AMD's plan to get back into the black? Fix the internal IT culture, and fast

AMD's plan to get back into the black? Fix the internal IT culture, and fast

Summary: AMD's chief information officer tells ZDNet how his company's efforts to compete amid financial downturn rest on its innovation efforts, because the chipmaker can't innovate without a solid internal IT system.

(Image: ZDNet/CBS Interactive)

NEW YORK — Think for a moment about how well you slept last night. 

AMD's chief information officer Jake Dominguez has one of the more important and anxiety-filled jobs in the chipmaker's recent troubled history — probably more so than anyone else in the executive lounge. He is in charge of overhauling the company's internal processes, which he hopes will make it more nimble and aggressive in the face of extreme competition, not least from archrival Intel, but also up-and-coming mobile chipmakers, including Samsung, Qualcomm, and Micron.

The Sunnyvale, Calif.-based company currently stands in 12th place behind international rivals, and five of its US-based chip-building giants. It's lagging behind not only because of a combative anti-competitive push by Intel, which landed it with a $1.4 billion fine by the European Commission — a penalty Intel appealed. AMD is lagging due to its inability to draw in the crowds with the best technologies it could make at a decent pace.

AMD's problem wasn't a shortage of new ideas. Rather, the company's own IT infrastructure was letting it down. Designers and engineers couldn't work any faster because the company's internal systems were sluggish and outdated, and costs were rocketing to levels that could not be sustained for much longer.

Enter Dominguez, who joined the company in September 2011, and was promoted to head up the company's own IT unit in November 2012 — about the time AMD's financial struggles hit a peak.

AMD's engineering and development operations were spread out over multiple datacenters around the globe. Its bold and daring plan to storm its way ahead was to shrink its operations down to two facilities, one in Atlanta, GA and the other in Cyberjaya, Malaysia. In shrinking its servers and condensing its technologies into a tighter space — enough to virtualize significantly more than it did before while saving power and operating costs — the company would save significant sums of money and ready it for its recovery push in the coming year.

"One of the biggest customer complaints is that we weren't predictable or reliable, and we did not execute," Dominguez told ZDNet in a phone interview. His job first and foremost was to begin "getting our house in order," he said. "Our IT system is one of the few organizations that touches every process in the company. And so us stabilizing our overall environment running, virtualizing and optimizing performance, it now shows through in how we're executing."

AMD's on-time delivery had rocketed to 92 percent at the end of 2013, but it falls short of its own expectations. This year the chipmaker is shooting for 95 percent. While that three-percentage point rise may seem trivial, it can save and generate the company millions in profit each financial year.

"My anxiety right now is centered around the fact that I know what our mission is, what we're trying to do, and where we're trying to stage it... but are we going to get there fast enough?" 
— AMD CIO Jake Dominguez

"Coming into this year," Dominguez said, "the conversation from the engineering house was, 'we need more grid capacity' — whether that means more cores or more storage. And what we have proven is that those needs aren't necessarily true."

Admitting that the datacenter consolidation and the internal IT shift wasn't a popular move at first, he justified it by pointing out too many existing inefficiencies. 

"We didn't invest in a lot of areas that we needed to, so we lost track," he said.

After months of finding new servers and applications, and dozens of "we've found even more" meetings in a widescale organization-wide cataloging effort, Dominguez likened the effort to going into the garage and throwing things out, and finding stuff that you didn't even realize was there and you've been looking for it for years.

The company's datacenter consolidation was led one designated program manager, who reported to Dominguez, to head up the effort, knowing full well in his experience that the traditional "run by committee" model doesn't work. Dominguez wanted a single internal person on point to guide the process, and be able to seek help to him when necessary.

Now as the company begins to pick itself up after more than a year's worth of declining profit and two years of operating income declines, it's back in the black and its operating costs are stabilizing and more than breaking-even.

But the whole process wasn't entirely smooth sailing. One of the more critical parts to the consolidation was physically moving the server racks and equipment from its Austin, TX facility to Atlanta. Logistically that involved moving tons of machines half way across the U.S. in trucks over a quiet public holiday, and within a timeframe that would minimize employee disruption.

"There were two big parts of the datacenter move that made me anxious," Dominguez explained. "One was the big engineering grid we had to move during the July 4th weekend, and then the SAP environment we had to move. And if any one of those went poorly, that was going to change the whole perception of how the datacenter move went."

The datacenter assets, he explained, hadn't been touched or moved in years. With concerns and nerves running high for reasons twofold: what if something happens on the move; and what if something blew up when the servers were fired back up in the new building? If anything fell down in the planned move, the entire company's workflow could have been hampered for days, or even weeks, costing the company potentially millions in lost revenue.

"It took us a little longer to get the network link in place than we wanted, and we were cutting it really close to the timing that we had to install and move our SAP systems. But we fought through that," he said.

"Amazingly enough, we were on target on Monday morning after the weekend to start bringing systems back up, and they did like clockwork."

Dominguez anticipated for about seven-days downtime, but the move went well enough to bring the systems back online after the 920-mile trip "on target," he said, first thing Monday morning, a day earlier than planned. He admitted: "If we had a major failure in the grid, it would've been game over from there and we would've lost the confidence of the company," forcing the company to bring in external hands with a fresh set of eyes.

The (so far) successful consolidation of its datacenters couldn't come at a more critical point in the company's history. Historically, Dominguez said, the company would have outages and coordination efforts, and the move would help the company pivot into new domains as a future-proofing strategy. He said reinventing the company's IT strategy will make it more sprightly in future when it comes to jumping on the latest technological breakthrough.

With the company's attrition rate at historical lows at present, he said, employee satisfaction is on the up. It comes after the company's former chief information officer Mike Wolfe and other high-level executives left the company in 2011 and later. "We had to rebuild the organization and do all of the things we wanted to do," he explained.

Make no mistake, there's still a great deal more to do, he said. "My anxiety right now is centered around the fact that I know what our mission is, what we're trying to do, and where we're trying to stage it — but are we going to get there fast enough?"

With the timeframe he has been given and the execution pace at which his organizational unit has to reach its monthly and quarterly objectives, that's what keeps him awake at night, he said. "That's really what we're chasing and what makes me anxious because if you have any hiccup or stumble in any part of that path, it just slows you down."

If an entire company's prospects for survival rested on my shoulders, that would probably keep me up at night, too. 

Topics: Hardware, CXO, Cloud, Laptops, Security, Storage

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • How does this happen?

    Like the old saying, the shoe maker has no shoes.
    • shoemaker's children

      I think it's the 'shoemaker's children'..
      Anyway, why Atlanta and Malaysia....
      I can understand Malaysia, but why is everyone moving stuff to Atlanta, isn't that
      city choked enough with companies, 'burbs'... traffic, polution, red necks.. etc... :)
      • That was insightful

        That was most insightful....references to rednecks, etc...

        I hadn't realized that data centers run more poorly in areas with higher concentrations of rednecks. Still, this infestation of rednecks doesn't seem to have impacted the decisions of several fortune 500 companies to locate to the poluted and over-crowded Atlanta area:

        Still, bad traffic and 'burbs' must have a determinative impact on location a data center, right? After all, driving conditions and the mere existence of rednecks can impact the operations of a data center....some how, right?
    • Microsoft bears much blame

      In order to promote growth in commercial use of its products Microsoft began to aggrandize box jockeys (the TechNet crowd) early on and molding them into shadow evangelists. Another prong of this marketing attack was to target customer management with disinformation that predisposed them to elevate the role of the box jockey... you know, they guy who gives the boss the newest and fanciest hardware even if he's the last person who needs it?

      So you have a cadre of low skill, low education technicians with a far too disproportionate amount of influence in decision making. Microsoft may have been behind it, but this spread to cover the *nix jockeys and such as well.

      So now you have techs with a ton of influence who don't want to learn anything new. It is much easier to smoke on the loading dock texting the SO on the company CrackBerry all day, plus they always did hate school anyway.

      This is the state of "IT" and why progress languishes, so many are still stuck on WinXP, and any number of other woes. And it is everywhere.
      • First time

        I've ever seen a username explain an entire post.
      • Seriously?

        AMD's chips run x86 code, just like Intel's. Indeed, it's not all bad since Intel licensed x64 from AMD.

        Noting that, let's look at an obvious issue: clock speed per clock speed, Intel will blow away AMD in terms of where a processor counts most: performance. Even the fx8350 octocore can't keep up with the i7-4770k quad core, how the heck is that Microsoft's fault? Or Apple's, since Apple runs on Intel CPUs rather than AMD (and for good reasons, like speed and heat generation...)
        • You are being scammed.

          The reason Intel had to shell out nearly 2 billion to AMD (which frankly shouldn't have been allowed as this let them get away from a DoJ investigation) and the nearly 2 billion in fines from the EU is because of how badly they have been (and still are) rigging the markets.

          Look up "Intel Cripple Compiler" to see how you've been had, also look up which compiler is being used on the benchmarks and see for yourself. in fact there was a stink last year because Cinebench was going beyond using the cripple compiler and was adding code that would purposely SLOW DOWN the bench if an AMD chip was detected. Guess where the majority of advertising for Cinebench was coming from?

          If you run programs compiled with GCC and see for yourself you'l see you are gaining MAYBE 15% on the high end (for a 300%+ price difference) and single digits on the low and midrange (for 75%-150%+ price difference) by buying Intel over AMD. After finding out about the market rigging I went AMD exclusively and the difference is pretty shocking, dollar for dollar any system you build under $1000 will have much better parts by going AMD and under $600 its not even close as you can have a really nice hexacore gaming system for the same price as a lousy locked dual core from Intel.
          PC builder
          • And, didn't AMD drive the 64 bit desktop CPU?

            I seem to recall something about the AMD64, and unless I was dreaming I thought I had read an article by someone from Intel mocking the idea of a 64 bit desktop processes.

            I also seem to recall some legit benchmarks that very clearly showed real performance advantages for AMD in the data centers, especially servers running SQL databases.

            AMD is also my preference. I understand that there are modest performance advantages with Intel. However, as you pointed out, the cost/performance curve is actually quite strongly in favor of AMD.

            I do wish, however, AMD would have acquired nVidia instead of ATI :(
          • Look man, to each their own.

            "If you run programs compiled with GCC and see for yourself you'l see you are gaining MAYBE 15% on the high end (for a 300%+ price difference)"

            We always hear things just like this. That if you only use "my" ways of looking at things you will see differently and choose the way I do!

            It would be one thing if there was some actual major conspiracy, but alas, there is not. This isn't a case where some say Intel is faster and others say AMD is faster and Intel is pulling a scam. At the end of the day, Intel is agreed to be faster than AMD. Its just a question if you want to pay for it or not.

            And on that, to each there own.

            I have been watching the idiotic yelling match around here for years between AMD and Intel fans, and it is idiotic. It wasn't that many years ago when AMD was eating Intels lunch when it came to the fastest processors. And believe me, one can easily look into the ZDNet archives to find all the lunatic talk that was going on back then. And much the same as the loony talk that goes on today around here it was stupid, it was dead wrong, and for the most part it was logically dead wrong and getting the idiots spewing it off was impossible to get them to listen to reason. They thought they were right and they claimed the were perfectly logical and yet it turned out they were wrong and completely obviously illogical.

            What the said back in those days; pre-Conroe, was that Intel was about to get their butts handed to them in a very definitive and permanent way by AMD because AMD CPU's were regularly toppling Intel's best chips in every test. A number of people around here felt an intractable position of Intel going broke was the only outcome.

            Even when Conroe was starting to rear its head, people said (and I know this because I read it right here on ZDNet numerous times) that Conroe would prove to be vaporware. It would never arrive with the promised power. And when early tests revealed it was practically scarry powerful there were actual claims by some poster, real live claims that the tests were rigged and the were done on prototype chips that run of the mill common releases would never run at that speed.

            Talk of Intel needing to charge huge amounts for such chips by the nay sayers, then the price quotes from Intel showing they would be pretty cheap; more denials by AMD fans. At that time the better AMD chips were fairly pricy.

            More talk by the Intel haters saying Intel would never be able to produce in decent numbers, that Intel would be stuck with vast amounts of older PIV stock on the shelf that would never sell and it would kill them. On and on and on.

            What I said at the time was that the Intel haters were making zero business sense. At all. It was like not one of them knew the first thing about the business of selling CPU's and what the CPU market was in the least little bit like in reality. And they probably didn't, but loved to talk trash as if they did.

            I said that it would be illogical for Intel to say they were coming out with a new powerful chip they were calling the Conroe if they were not. There were numerous great business reasons why they would not do such a thing. I said that I couldn't see any logic or any helpful reason, but only colossal failure if the prototype chips in early reported testing were purposefully made significantly better in ANY way then the regular production run chips. Again, to do so seemed like an almost comical colossal business error if they did.

            I pointed out that Intel was not likely to get stuck very long with excess PIV stock on the shelves because they would likely drop prices to "fire sale" lows, because they were a big company and could do so, and because it would make even Intels PIV chips that were inferior to AMD's chips look at least a better bargain than AMD's best chips.

            And that's exactly what Intel did, not because it was some kind of genius move, but because its straightforward obvious business practice. And between the now very inexpensive PIV's and the new decent cost Conroe's, it practically killed AMD. They have never been the same. Not close since.

            It was like it was a one two business punch combination that was deadly and inflicted what has been up to now, permanent damage.

            And of course the biased idiots around here couldn't see it coming. They claimed it was impossible and anyone who said it was coming was a fool. The problem was that in order for the biased idiots to be right the fact also would have had to exist that Intel would have also have been just about to very purposefully and carefully shoot themselves right through the foot for no apparent reason other then they wanted to engage in business suicide by slow and embarrassing death.

            And while that makes no sense to anyone who is not biased into madness, and there for could not possibly have had anything to do with Intel announcing the Conroe chips, and therefore should have excluded considerations of Intel making knowingly unsustainable claims that would lead to inevitable destructive failure and embarrassment, in the minds of the biased, because it was the only possible explanation that could keep AMD from getting hurt really bad, it was the theory the Intel haters grabbed onto.

            As utterly idiotic as it was.

            So go ahead and use what you like. Just don't talk like an idiot.
          • how about this...

            The reason why Intel is the favored corp is because the government has deemed controlling CPU architectures a matter of "national security". Think just a little (it's all made in China) and you'll see why. The government doesn't have the resources to ensure every possible competitor in the marketplace is "secure".
            Mahstehr Blahstehr
        • Invalid comparison

          FX-8350 costs approximately $200 USD. i7-4770k costs about $300. So yeah, obviously the more expensive part will beat the pants off of the less expensive part. If you want to compare processors, clock speed is almost meaningless now, even within the same manufacturer. I'm willing to bet a given Intel i7 will beat just about any i3, even at lower clocks. If you want to compare like against like, what you're looking for is the FX-9590. I'd still buy the Intel in that situation, but not because of performance. The benchmarks are about even, despite the fact this FX is about 1.2 GHz "faster." However, the FX consumes way more power.

          That being said, the FX-9590 is AMD's "top of the line" consumer processor. Intel has consumer offerings that both cost more and outperform it. You want absolute performance? Buy Intel, and be prepared for your wallet to scream. AMD has performance/price nailed, except you'll pay more over the long run in energy costs.

          It'd be better if AMD just stopped claiming these parts as 8 core. They're really not designed that way architecturally, it's far closer to a version of 4-core with hyperthreading in terms of performance.
  • Great opportunity

    He could easily turn AMD around by moving away from the "trusted computing group" creating a NSA / hack proof motherboard/processor combo. Offer full product transparency and embrace open source. People are waiting for the opportunity to replace Intel.
  • I think you a word

    The company's datacenter consolidation was led one designated program manager, who reported to by Dominguez,
  • amd back on top

    I don't care where they keep their data. I just want to see amd back making the best cpus at the best prices, like they once did. I've been buying Intel since 2006 and it still annoys me. I don't care how it is done, I just want amd to offer me the best bang and the best bang f for my buck. If you can't beat Intel with individual cpus, then build a multi cpu architecture that runs seamlessly and is better value. Man up and think like winners again. Make Intel react to you.
    John in Brisbane
    • Intel hit a homer with the Core2Duo

      And has maintained a solid lead since. Apart from x64, which Intel licensed and adopted, AMD hasn't come out with too much. Not that a company or person can wake up and say "Today I will be brilliant and come out with big groovy changes", since real life doesn't work that way, but Intel has managed to make huge leaps with performance. Why not AMD?

      Indeed, reducing the number of datacenters, R&D centers, etc, seems more like AMD circling the drain. :(
      • but just before Core

        AMD was wiping the floor with Intel, but couldn't fully capitalize upon it, as their supply pipe could not be kept full - combined with the threats, real or perceived, from Intel to the OEMs.
        • Your right on the money.

          I recall those days oh so well, see my other post here!

          And as always, even back then, we had trolls. Trolls who absolutely unwaveringly refused to see the blatant writing on the wall for AMD because it was very bad words for AMD when the Conroe was coming out. Some really stupid stupid things were said by many posters who let their brand bias get in the way of what was becoming an ever more stark reality for AMD as reports of Conroe performance tests were coming in.

          I read so many comments here about how Intel was going to go down for the count, and that AMD was about to deliver the killing blow, but not despite of Conroe, but even because of it! That in fact Intels efforts were, according to them nothing more than a pack of lies crafted to save themselves, even though its never been shown anywhere in business history how lying about releasing a new product that is better and cheaper that you know will never materialize or will fail miserably if it does has ever saved a big company.

          So ya, I remember the times well, particularly the nonsense it inspired around here.
  • Quality Control?

    I personally think at least part of AMD's problem is a lack of long-term quality control and reliability.

    I own a PC repiar shop. About 75% of the motherboard issues I see involve AMD/ATI products - most only two or three years old.

    Since I farm out such repairs, I can't give an exact diagnosis of why this happens. It does, however, seem suspicious that a company that makes the minority of chips for Windows-based hardware has the majority of repair issues.

    I'm at the point I'll only use Intel for custom builds and I'm guessing many others feel the same way.
    • Since your tale is anecdotal, I really can't dispute it

      but, in my experience, I see many more Intel failures [which is to be expected, as there are more machines being built with Intel architecture].

      If you're referring to whitebox machines, then I could see it, as these are usually homebuilt, and many times abused by the torture of gaming, pushing parts beyond their thermal limits, and also using PSUs which are problematic due to the economy of their build.

      I do know that, over the years, I see so many more H-P and Compaq machines needing repair than other brands, and these are usually due to inferior hardware and too little consideration for thermal problems [all ostensibly in the name of cheap and quiet].