My view has been that AMD has had the party all to itself with its low-power consuming high-performance Opteron server chips--and now the Empire Strikes Back, as was said by Nathan Brookwood, analyst at Insight64 and a top Intel watcher. Intel is primed and ready to rumble.
AMD deserves recognition for having done very well over the past four years in driving a truck through gaps in Intel's product line. Its Opteron server microprocessor has established itself in IT data centers. And it has won some large customers for its PC microprocessors, notably Dell, which had been an stalwart Intel loyalist.
The Opteron chips have been AMD's spearhead into new markets, enabling it to capture as much as 26 per cent of the sever market. Computer data centers often cannot get more electric power. By using low-power consuming hardware based on Opteron, data centers can boost their computing power.
Now with Intel's multicore Core 2 Duo microprocessors coming onto the market, AMD faces the full competitive brunt of a massive competitor that is almost 10 times its size in market cap ($12.5bn versus $112bn).
Intel's chips are fast and low power consuming, and packed with other goodies. And Intel has state-of-the-art fabs all over the world capable of pumping out millions of microprocessors with a low cost of production. It is clear that times are going to get tough very quickly for AMD, it is going to be difficult to hang onto market share-- let alone increase it to 40 per cent of the server market, as AMD has said.
AMD also has the tough task of integrating its $5.4 bn merger with ATI Technologies, the Canadian graphics chip maker. And it has to migrate chip production to the 65nm chip process, a very challenging task and one that is vital to its ability to compete against an Intel already using 65nm.
Chip fabs are the most complex industrial facilities of our times, they are filled with hundreds of expensive, finely tuned machines. Every time a chip maker switches production to a smaller geometry, or to larger wafers, much of the machinery changes and everything has to be recalibrated and retuned for maximum yield.
This is always a risky endeavor because if you cannot get high enough production yields fast enough, that chip fab becomes a massive liability.
Can AMD continue to give its giant competitor a run for its money? I recently visited Henri Richard, AMD's chief of sales and marketing, at AMD headquarters in Sunnyvale. Mr Richard joined AMD in 2002, is an ex-IBM VP, among his many senior positions in the tech industry.
The dapper Mr Richard greeted me warmly and we sat down in leather armchairs in his sunlit office. He's keen to give AMD's position.
"There is a side to this story that no one has written yet," he said. "There is a belief that history will repeat itself, that Intel will recover its markets, and things will go back to being the way they were. But that is not true."
"Things won't go back the way they were because there has been a fundamental shift in markets. Four years ago we were not in the commercial market and we were not in the server market, or the mobile market. Now we are, and that is something that is not going to change."
Fair point. AMD couldn't get a toe or a pinky into the server market, a very lucrative microprocessor market dominated by Intel's Xeon family. And US businesses did not buy AMD PCs or laptops--the commercial PC market was Intel's.
AMD was largely confined to the consumer PC business, a tough business with manufacturers struggling with razor thin margins and always pressing hard for price concessions from suppliers.
Now, four years after Mr Richard joined AMD, it is firmly in three very significant markets. That position won't go away, especially in the server and commercial markets.
Once those conservative, risk-averse IT buyers are comfortable buying non-Intel gear, they will buy more. And they like buying from two competing suppliers when they can--to keep both on their toes.
What about the manufacturing challenges, with Intel's prowess and its pioneering 65nm production process already pumping out chips? At 65nm, Intel potentially can produce more chips per wafer than AMD and that means lower production costs.
Mr Richard claimed that AMD is currently getting better yields and producing chips at a lower cost than Intel despite not yet being at 65nm.
What about AMD's move to 65nm, it will face the same challenges as Intel, that Intel has already faced and figured out.
Mr Richard said he is confident AMD will manage the change to 65nm, because the challenges are better understood if you wait awhile. He confessed he was not a chip manufacturing expert but I could talk with one.
What about the ATI integration challenges? Mergers are risky, and a huge distraction to management. This is a distraction that AMD doesn't need right now, with Intel pulling out all the stops to stop AMD.
Mr Richard said that AMD and ATI know each other very well. "Culture is always the stumbling block but we know ATI, we have been working together for four years and we meet with them regularly. Our businesses do not overlap by very much at all. It is mostly in my group that the integration has to take place. In terms of sales and marketing, this means that I now have more resources."
He pointed out that merging with ATI, AMD is a company that has its own fabs, and is also one of the biggest fabless companies--using fab services from TSMC, the largest chip foundry.
He said that there was no change in ATI's relationship with TSMC. However, I could see how it could be advantageous for AMD in the future, if it could mix and match production. This could keep its own fabs running at full steam, should it have a drop in demand. In the chip business it is essential to keep the hugely expensive fab production lines filled otherwise you will lose your margins very rapidly.
The ATI deal also brings very strategic customer accounts to AMD. Mr Henry said that ATI brings Microsoft and the Xbox, and it also brings Nokia.
"We believe that most people in the world will access the Internet from non-traditional computers. Whether it is a thin client or a fat cell-phone, I don't know -- but companies like Nokia will be important. The PC was designed by engineers to solve equations, but companies such as Nokia understand consumers and they know how to make the user interface easy to use."
Margins in the consumer business are not that good, and the average price of the silicon in cell phones is a lot less than in PCs. Plus the cell phone business won't allow itself to be PC-ized in the way that Microsoft and Intel managed to commoditise the PC business and aggregate much of the value-add within their products--leaving thin margins for the PC makers.
Mr Henry agreed, saying: "We know very well where the margins are and that is in server and commercial markets. That is our focus"
I asked about AMD's antitrust suit against Intel filed more than a year ago, which has already revealed some salacious tidbits of how the PC industry does business. It shows how PC makers sometimes say they will buy from AMD to get price concessions from Intel.
"Intel's slide is not over. The discovery process is going to expose some very unethical behavior by Intel."
Antitrust legal actions are long and very expensive, especially the discovery process. I'm not sure what AMD has to gain by continuing with the suit.
AMD had prepared the suit to help it gain entrance into key markets. Now that it has through other means, the suit has lost its original purpose, yet it remains a big distraction for management, and a drain on profits.
As we wound up the interview, Mr Richard took an opportunity to repeat the message that there has been a fundamental change in AMD's business, that the markets have changed.
"This time, the chess board is different, it is larger and we have more powerful pieces in the game," he said.
What if AMD is being used as a pawn against Intel by other companies?
"That could be," he said. "But now we are like the pawn that got to the other side of the chess board and has been converted to a Queen."
- - - - -- --
Please also see related: