AMD vs Intel : What’s next?

Economics stuff wasn’t really what I have in mind to kickstart my blog again, but I thought (honestly) that this may at least get some readers to my blog.

AMD, the perennial underdog to giants Intel, just posted a straight 7th month loss, accumulating more than 1 billion dollars in losses and resulting in a change in leadership from Hector Ruiz to Dirk Meyer. Hector Ruiz was the one who led AMD to the heights of the Athlon first 64-bit, integrated-memory-controller fame, only to see it crumble now, because of a controversial, expensive buyout of graphics company ATI and the late release of new processors (Barcelona) which churn out lower-than-expected performance. Dirk Meyer sure do have an uphill task on his shoulders. But to any fanboy onlooker yearning for AMD to pick itself up again and provide real competition to Intel’s domination, the things AMD need to lookfor now cannot have been much clearer. As someone who hasn’t got a formal education in economics but is somewhat aware of its dynamics, I thought the following can help AMD get out of the blackhole it has gotten itself into.

For the short term, AMD really needs to:

1. Something already in the works…Sell away non-core business i.e. the consumer unit making graphics solutions for handphones and high-definition TVs (Imageon and Xilleon). By selling them, AMD can raise more money to write off the goodwill debt arising from the purchase of ATI and start a period of profitability for AMD.
2. Get that Asset-light strategy right. Whatever that means. It may be good to get on with selling away some of the under-utilised fabs or fab equipment that would incur maintenance costs and leave the actually fab manufacturing to third-parties.
3. Now that consumer graphics are off AMD’s headache, time to make out which markets to really zoom in. Mid-range desktop, laptop and server lines are the way to go.
4. Change its stance in marketing AMD platforms (AMD processor and chipset). Sell their platforms not just by their superior integrated graphics performance (which is undoubtedly much better than Intel’s), which may only confuse consumers. Make a point that an AMD chip can do what it does well too. Why a Ferrari when you can also drive a Hyundai to your destination?
5. Launch 45nm Shanghai chips successfully and convincingly in big quantities in the channels before 2008 ends. Paper launch is unacceptable. This would put a foot into the market before Nehalem arrives in 2009. Shanghai has shown about 15% improvement over Barcelona and this would at least match AMD’s quad-core chip to Intel’s hugely popular Q6600 and allow AMD to sell higher-clocked chips. 45nm dual-core flavours will also be available to give consumers more choice and compete with Intel C2D.
6. Continue to build on the recent home-run successes of the graphics cards line led by 4850 and 4870 by getting 4870×2 on time.
7. Actively release drivers that will really uncover more performance out of the 4000 series graphics cards, especially in CrossfireX mode, to sustain the appeal and momentum 4850 and 4870 has generated.
8. Release low-power, very affordable graphics cards which promises good performance (inherited from 4850) for the OEM market a.k.a. 4650 that would set a new benchmark for entry level gaming cards.

The long term goals should roughly be..

1. Guide 45nm process to perfection, with high K-metal gates ready to go. This would not result in a repeat of the clock scaling and power consumption problems of 65nm process for Barcelona and Phenom.
2. Keep on track for Fusion which would be a true 64-bit, x86 CPUGPU chip, not a sytem-on-chip like that of the impressive Nvidia Tegra and upcoming models by Via and even Intel. Fusion will constitute a whole new line of chips (and source of income) that may revolutionise computing with its promises on radically better performance per power consumption.
3. Produce a totally, radically new microarchitecture that will not at all be a follow-up design to K10. Just like what Intel did from Netburst to C2D.
4. Get Fusion on the right trackfor laptops, then the desktop.
5. Fusion.
6. Continue on with Fusion no matter the outcome.
7. More creative and convincing advertising, which has been seriously lacking thus far.

The rise of Dirk Meyer is not without its critics. Dirk Meyer is a great and proven chip designer, but even though his resume include a degree in business, a chip engineer and hard-core business pro are not the same. It remains to be seen how Meyer will guide AMD back to its heydays, let alone out of the current situation, but he will certainly give AMD a fresh voice and younger drive to survivability and sustenance.