock for tiny NexGen in 1995. Then, the microprocessor market's smaller players were under pressure to come up with a chip that could compete with Intel Corp.'s {INTC, News, Boards
Advanced Micro Devices 52-week stock performance
} new, high-performance Pentium line. AMD didn't quite have one yet; NexGen did but it didn't have the manufacturing facilities to meet demand. NexGen was losing money, AMD was losing market share. The merged duo continued to lose both together for years following the marriage, but when AMD rebounded spectacularly in 2000, analysts were quick to point out that the company couldn't have done it without NexGen.
AMD doesn't, by anyone's estimation, need highflying, cash-burning $3.2 billion market capitalization Transmeta right now, and the only confirmed deal the two have made simply involves AMD using the upstart's code-morphing software to simulate its not-yet-manufactured server chips for software development. Code-morphing is a process that shifts much of the work that a microprocessor does from hardware to software code.
That is a far cry from the M-word. But Needham & Co. senior analyst Dan Scovel thinks there is one thing investors can learn from the company's NexGen deal: "AMD is not afraid to buy."
"Everyone told [AMD] they were wasting their money on NexGen," Scovel says. "The joke around the Valley at the time was, if you're a small microprocessor design house in need of an exit strategy, just get bought by AMD. But AMD wouldn't be where it is today if it hadn't bought NexGen. It took a few years for the two companies to really converge, but it worked as a long-term strategy."
Transmeta stock performance since IPO
If the relationship between AMD and Transmeta gets any further than flirtation, it will also be part of a long-term strategy. Transmeta's strength, low-power software-based microprocessors, are tailored to a segment of the market that is still incredibly immature: mobile mini-computers and Internet appliances.
And high-margin server chips, the product that brought the companies together in the first place, are still in drawing-board stages at both companies. AMD will simulate its Sledgehammer server chips, due out in 2002, with Transmeta's code-morphing software and Transmeta, which is working on its first low-power server chips for three server start-ups including Houston-based RXL Technologies, will get a license to use the Sledgehammer design in other chips. Intel, too, has been trying to break into the server market with its beleaguered Itanium chip project, and the AMD-Transmeta tie-up may be aimed at beating the behemoth in the specialized server chip market.
But forays into the server business, successful or not, won't be significant to the three companies' top lines for a few years. What will be significant, for AMD, is whether it can keep doing what it has been doing: chipping away at Intel's microprocessor hegemony. The company has increased its market share to 17% from a low of less than 6% in 1997, and its earnings last quarter more than doubled from the year before, thanks to big gains in the high-end corporate market. No one expects the company to pull off that type of growth during this year's PC slump; in fact, Scovel's forecast of 15% to 16% growth in operating profits is "bullish," he says. But if the company doesn't stumble and continues meeting its short-term goals, that could be enough to send value hunters toward the stock, he says. The stock has rallied about 65% so far this year, but AMD still trades at a measly 10 times expected 2001 earnings, a deep discount to its peers, thanks to its five years as the chip sector's most-notorious underachiever. Scovel thinks that multiple should reach 20 this year, and he has a $50 12-month price target on the stock.
"As history gets older, it counts less," Scovel says. "AMD's growth prospects are a lot better than much of the chip sector, it has a one-and-a-half year record of not making mistakes, and the task at hand for the next few quarters is to just keep its head down, keep meeting its short-term goals, keep its eye on the ball."
AMD doesn't, obviously, need Transmeta to keep its eye on the ball.
"At this point I think any cooperation is more of a symbolic, enemy-of-my-enemy kind of thing," Scovel says.
Mark Feibus, a principal analyst at Scottsdale, Ariz.-based Mercury Research, says Transmeta is overhyping the deal but that cooperation between AMD and Transmeta could be good for AMD's margins.
"The reason [PC makers and Transmeta stakeholders] like Sony {SNE, News, Boards}, Compaq and Quanta {PWR, News, Boards} will back microprocessor start-ups like Transmeta is because it's good for them when Intel and AMD have competition and start waging price wars. It probably makes sense for them to cooperate," Feibus says. "But, I don't really see how AMD would be getting any value by buying Transmeta."
At these levels, AMD shareholders would be better served if the company stayed away from that thought. Transmeta, which guarded its Crusoe project like a state secret until 1999, lost $17 million last quarter on $12.4 million in sales last quarter. Its share price of is about 50% off the post-IPO highs it hit in November but with a market capitalization of $3.2 billion -- compared with AMD's $7.1 billion -- it is still a pricey conquest.
Of course, many analysts doubt Transmeta's ability to support those valuations in a market like this. The company made a big splash last week when it announced its low-power server ventures, but the move left many industry watchers scratching their heads as to why anyone would want to power a server on a watered-down chip like the Crusoe. Similarly, the company's ability to take off even in its original target mini-computer market is hardly a given. While its Crusoe chips only use a fraction of a plain-vanilla microprocessor's power, it comes at a major performance sacrifice. And because the average notebook computer uses only about a quarter of its power on the microprocessor -- the display screen uses up most of the rest -- the chips don't even improve battery life all that much yet.
Quality control, too, is a concern. NEC Corp. {NEC, News, Boards} had to recall some of the first batch of notebooks it shipped using the processors, Sony issued a warning to some customers who bought them, and last month IBM cancelled plans to introduce a Crusoe ThinkPad notebook. Finally, competition will be fierce; AMD, Intel, National Semiconductor Corp. {NSM, News, Boards} and VIA Cyrix have all introduced power-saving "mobile" microprocessors.
"Transmeta's got some intriguing technology, but the market its chips are tailored toward really doesn't exist yet, and it's definitely got an uphill battle ahead," Scovel says. "If the shares fell pretty low AMD might be interested in buying it. But at this point, I think they're more savvy than that."
