The author, and many others, fail to note that in the 90s Microsoft made every effort to include other chip vendors - Windows NT ran on DEC Alpha chips, and people claimed to be working on a PowerPC port, until the IBM-Motorola alliance died.
Modern Windows has always had a Hardware Compatibility Layer - this is the only part that needs to be processor specific. Intel is the real villain - Andy Grove took great pride in stopping audio and video coprocessors from being used in PCs. Everything had to go through the processor. This philosophy has resulted in $200 chips in $600 desktops. The Intel CPU is responsible for 35% of a system's cost; an ARM chip is only 7-10% of the total cost.
Short answer - hardware vendors, and consumers, are tired of Intel holding back progress to keep their margins high.
My 2 cents and a great Sunday rant,
Finfets and 22nm is perfect for big honking processors used in servers or desktop computers. I don't think the extra power purchases you much in a cell phone or tablet, where the market growth is taking place. ARM's challenge to Intel in the server market may be Gulliver and the Lilliputs, but it would appear Intel has lost the mobile market.
Intel won the last processor war against the power-pc, sparc, mips architectures by out-manufacturing the other suppliers by riding the bleeding edge of mooores law. But now with moores law slowing down (is it a law anymore then?) the finfet transistors are Intels salvation at least for now to keep ahead of the other suppliers.
But what if finfets do not pan out for Intel - I have to imagine that such a radical new transistor will be more expensive to produce, more difficult to produce defect free in volume. Then the little lilliputs armed with their sling-shots will soon surround and suffocate the goliath.
Or maybe I am being too dramatic!
Intel vs the ARM ecosystem is like Gulliver and the Lilliputs - and we have seen that played out in the computer arena since the days of the Mainframes.
The low voltage - low power CMOS that the author's company SuVolta is promoting as an alternative to the FinFET by Intel ( his previous employer ) definitely looks promising for Foundries serving Fabless chip makers in the ARM based Mobile segment. But Intel too might have a few tricks up its sleeve to counter the Lilliputs.
Growth of mobile computing in latecomers like India, Africa etc. could prove to be a wildcard. Intel may suddenly drive down the cost of the chipset by integrating many of them ( incl. analog & RF )into a SoC that Foundries may not be able to duplicate.
And if nothing technical works then there is always the "nuclear option" of buying out ARM since anti - trust laws in England are far weaker.
Blog Doing Math in FPGAs Tom Burke 2 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...