Logical conclusions

Written by Brett Thomas

July 29, 2007 | 14:22

Tags: #commission #competition #ec

Companies: #amd #eu #intel

I love logic. I love that at the crux of every decision, there is a tree illustrating cause and effect. At the base of every good decision tree, you can find fundamental pieces of evidence which help support the conclusions drawn.

What's even more fun than logic to me, though, is where it fails. It seems that the more simple the concept, the less useful it actually is. Nowhere is this more true than in trying to define something by using the definition of something else, like mathematical substitution.

For instance, try this: "Business is war," and "War is hell." Logically, one could assume that "Business is hell," right? Okay, so maybe that's a bad example...most days, I would be pretty hard-pressed to argue.

But, I digress...so let's try another one: "Intel dominates the processor market," and "monopolies dominate the market by unfair advantage and anti-competitive practices."

Doesn't exactly roll of your tongue, I know. In fact, the two statements really don't have all that much in common...but the European Commisssion seems to think they might. What a shock.

All snide comments aside, this is serious business with serious consequences. Monopolistic practice is a dangerous thing, I think most of us agree. But where does it stop being about fair competition and becomes handicapping a bigger opponent?

It seems that since the release of Intel's Core microarchitecture, AMD has been falling farther and farther behind. At the same time, it's been getting louder and louder, screaming unfair competition, monopoly and so many of these other buzzwords.

"I think we may need to take a long, hard look at the unfortunate victim role we've been giving AMD as of late."

Is there any truth to these allegations? Probably some. But it's all about how you set up your argument. To AMD, Intel is a bully making back-room, exclusive deals and engaging in mutually damaging price wars. To Intel, AMD is a whining kid who's upset he hasn't come up with an original idea since the K8 architecture.

Who's right? My bet is neither...the truth is somewhere in between. But, looking at the numbers, I think we may need to take a long, hard look at the unfortunate victim role we've been giving to AMD as of late. It's a sympathetic plight, much like that of Immersion vs. Sony...but that doesn't mean there's not more behind the scenes.

Now that I've said that, please put down the pitchforks. I'm not just inventing this - but as usual, it's a game of numbers. And it all comes down to two primary numbers: market share and capacity.

Let's start off with a look at market share. Are you aware quite what a booming business PCs are? Not including the DIY market (read: us), there were almost 68 million sales of x86-based machines just in Q4 of 2006. That's up by nearly 6 million from the same quarter in 2005, and finishes off the sales of almost 240 million PCs over the year. Again, that doesn't do justice to the fact that all of us exist - and the DIY population accounts for a few million more in sales worldwide.

Over the course of 2006, AMD accounted for roughly 45 percent of US desktop and notebook sales, which make up roughly a quarter of that 240 million. If you do the maths, that's about 27 million chips in the US alone. Even if AMD captured only 10 percent of the remainder of the world's sales (well under its actual share of the market), that would be another 18 million chips, topping the company out at 45 million chips sold.

I suppose in relation to the 195 million chips that Intel would have had to sell, it's easy to see why AMD might feel a little jaded. And, indeed, this is enough information for most people to stop digging any deeper. Here is a company that is being outsold by its rival over four chips to one...and that rival is offering price incentives, contract bonuses and other benefits to push those numbers down even further.

"it's hard to say your business is being injured when your sales are outstripping your production capacity."

When you put it that way, it's hard not to see a monopoly. I guess it's a good thing I don't like to stop digging, because these seemingly logical numbers are lacking a vital component. See, in order for a company to be a monopoly, you have to prove that the bigger company's practices are injuring the smaller company's business.

Think we just proved that? Think again. See, it's hard to say your business is being injured when your sales are outstripping your production capacity. Yep, that's right - year after year, AMD is near to or actually does sell out of chips...a "problem" that Intel only wishes it could experience.

Each year, Intel has a warehouse stacked with chips that need to be sold, encouraging the company to cut costs aggressively to get rid of its old inventory. Remember the changeover to Core and how cheaply you could buy some Pentium D models? That was excess inventory at work.

In the meantime, AMD has had to "compete" by making the chips one day and selling them the next. The price wars are hard on the company, don't get me wrong - but AMD isn't sitting there with warehouses full of old chips that nobody wants to buy. Quite the opposite - a truckload is made, and a truck gets loaded. It's really a testament to good management, functioning on what we in accounting call the "Just in Time" inventory management.
By functioning this way, AMD avoids the costs of warehousing either the excess raw material or the completed chips - it's a money-saver all around, and takes a sharp eye and steady leadership. The company must have both, because its production capacity is growing year after year, and seems to be hitting its sales figures like Robin Hood firing an arrow. In fact, business is so booming that AMD has had to contract outside production help just to keep up.

Indeed, in proof of this, the company just announced it's writing off $30 million in old inventory for the first time in 2007 - and $30m is almost nothing. This is such an unusual occurrence for AMD that it actually required special mention on the financial statements. And this doesn't really even specify just what that "inventory" is - it can include old raw materials, dead-die SKUs, or even some production machines - after all, Fab 30 is undergoing a conversion now.

So, let's talk about production capacity for a bit. AMD has two fab plants, both of which switched to 65nm only recently. In 2005, the company made roughly...drum roll, please....45 million chips. Now, where did we see that number before? In 2006 (mind you, both fabs were still running 90nm chips on a combination of 200mm and 300mm wafers for almost the whole year), the company produced about 57 million - leaving a nice cushion for its increase in sales this year. Lo and behold, sales are again up for the green team in 2007.

In the meantime, Intel has around sixteen plants with four dedicated to running 65nm CPU dies on 300mm wafers (three from the end of 2005, a fourth late in 2006). Now, a 300mm wafer is actually double the surface area of a 200mm wafer (the joys of geometry). On top of that, the chips themselves were smaller - Intel's Presler was only 162mm² and Cedar Mill was 81mm² across the die, while the whole K8 line has hovered around the 200mm² mark.

So even if we don't include the die shrink, we're looking at twice the plants making around twice the number of chips - or four times larger figure. Once you add in that shrink to 65nm, you once again have a glut of Intel processors being stored in some warehouse, not being sold.

"I don't even think Intel's pricing choices and "shady" deals have much to do with AMD at all."

All of these are rough figures, but I'm erring on the side of caution - the real situation is undoubtedly even a bit better for AMD. Somehow, in spite of Intel's size, AMD is selling out its inventory year after year. About the only thing AMD can really complain about is cost - Intel's price wars undoubtedly do damage...but if you compare the way the companies themselves perform and not just gross sales or market share, it's easy to see why Intel has to do it.

I don't even think Intel's pricing choices and "shady" deals have much to do with AMD at all. It's pretty clear that, due to the production issues, the companies can coexist relatively comfortably without much of a battle. It makes perfect sense to offer package and volume deals to resellers just to get your own stock down and out of your warehouse. Do the cuts hit AMD's bottom line? Sometimes - but I think the effects of that are a tad over exaggerated in favor of the raw sales numbers, which can paint a much more "grim" picture.

If you turn back the clock to 2004 (when AMD alleges that the worst issues happen), the situation gets even more interesting. There was no Fab 36 at that point and also no 300mm wafers - AMD had one plant and a production capacity of (at best guess) less than 25 million chips, in a year that sold 189 million PCs. At the time, AMD had even told us here at bit-tech that it couldn't openly advertise the Athlon 64 - the chip was already selling as quickly as it could make it.

Obviously a company that's as large and established as Intel will outsell you. But when you're selling 100 percent of your inventory, the only way to make that not happen is to either get bigger yourself or slow the sales growth of computers for a bit while you catch your breath. That way, you can force them to idle their plants; or, even worse, get them to build lots more chips that won't sell, which they'll have to pay to warehouse until they can liquidate them well below cost.

Now, if you've just spent $5.4 billion acquiring an extra company with no fabrication abilities of its own and then need to borrow another $2.2 billion just to keep your production up with your orders, you certainly aren't able to get much bigger just yet.

But, ya know, people have a soft spot for David vs. Goliath stories, and there's always someone willing to hear it. And nothing can slow down a market while giving you extra cash like your competitor having to raise prices to pay off some absurdly expensive fine. After all, then you can raise your own prices - and you still sell all of your chips.

"Good afternoon, EC Competition Commission, how may I help you?"
"Hello, this is Mr. Hector Ruiz. I'd like to place a complaint."
"And the nature of this complaint, Mr. Ruiz?"
"Yes, I believe there are uncompetitive business practices being taken against my company..."
"And how did you determine this?"
"Well, it's the only logical conclusion..."

Discuss this in the forums
YouTube logo
MSI MPG Velox 100R Chassis Review

October 14 2021 | 15:04