Back in 2000, AMD's vision for next generation PC memory was DDR SDRAM. Intel had a different vision: Rambus. You know which won the day.
In October of that year, the AMD-760 chipset enabled the world's first commercially available DDR SDRAM PC. Intel platform desktop customers had to wait over a year before the chip giant followed the leader's tune.
It's not just about adopting technology first though. It's also about supporting what the market wants.
Intel didn't want to adopt PC133 SDRAM, because it wanted its customers to use higher priced Rambus memory instead. But market reality forced the chip giant to get onboard.
DDR2 SDRAM is another case in point. Intel needs to adopt it to enhance its Netburst architecture. Unfortunately, its cost is prohibitively expensive, so market interest isn't strong. AMD, on the other hand, is doing just fine with the current technology. It will move to the next generation memory when the market is ready for it.
Because Intel's 90 nm Netburst processors are generating much more heat than their 130nm brethren, the chip giant needs to move its customers to the BTX PC form factor, which will help alleviate that problem. Of course, for Intel's partners, this platform change doesn't come for free. Again, because of AMD's technology choices, the chipmaker doesn't have a need for BTX today.
For x86 microprocessor innovations, AMD has shown the way as well.
First superscalar RISC - K5
First to use "Flip-Chip" technology - K6
First on-chip L2 cache - K6-3
First use of copper interconnects - K7
First fully pipelined, superscalar floating point unit - K7
First to extend x86 to 64-bits (AMD64) - K8
One wonders if Intel will use the other K8 technologies that differentiate it from its own designs - HyperTransport technology, silicon-on-insulator technology (SOI), and the integrated memory controller.
Of course, I'm sure Intel could provide a whole heap of firsts that the chip giant had delivered. But isn't that what one would expect from the dominant player in the market?
The point I'm trying to make is that AMD, a company that for last year had processor sales that were eleven times less than the 800-lb gorilla it competes against, is putting on a pretty good show.
Role reversal: How it might have been
If there had been a role reversal between Intel and AMD - that is, Intel had developed what AMD had brought to market and AMD had done the same with Intel's technology - AMD would have failed as a going concern long before this day.
Imagine some of the turkeys that AMD would have been saddled with. Rambus memory, Willamette P4 performance and its 217 mm^2 die size, the product recalls - Pentium, Caminogate, the 1.13 GHz Pentium III, and the biggest turkey of them all - Itanium.
If AMD had had the temerity to bring these albatrosses to market, it would have been buried alive with the birds. The only reason why AMD is still in business today is because its strategy has proven to be the best course of action for AMD. One can argue that its past monetary losses are no measure of success, but it certainly beats not being in business at all.
So where AMD has chosen wisely, Intel has definitely erred.
One small step for Intel, one giant leap for AMD
When you compare AMD's K7 and K8 platforms, and then adds to the mix Intel's technological progress, it really does look like one small step for Intel, and one giant leap for AMD.
The problem that this may cause Intel though is ridicule, especially if it's seen as a company that is dependent on AMD technology and strategy as backstops of last resort. If that should happen, what faith would there be in the chip giant's strategic technological planning?
As the sub-head says: One can't be a follower and expect to lead the industry. If confidence in Intel's ability to innovate were eroded, the opportunity door would be left wide open for AMD. It would give the pretender to Intel's throne an opportunity to turn silicon into gold.
Can the chip giant turn the
good ship SS Intel around?
It seems like all the concerns that former Intel chief architect Bob Colwell had about where the chip giant was going are all coming home to roost. There may be a lesson that Intel will learn from the Colwell experience: A strategic plan that goes against the advice of its chief architect may prove very expensive indeed.
At Intel's 2003 spring analyst conference in New York, COO Paul Otellini discussed the challenge of getting companies to replace their aging infrastructure. He shared a 1933 quote by Charles F. Kettering, founder of Delco, and then Director of R&D for GM: "I believe business will come back when we get some products that people will want to buy." Great quote, but are Intel's product offerings sufficiently broad and relevant?
Centrino has made great inroads in the commercial space, XScale was reported to be going exponential, and the P4 and its Celeron sibling are still bringing home the bacon. But when one looks at Xeon and Itanium, the enterprise world had been holding its breath.
Intel has partly resolved the Xeon issue by embracing the AMD64 instruction set. But Xeon MP won't gain that enhancement until sometime in 2005.
The other problem Xeon faces is eating Opteron's dust. For their dual Opteron offerings, Sun and HP are touting performance claims that are up to 45 and 57 percent better than Xeon. When these companies launch their quad Opteron offerings, their performance claims over Xeon MP won't be far off the three-figure mark.
On the Itanium front, the majority of IT decision-makers are still looking the other way.
When flagship products have question marks against them, it can change the whole perception of a company. Centrino, XScale, and the P4 are more than paying their way, but if Opteron and Athlon 64 make the sales inroads that their performance should allow, Intel's high revenue offerings could be in for a very rough ride.
AMD technology, strategy and
stock: Intel backstops of last resort
Because of the cross-licensing agreement that Intel has with AMD, when its strategy fails, it can look to AMD to get it out of a hole.
There are no quick fixes for the challenges that are on Intel's plate. But there is another backstop card that it can play - invest in AMD. Intel has bought AMD stock before. But if it decides to invest in today's climate, or has done so already, it will be seen as another AMD endorsement.
I know, very painful for Intel to do after embracing the AMD64 instruction set. And then adopting model numbers for its processors - after denigrating AMD and telling its customers it would never do such a thing. But if Intel is serious about looking after the interests of its shareholders, this may be one investment that could offset AMD market gains. Better to be safe than sorry.
Intel arrogance has already proven costly
Arrogance has already been costly for Intel. In Q402, Intel told its customers that it would raise flash memory prices by 20 to 40 in January 03. Some of those customers told the chip giant to get on its bike. For Q103, Intel's flash memory sales suffered a 29% sequential revenue decline.
Shareholders can be forgiving for one act of stupidity. But if arrogance should be the reason for another costly decision, Intel investors will want their pound of flesh.
Will Intel catch the green virtual giant?
An industry source has said it will take Intel 18 months to catch AMD up. If that should prove to be the case, Intel has been seriously caught with its pants down.
But let's not forget that as Intel works furiously to make up for its shortfalls, AMD won't be taking a vacation.
So for both performance and technology leadership, will Intel turn out to be the follower in 2004? If the chip giant can't get a handle on its challenges, and AMD executes smoothly to 90 nm, then yes, from AMD's perspective, Intel will most definitely be behind. µ
Intel may introduce CPU model number scheme
AMD cautious about committing to DDR2
Intel "persuades" hardware makers to shift to one kilo BTX heatsinks
Former Intel chief architect shoots from the hip
Slowest Athlon 64 humbles fastest P4 in gaming
Intel "will take 18 months" to catch AMD up
Sign up for INQbot – a weekly roundup of the best from the INQ