The rise and fall of AMD: How an underdog stuck it to Intel

It's not so much that AMD "stuck" it to Intel as Intel shot itself in the foot. The only reason AMD was successful was that Intel made a gigantic mistake in NetBurst ("10 GHz or bust!")<p>When Intel finally rectified that mistake and released Conroe it was game over for AMD. AMD simply can't compete with Intel whether in architecture or fabrication.<p>What AMD could've done was to concede the PC market to Intel and focus on the emerging mobile device market - the classic disruptive attack. ARM went that route and now has a market cap that's 10X bigger than AMD. Qualcomm adopted a similar strategy with Snapdragon and now has a market cap equals to that of Intel's.<p>Lesson: never fight a dominant incumbent in its own game. You will get killed. Play a different game. Better yet, invent a new game where you know the rules best.

&#62;AMD began life as a second-source supplier for companies using Intel processors.<p>Ars should know better. AMD started in 1969, making their own logic parts. In the late minicomputer era, their Am2900 series of bit-slice[1] components was king, being used to build CPUs for models of DEC PDP-11 and PDP-10, DG Nova, Xerox Dandelion ("Star"), Wirth's Lilith, Atari vector arcade machines, and countless other machines.<p>[1] <a href="https://en.wikipedia.org/wiki/Bit_slicing" rel="nofollow">https://en.wikipedia.org/wiki/Bit_slicing</a><p>(Not especially relevant disclaimer: I work for Intel.)

What a wave of nostalgia. I took a job at AMD in 2005, right at the zenith of their success. I was totally enamored of the great technology that went into K7 and K8, and I was ready to help this underdog company stick it to Intel and turn the microprocessor world on its ear.<p>I worked there for six years, through fumble after disaster after boondoggle. I could go on and on about all the reasons I believe AMD went down the tubes -- and I'm really looking forward to reading the second installment -- but I think a lot of it reduces to the disfunctional corporate culture alluded to in this piece.<p>During my time at AMD -- and the old-timers confirmed that it was ever thus -- there was <i>always</i> the sense that every project was make-or-break for the company, that we were always on the brink of disaster. Long-term strategic planning is simply not in the company's DNA. We lurched around like a headless chicken, and when -- through a combination of good products and missteps by Intel -- AMD finally got a taste of sucess, we squandered it in the most ham-handed and disastrous (and predictable) way.<p>P.S. Bulldozer project was a total horror show, beginning to end.

From my perspective, AMD really screwed up the AMD/ATI merger in the worst way possible. When AMD bought ATI they were well positioned to beat everyone to market to release their APU chip. However, the problems started almost immediately when neither AMD nor ATI did anything to combine resources. They should have put in the time and money to merge teams at all levels of the company. Instead they built a few small groups that included senior personnel from both units, then they left most of the lower level teams to work on whatever they were working on previously. Never mind the fact that there were a lot of really clever people all over the place ready to contribute really great ideas.<p>In other words, there was zero global direction down the ranks. It was just business as usual; keep doing what you've always been doing, and maybe we'll show you some nice slides a few times a year about how great APUs will be. This lack of organization meant that no one had any idea what anyone else was doing. Worse-- even if you wanted to find out there was absolutely no company-wide documentation or organization on anything. Your only hope for getting information was hoping one of your co-workers had bookmarked some magical page with the info you required. l This just got worse when you accounted for the problem of elitism. The hardware teams were just so much better than the software teams. After all, software is easy, so what sort of useful input could those code monkeys offer. And far be it from the software teams to actually talk to someone from the QA teams; those QA people were beneath notice. Finally, add in a very wide distribution of personnel seniority, insane levels of paranoia about job security, grade-school level office politics, and completely disparate management styles, then hit blend.<p>So really, the results are not at all surprising. You can't have two companies pretend to be one while playing tug-of-war, and still be competitive.

How I remember those days! after the complete disaster that were the first PIIIs comes the Athlon. It was just so amazing, so unbelievable that you could get that much power without breaking the bank.<p>And it was the same story until about 2006, you could get an Athlon or a FX and get more than enough CPU power to run almost anything in the market at the time. Even the Sempron which was the cheap option was <i>good enough</i>, I remember guys in forums getting mobile version and OCing 'em to the very limit.<p>But then came Core2Duo, and AMD literally had nothing against it. The first Phenom sucked, big-time, there is no other way to put it. The Phenom II was much better but not good enough and the only talk about it was how you could turn on the disabled cores in the X2 and X3 variants since they were the same silicon than the quads.<p>I really wanted the FX to be as good as the Athlon-era FX, but it wasn't, not even close.

Hang on, what about consoles? AMD are doing the CPU and GPU for the PS4 and presumably the next Xbox if rumours are true.<p>Surely this is quite a windfall? Particularly if they're pumping out the same part for ~5 years, much longer than the average CPU stays on the market. Is this not a cash cow?<p>I'm disappointed AMD are no longer competitive on the CPU desktop performance market, but I'm not sure I understand <i>why</i> they cant compete with Intel on CPUs given the stable revenue offered by consoles and their competitive GPU line. Is it Pure R&#38;D budget? Or Intel are too far ahead tech wise? Intel have the best engineers? What is it?

Apparently AMD will have the upper hand in the next gen video game consoles PS4 -&#62; <a href="http://en.wikipedia.org/wiki/PlayStation_4#Hardware" rel="nofollow">http://en.wikipedia.org/wiki/PlayStation_4#Hardware</a> xbox 720 -&#62; <a href="http://www.digitalspy.co.uk/gaming/news/a471564/xbox-720-to-use-amd-processor-lacks-backwards-compatibility.html" rel="nofollow">http://www.digitalspy.co.uk/gaming/news/a471564/xbox-720-to-...</a> or at least that's the rumor.

AMD has only stayed in the game to begin with because Intel was 'forced' to license x86 to them in order to avoid government anti-trust prosecution.

I found it interesting that Bill Gates had a hand in AMD buying NexGen, the company that gave AMD its' K5/6 technology and brought them Raza.<p>I don't really know what was going on with Microsoft and Intel around 1995 but it appears pretty clear that Bill wanted to have atleast two x86n chip suppliers available for PC manufacturers.

A book recommendation relevant here: Inside Intel is a great book about the founding and early-days sparring with AMD - it's old - I bought my copy in the late nineties en route to Silicon Valley from London. But it's a great scene setter and backgrounder for these sorts of stories. Great read.

AMD also had a niche in supercomputers, for example the Cray XT/XE series. Not sure the absolute volume really made a difference there though, the number of Cray sales would be dwarfed by ordinary PCs and servers.

This is a very scary image for AMD:<p><a href="http://i.imgur.com/pCiDEKP.png" rel="nofollow">http://i.imgur.com/pCiDEKP.png</a><p>Their revenue is falling, and falling consistently.<p>They pretty much lost as a CPU manufactured in every area - desktop, laptop, tablet, phone.<p>Their only option at the moment is to keep making the best damn APUs and GPUs, and try to invent something new. I wish they came out with a 1000-core CPU or something.