Intel’s ARMageddon

In ten years we’re going to recognize this week as the moment Intel ceded its leadership position in the global chip business. Microsoft’s decision to embrace ARM signaled more than a desire to take on Apple’s iPad — we are witnessing the end of the 'WinTel' era and the emergence of 'ARMDows' - a seismic shift in the IT industry is about to take place.

Intel and Microsoft are like a married couple who can’t stand the sight of each other and yet whose fortunes are so intertwined that they cannot afford to get divorced. Despite what you may read in the press releases there’s no love lost between the management teams of the two companies. Rather the opposite might be closer to the truth. Both have been seeking a way out of their co-dependency but the urgency has been greater for Intel.

Chip making is a commodity business. The only thing that matters is performance and price per watt i.e. how cheaply can they provide the most performance with the lowest energy consumption. That reality is borne out by the dizzying roster of processors Intel produces. Each is uniquely 'tuned' by the marketing department to target a very specific economic demographics of the computer buying public.

Intel’s constant nightmare is the relentless path to commoditization for x86 architecture processors. The pressures created by AMD in this regard lead to behavior which the European Commission’s competition authority deemed to have «diminished competitors' ability to compete on the merits of their x86 CPUs.» In desperate circumstances, companies some times resort to desperate measures.

Intel’s dreams came true when it started playing around with Linux. Here at last was the perfect partner; no one really controlled Linux. It could be molded to Intel’s needs and would finally offer the opportunity to escape Microsoft’s cold embrace. To say that Intel’s increasing infatuation with Linux didn’t go down well in Redmond would be like saying the gulf oil spill was a small leak.

The day Intel announced their engagement with Linux was the day SteveB and the boys in Redmond started working on the divorce paperwork. At CES in Las Vegas this week the writ was served.

The ARM architecture can now be found in over 90% of all mobile phones. It powers the iPad the fastest selling computing device of all time and it is now about to become a very direct alternative to Intel chips powering hundred of millions, if not billions of devices which will run Windows.

Microsoft’s ability to craft a compelling response to iPad and the emerging tablet market in general still remains very much in question but that was not the most important issue at play in this week’s announcements. Intel has already been written off in the low margin, low power, high volume, internet connected device market. Despite being demonstrably unable to curb the ARM architecture’s growing dominance of this market this is not — I suspect — what keeps Intel’s senior leadership team up at night.

The very profitable end of Intel’s business comes from another domain altogether; servers. The inexorable growth in demand for more and more server computing capacity enables Intel to charge a huge and profitable premium for their high end — server targeted — processors. For the last 10 years, Intel have had almost zero competition in this market killing off competitors — like Sun’s Sparc and IBM Power series — by leveraging very compelling volume economics and — most importantly — by riding Microsoft’s growth in the enterprise software business.

The scale economics of the WinTel platform — Intel processors and Microsoft’s Windows Server stack — has decimated competition and established the dominant regime to be found in most data-centers around the world today. IDC’s server numbers — the most respected for this type of data -make the point. In December last year IDC published their third quarter server market share numbers — Intel servers running Microsoft’s Windows Server operating system accounted for 47.7% of overall quarterly shipped factory revenue. The nearest competitor was Linux based Intel servers with 17.5% of overall revenue.

What seems to have been lost in this week’s Microsoft and ARM announcement is the implications for the server business. As this BBC report points out, ARM have been quietly working away at an optimized processor design for server applications for the last last eighteen months. You can bet that the Windows Sever team is now very engaged in that development process. In my view, we’re about to witness a fundamental transformation of the server business and it looks like Microsoft has just lined itself up to be the prime beneficiary. As ARM boss Warren East points out in the BBC report «They [Microsoft] are ambitious to put that operating system everywhere … It works for both of us.»

Over the next ten years we will witness a massive relocation of compute capacity out of private corporate data-centers into high density cloud compute facilities. In ten years time it is very unlikely that any CIO will want to carry the cost, risk and complexity of running their own server farms. Cloud computing will become the utility IT service provider of choice. As this transformation moves forward old Intel servers will be retired from traditional data centers while a new generation of low power, highly efficient — ARM based — servers will come online in the cloud and — in Microsoft’s view of the world — the majority of them running some version of a Microsoft operating system — Microsoft’s Azure Cloud stack is after-all just a tuned version of Windows Server with some very clever and innovative management software running on top.

These market dynamics combine to become Intel’s worst nightmare; a classic market squeeze between multiple ultra-high volume, low margin providers of ARM based processors at the low end and a compelling ARM based value proposition — based on efficiency and low operating costs on the very high end. Intel ends up with what is left of the so-so traditional PC middle market. Even that gets increasingly cannibalized by ever more powerful and capable low end devices. Ultimately Intel neither commands the volumes or the high-end margins required to sustain the R&D expenditures needed to stay ahead and the inevitable downward spiral becomes inescapable.

Of course Intel’s management is far from being unaware of these strategic challenges and with their formidable resources you can expect them try their hardest to compete and innovate their way around this challenge. They need to because the very future of Intel will depend on it.

4 responses

  1. It looks like you are trying to kill intel and are desperate to do so. 10 years is too long and in 10 years, things changes drastically. Intel is already throwing a lot of weight behing processing power per watt and you can see that shift is happenning at intel. There porcessors are using less energy with every new iteration. Intel used to change their processor line up in every 6 month or so and their processors are not single core, they are multicore. I believe, it is not the ARM who will benefit, companies like Nvidia will benefit more. Think on this.

    • @chocoajay — Thanks for the comments. What ever gives you the opinion I am trying to kill 'Intel'. I have no vested interest one way or another. I am merely sharing my views on recent developments. Whatever transpires in ten years time will be as a result of the various competitors actions not my opinions :-)

      Maybe I’m adding 2+2 and getting 5 or maybe not. Everyone’s processors: Intel, IBM, ARM, NVIDIA use less and less power every iteration. That is not an advantage that Intel had to itself. In fact one of the interesting things about the chip business is that everyone works off the same basic roadmap.The benefits of each generation eventually go to every producer because each advance gets captured in the fabrication technology sold by companies like Applied Materials.

      There is no sustained advantage to be had in chip manufacturing technology. That is why everybody makes multi-core chips today. That has become an industry wide capability and not a unique competitive advantage of a single company. This further re-reinforces my view about the problems facing Intel because it removes yet another potential area of differentiation.

      Ultimately chip making is a comoditized business. You either differentiate on the high end with specialized high performance chips for which you can charge a premium or you win on ultra-high volume by being the most efficient producer. There are few other paths to a sustained profitable business model. I stick by my view that the rise of companies like ARM potentially places a very uncomfortable squeeze on Intel at both ends of the business model. Microsoft’s endorsement of the ARM architecture should be a very worrying development for Intel.

      Thanks again for the comments.

Leave a Reply

Your email address will not be published. Required fields are marked *