The war for the control of future system architecture is almost over. It is a replay of previous such struggles. History is a useful indicator to the future where it repeats.
The first important battle was won by Intel (NASDAQ:INTC). By wresting control of the business from IBM (NYSE:IBM) who did not know what it had in the 1980's and by then working with its partner ecosphere, in this case just Microsoft (NASDAQ:MSFT), between them, Intel and Microsoft monopolised the PC market and decided what consumers would get. Once you define the standards, e.g. homogeneous x86 architecture, and once the market accepts that as the standard, you can almost charge what you like and the consumer gets what is given to them. That is how Intel became a hugely profitable and seemingly dominant company with a monster mark up on each chip sold. Paying hundreds of dollars just for a hot CPU was commonplace. That time is disappearing, CPUs are on a path to being a commodity designed to a new heterogeneous standard.
Winding the clock on, dominance led to apathy, arrogance even, and Intel ignored some blindingly obvious moves such as reducing power consumption for mobile applications. The fact it is now doing this as a belated competitive response when it could have done it earlier simply because we were living in the era of declining resources leads one to conclude that it had absolutely no morals when it came to wasting vast amounts of energy. That selfishness and crass stupidity left the door wide open to a company that had watched and learned from Intel how to do the same business model but better.
ARM (NASDAQ:ARMH), known in the UK as just ARM (Advanced RISC Machines), had no choice but to focus purely on design as a 12-man start-up in 1990. It had no choice but to give customers what they wanted but was astute enough to see 'the future is mobile' as far back as the very early 1990s and focused on minimal CPU power consumption. An early ARM CPU powered the first 'Tablet', the Apple Newton. Whilst it was not a commercial success, Psion, amongst early PDAs which followed on, was popular and powered by ARM. Apple (NASDAQ:AAPL) became a significant early investor in ARM
The lesson on standardised chip design from Intel was not forgotten but how to give customers choices as well as what they want? That led to:-
'Heterogeneity of microarchitecture whilst maintaining homogeneity of the software visible architecture', this is a phrase you may see if you ever attend ARM seminars or watch an ARM keynote speech on YouTube. This is key to ARM's success and has allowed the building of an ecosphere of 1,000 companies that contribute to the ARM evolution. Customers can have exactly what they want in an SoC and can call on ARM to help move things into new spheres. The Lilliputians are tying Gulliver down as between them all the ecosphere now own the Heterogeneous standard the market demands, and in the end, it controls the market together with ARM. For example LINARO became a standard in 18 months where previously it took 6 years to create a standard. Gary Smith (EDA) is the source for this and has many other such insights to offer.
The ARM ecosphere licensees get new ARM designs in fully functional and software testable, simulation form one to two years ahead of the release of its silicon form. That means within one day of releasing a chip it works exactly how it was expected to. ARM have had no failures of released chips. The very largest ecosphere customers can adapt the SoC design to get exactly what they want, made where they want, differentiating themselves and using a rapid development environment whilst staying true to the software visible architecture. The genie is out of the bottle and isn't going back to the homogeneous, one-design-fits-all, one monopoly supplier, x86 model even for a temporary power advantage. This Intel iMac I am typing on, is six years old and is plenty fast enough until 4K is commonplace.
Multi-core processors, correctly implemented lead to reduced power consumption! big.LITTLE does the same whilst both extend the reach of the ARM architecture upwards. All of these things can be and are tested prior to release of the silicon form. Hard to believe and comprehend so look at these linked videos.
The world's SoC architecture of choice is now heterogeneous and will never revert. Intel's homogeneous x86 architecture for the big chip that does everything will gradually decay as it is no longer the standard and is carrying 30 years of often useless, historical compatibility baggage that they MUST, but dare not, cast out.
This is why the forty plus billion highly optimised ARM CPUs are everywhere; Tablets, iPods, phones, cameras, E-readers, set-top boxes, TVs, Routers, HDMI sticks, games devices, cars, smart meters to name a few areas. They will have a bite at Intel's heartland of servers, PCs and laptops via partner initiatives that ARM has no ability or desire to control and will take a slice of that market too, particularly if ChromeOS gains traction. For a larger list of new devices, videos and links see:- armdevices.net
Life does not progress in straight lines and Intel have some much anticipated and discussed moves to make as Seeking Alpha author Ashraf Eassa keeps telling us. But each move will reduce previous levels of profitability and it will only make moves forced by having to defend a market or where survival of profitability in a market is threatened. Increasingly Intel are now playing in markets where the best it can hope to be is an option as opposed to a standard. Intel already make ARM chips (via Infineon) and I predict, will make a whole lot more in years to come which is their future destiny.
We should all thank ARM and the ARM ecosphere for bringing down both power consumption and the cost of computing but most of all for enabling the mobile era.
I am sure the Intel devotees will flame this article for being against the faith. Some people just don't get how things have changed. Maybe they still drive 6 litre Buicks. If you didn't understand ARM's momentum before and if you miss a chance to buy ARMH cheaply on a dip such as is starting to appear as I write this, don't say you weren't told. Gulliver has been tied down by 1,000 Lilliputians; he may struggle to rise up again for a while but he isn't coming back, the world has moved on and the little people are growing into big ones fast.
Additional disclosure: I sold most but not all of my ARM shares in February and plan to reinvest on a dip. I also have some Jan 2014 call options on Intel as the release of their new chips in the next few months should temporarily change sentiment. I am long-term bullish on ARM and bearish on Intel subject to changes of strategy by both companies.