By  JOHN MARTINEZ

Only AMD hasn’t been making money recently. With a stock price that’s been bludgeoned to under $2 and a debt load surpassing $5 billion, AMD’s new CEO, Dirk Meyer, had to ponder his manhood and decide that seeing AMD survive into the next decade was a manlier pursuit than maintaining the fab strategy that was bleeding the company into oblivion.

Briefly, the deal is that the emirate of Abu Dhabi, part of the United Arab Emirates, will be investing in both AMD and the fab business that AMD is spinning off, called the Foundry Company. (A foundry is a manufacturing plant that doesn’t do design.) Abu Dhabi already owns over 8% of AMD; this new round of investment will increase that. The Advanced Technology Investment Company and Abu Dhabi will plough additional money into the Foundry Company over several years. The upshot is that AMD will find itself with $1.2 billion less debt and a major cash infusion.

Fab 36, Auf Wiedersehen. This is one small portion of what was AMD’s Fab 36 facility in Dresden, Germany. Whether the passing of ownership to the Foundry Company will be a pro or con for AMD remains to be seen.

That’s the good news. Of course, the real question is if this is also bad news. Was Jerry Sanders right? Are fabs essential to a successful processor company? It seems like only yesterday that AMD was telling me that how breakthroughs like its new overclocking technology, ACC, are a direct result of engineers from both sides of the AMD and ATI (which handles chipsets) fence being able to work so closely under one roof. Such proximity is supposedly vital to staying competitive and making sorely needed leaps in processor technology. Haven’t we heard the same sort of thing from Intel and its various Nehalem teams? Hasn’t AMD been preaching the same gospel for its upcoming graphics-plus-CPU Fusion processor?

Indeed, up until a couple of months ago, you might have sworn that Sanders had to be right. The proof is in the products. But then you would have been ignoring NVIDIA, one of BusinessWeek’s Top 50 for 2008 and winner of the Fabless Semiconductor Association’s Most Respected Fabless Semiconductor Company award in 2007 and 2006. Whatever its present situation between the Scylla and Charybdis of Intel and AMD, NVIDIA has had a tremendous 15-year run with zero fab resources in-house. Like everyone else, the company has had its share of botched and late launches, and this year’s brouhaha over flawed mobile video chips might be fodder for a counter-argument that having your own fabs really does help. But how about the Pentium floating point fiasco or AMD’s more recent TLB setback with the Barcelona launch. Problems happen whether your fabs are internal or outsourced. Perhaps the more critical issue is how well your engineers are able to work with the fab engineers, regardless of where all of those people are located.

If we’re all really honest, maybe we can agree that AMD’s greatest strength has always rested with its engineers. The marketers, executives, and everyone else are well and good, but the crew that brought us HyperTransport and Direct Connect Architecture five years before Intel got around to it with Nehalem and brilliant. And ATI—another fabless company before the AMD acquisition—has equaled if not exceeded NVIDIA’s innovation over the years. Consider the company’s work in digital television, mobile graphics, gaming consoles, and various highlights from the Radeon line.

	X Marks the Spot In AMD’s hunt for future riches, it seems likely that there’s more gold to be found in the creation and marketing of chips, such as the Barcelona dies shown here, than in their manufacture.

So it doesn’t seem far-fetched to imagine a scenario in which the AMD honchos sat around and mused, “We’re on the ropes. We can’t do everything. We need money, and we need to use that money on the one thing we do best. What is that?” The answer, of course, is design computer chips. Not TV chips—and so went that business to Broadcom in August. Not manufacturer chips—and so went the fabs to a spin-off. No, AMD is about designing CPUs, core logic, and graphics chips for computers.

Could the graphics part be spun back off? Not on your life. Hector Ruiz may not be the most popular guy on Wall Street these days (although a sweetheart setup as chairman of the Foundry Company sure can’t hurt), but his foresight in pushing through the ATI acquisition will likely remain his greatest contribution to AMD. It seems very likely that having a graphics engine incorporated into the CPU will elevate integrated graphics to a whole new performance levels. Intel will have its first generation of such a design with Nehalem parts out next year. Without a hybrid approach of its own, AMD could have kissed the consumer PC market goodbye within three years and become the next VIA.

Because AMD made its recent decisions, the company now has a fighting chance. Our cover story this month is on Shanghai, a formidable product if not a breakthrough revolution like the original Opteron. Truth be told, Barcelona didn’t meet anybody’s expectations, least of all AMD. Barcelona was a good design that got caught between the pincers of a botched launch and Intel’s leapfrogging Core microarchitecture. Shanghai appears to finally be what Barcelona should have been, only now you don’t hear the staunch bravado of supremacy in AMD’s voice as we did in 2006. This is a humbler company that understands it probably can’t go clock-for-clock against Nehalem and expect to win. But there are other ways to find victory..

I think we’re revisiting the days of the Radeon X2000 and X3000 families. AMD/ATI came to market with a worthy GPU redesign that simply couldn’t beat NVIDIA at the high-end. Solution? Give the hilltops to the opponent and spread out across the plains. AMD undercut NVIDIA on price/performance throughout the low-end and mainstream and racked up sales left and right. This bought AMD enough time to regroup and prepare for the X4000 family, which finally trounced NVIDIA...for a while. It’s still a toss-up between the two vendors when it comes to multi-GPU setups.

Success in Modesty You don’t have to set records to win. Mid-level cards like this ATI Radeon HD 3450 excel in delivering the features most people want at price points below the comeptition.

The point is that AMD has had low ebbs before and made the best of them. We need AMD—you, me, the channel, everyone. I’ve even had Intel reps say to my face that they’re glad AMD is here to help foster a competitive marketplace. (A healthy buffer against antitrust suits doesn’t hurt, either.) Would Intel have come up with an architecture as solid and efficient as Nehalem without Opteron? Would Nehalem have ever gotten embedded graphics without the prospect of AMD’s Fusion around the corner? Maybe, maybe not. If nothing else, AMD continues to offer surprisingly good value at the mid-range and below. Compare pricing on the Phenom Black Edition against a Core 2 Extreme if you need proof.

I’ve spoken with resellers who are either an Intel house or an AMD house, and I think both approaches are self-destructive. We need at least two CPU vendors just like we need at least two political parties. It’s about maintaining balance, fairness in the market, and a selection of choices that can accommodate the widest variety of end-user priorities. Quite often, it doesn’t matter which component is “best.” What matters is having the pieces necessary to make the buyer an offer he can’t refuse. You’re going to find in many cases that it will take an AMD product to make that happen. How do I know? Because AMD needs the business. Because AMD is determined to be the best channel partner it possibly can be. And because if AMD doesn’t enable you to make customers that irresistible offer, then its recent spate of tough decisions will have all been for nothing.

As a channel and an ecosystem, we owe it to ourselves to have faith in AMD. The company will create value propositions in many segments, and all of them can translate into profit for you. Besides, you never know what 2009 will bring. The market does love a pendulum.