Message Font: Serif | Sans-Serif
 
No. of Recommendations: 8
I guess I'm the first to post about this Duel, which surprises me. First, I voted that both arguments were excellent. Intel has some rough water ahead, Intel seems to have a long term plan to deal with it, but the company tends to get distracted by irrelevancies.

Of course, the big irrelevancy is the perceived competition with AMD. Yes, it is nice to have the bragging rights for the fastest processor, and yes, AMD has moved from the low-end and value markets to concentrating on the high-end desktop and performance markets. All this is very good for AMD, but it has little relevance for Intel. AMD can grow production faster than the PC market as a whole is growing, while Intel continues to be capacity limited and unable to fill demand. My guess is that at the current rates of growth, it will be 2005 or later before Intel or AMD will have to worry about overproduction in the microprocessor market. (And both companies are heavily into flash, where the demand is growing even faster.)

What Intel does have to worry about, and worry big time right now, is that Intel lost control of the direction of the PC market. No that is not right, they signed it away to Rambus, and then Rambus through their hamfisted arrogance gave it away to AMD.

Intel can live happily in a world in which AMD is the the technology leader or one of several technology leaders. But that phrase, technology leader, doesn't mean what most people think it does. The technology leader is the company or organization that brings together the various segments of an industry behind a coherent vision of the future. AMD clearly was willing and able to take the lead in bringing DDR DIMMs to the PC and server market. But technically, the first DDR chipset released was for the Pentium III, even though AMD DDR systems are shipping first.

But what should Intel do about this? Come out with a new proprietary standard? IBM's Microchannel is just one of many examples of how well that doesn't work in this industry. (And Itanium and RDRAM may soon be two others.) The right approach, if Intel wants to be seen as a technology leader again, is to look beyond tomorrow, and see a need, then prepare a solution and an infrastructure to meet that need. Do this right, and others will buy in to your solution. Do it wrong, and even if you have the best solution, the companies you need as partners will go elsewhere.

SDRAM is in the process of being succeeded by DDR memory. Everyone other than a few diehards on the Rambus board can see this. (And for the purpose of this discussion, whether or not Rambus collects royalties on DDR memory is irrelevant.) Rambus and Intel can try their hardest to keep RDRAM competitive with DDR, but that is only a way to bleed the company's treasury dry.

The right solution is to realize that in a few years, DDR400 (PC3200) will be running out of steam. What will the successor be? If Rambus were to develop a workable quad-clocked version of DDR, or even better a technology with a potential 10 or 50 Gigabyte per second upside, then the PC industry would beat a path to their door. (Although there may already be too many hard feelings for some companies to be willing to be led by Rambus.)

There are lots of other opportunities to take leadership roles in the PC industry, and as I said, Intel can, if they feel like it, leave the leadership to others and try to innovate elsewhere. The only thing they can't successfully do is to pretend to be the lead gorilla, while not taking the leadership job seriously.

For the record, I have been around this industry for a long time. I have seen many companies fail in exactly that way. IBM has done it not once in the PC industry, but several times. To take one of the less obvious examples, when IBM and Microsoft split over the future of OS/2, IBM should have made one of two choices. They could have written OS/2 off, or they could have gone out aggressively to get other PC manufacturers on board, and not only taken ownership of the marketing rights, but put enough of IBM's good software people into bringing out a quality product quickly.

Instead IBM allowed Bill Gates to get other companies to buy into his roadmap, and look at where we are now.

So what do I think Intel is doing wrong? They should gracefully hand IA-64 over to HP, or just abandon it. If necessary they should license x86-64 from AMD, I'm sure AMD would be happy to welcome Intel aboard. (I wouldn't expect any anti-trust problems, neither AMD or Intel holds a dominant position in the 64-bit server market.) If Intel is unwilling to do that, why not license PA-RISC from HP, or even the Alpha architecture. At this point the world does not need a new incompatible 64-bit architecture, especially one that Intel is having trouble getting right. And Intel doesn't need to waste their best developers on a product going nowhere.

Next, Intel needs to get out from under Rambus's thumb. If the cost is buying Rambus, it is cheap at the price. Or to phrase it another way, the Rambus deal has cost Intel more in market cap than Rambus would sell for, even if there were other bidders. And if someone else really wanted the company, Intel could gracefully bow out as long as the other suitor was willing to terminate the Rambus/Intel contract. And if Intel did succeed in buying Rambus, they might actually collect enough in royalties on Rambus IP to justify the deal.

Finally, in the processor business, Intel has to decide which market segments they want to be in, and which they don't. Intel cannot make enough processors to meet all the demands. Once upon a time it was important for Intel to have products in every market segment so that no competitor could get established by being where Intel wasn't. Intel blew this with the i820 Camino chipset, and it can never again own all the market segments. Via got a major share of Intel's chipset business, and AMD took over the value segment with the K6/2 long enough to get out of the red, and to get the Athlon established in the high-end. But the real problem was that Intel could no longer match demand for PCs, and now the market as a whole is growing much faster than Intel can.

What is Intel's answer? To sell a product for which there is no requirement! The Pentium 4 at introduction will just not be a viable product in the high-end and performance markets. Later versions of the Pentium 4 will probably be very competitive in the workstation, server, high-end desktop, and performance segments. Maybe in 2002 they can even make a mid-range P4 product.

But right now, the Pentium 4 is aimed sqarely between four existing market segments, while not matching the requirements for any of them. I blame this entirely on the terms of the Rambus deal. The Willamette can be competitive in some of these market segments, and I suspect it could be a formidible opponent in the workstation and server areas. But not with the current chipset and memory architecture.

So that is my read. An Intel with fire in its belly can be the stock market darling of old. But if Intel wants to be a market leader, it has to show leadership. If Intel wants to be a team player, it has to show a willingness to cooperate. But most important, if Intel wants to have a future, it has to have a plan for getting there. If they have one right now, it isn't obvious to me or to many other observers.

This year will be very good, and next year may be better. But unless someone steps up to the plate, the company will need to sell off lots of unprofitable pieces in 2003 or 2004. Which pieces? I don't know. But I do know that Intel is trying to diversify and be all things to all potential customers, while trying to enter market occupied by very lean and focused companies.

Intel is a large company, and can afford to be in more markets than say AMD. But like AMD, three of those markets have to be microprocessors, chipsets, and flash memory. My guess is that the old five plus or minus two span of control rule applies. I don't know how many markets Intel can afford to be in, but I do know that if top management is not constantly aware of what is going on in all those markets, the company is in trouble. The last year or so is an excellent example of the potential problems. (Um, make that an excellent example of the current problems.)

It all started when the 820 chipset was not going to meet schedule and performance requirements. This problem seemed to have no management visibility until the product was shipped late, and with one RIMM slot capped. At this point management seemed to wake up, and Solano the i815 chipset was rushed to completion. But the Solano, even though it is a nice low-end chipset, did not fill the requirements of other market segments. The effort required to make a chipset that addressed the needs of other market segments that the 820 was supposed to address was next to nothing. But it would have required top management make the commitment. But by then, management was trying to deal with a shortage of Coppermine chips by reopening Katmai production. And while that was going on, no one was at home to deal with the issue of Itanium prototype chip speeds, and, oh yes, someone should have talked to potential customers about Timna, and what could be done to make it a salable part.

I could go on, but you get the picture. Remember, those guys on Ed Sullivan who would spin plates balanced on sticks? Eventually they would have too many plates spinning, and while they were trying to get one back up to speed, two others would fall down. That is what Intel management is doing now. The are reacting to crises, rather than anticipating problems.

I'll finish with a different example, from AMD. AMD recently--as in this week--announced it was dropping the Mustang processor which was expected to debut this month. The reason for cancelling it was not that it didn't fill a market need, but that an innovative new Northbridge design from Micron would render it obsolete within a few months of introduction. So AMD figured that they could get a better ROI if they didn't build Mustang. Zap, its gone, and the Palomino is the new flagship of what was the Mustang family.

If AMD had cancelled the Mustang earlier, it would have saved more money. Not much, since most of the design work is also part of four other chips. And we don't know how long AMD's management may have known about Micron's plans before they were announced. But based on other clues, AMD's management reacted to this new development within a few days of AMD becoming aware of the issue. And AMD made the decision that so many managers have trouble with: The money spent is gone, what is the best decision now? (I also used to know some poor poker players with the same problem. Then they got smart and stopped playing poker.) And they evidently made that decision pretty quickly--my guess is that the only delay was to negotiate with Micron, and to get further schedule information. (And unless I miss my guess, to find out what would cause Micron to accelerate that schedule.)
Print the post  

Announcements

When Life Gives You Lemons
We all have had hardships and made poor decisions. The important thing is how we respond and grow. Read the story of a Fool who started from nothing, and looks to gain everything.
Contact Us
Contact Customer Service and other Fool departments here.
Work for Fools?
Winner of the Washingtonian great places to work, and Glassdoor #1 Company to Work For 2015! Have access to all of TMF's online and email products for FREE, and be paid for your contributions to TMF! Click the link and start your Fool career.