> “The thing you have to remember is that this was before the iPhone was introduced and no one knew what the iPhone would do…At the end of the day, there was a chip that they were interested in that they wanted to pay a certain price for and not a nickel more and that price was below our forecasted cost. I couldn’t see it. It wasn’t one of these things you can make up on volume. And in hindsight, the forecasted cost was wrong and the volume was 100x what anyone thought.”
This was an interesting time for embedded microprocessors. I was a big fan of DEC/Intel StrongARM and XScale, and the team I worked with preferred using them in designs.
I disagree that it was simply an economic choice, and I think anyone who worked in the ARM embedded industry around that time would also disagree. The move to sell off XScale was after years of neglect. The most obvious defects were the PXA250, which could only cache write-thru due to a bug (Wikipedia does not mention this), and the lack of USB2.0 high speed, despite it being a key feature of every competing SoC.
It was obvious to everyone in my industry that ARM was going to take off in a big way, and exponentially -- "microcontrollers" had grown up, and reasonable compute power could be put everywhere at home, and even carried around with you.
Intel, on the other hand, was seen by everyone I knew, to be more concerned with cannibalization of its low-end x86 market. Stealey and Atom were coming down the pipeline (Silicon Valley secrecy wasn't what it is today), and Intel was looking to make a play in the market. Except, Intel's designs were off the mark by orders of magnitude in power consumption, and could never fill that niche. Still, the word in the street was Intel had bad internal politics at play, and its ARM offerings had to go.
So yes, it's a great shame Intel missed that incredible opportunity at a pivotal moment. But no, it was not an economic decision.
I can remember how RISC chips were going to take over the world, Linux/RISC was going to be the greatest threat to Intel. 10/20 years later x86 is everywhere from the Macs to high end servers to lowliest tablets. The latest Atoms are actually very good.
ARM meanwhile dominates the lowest end, but no one really makes money out of it. I think Intel was wise to avoid.
> 10/20 years later x86 is everywhere from the Macs to high end servers to lowliest tablets. The latest Atoms are actually very good.
10/20 years later, ARM represents the vast majority of both volume and profit in home personal computing. x86 is only "winning" if you define a very small category to compare against ARM.
> ARM meanwhile dominates the lowest end, but no one really makes money out of it. I think Intel was wise to avoid.
Edit: Meanwhile the "PC" category became largely irrelevant. That's what Intel missed.
> Edit: Meanwhile the "PC" category became largely irrelevant. That's what Intel missed
That's a ridiculous claim. 2015 shipments of 288.7 million units (Gartner) may be down but they're still huge: that's 791,000 per day. And sales of Windows tablets and 2-in-1s are growing very rapidly.
Are tablet sales "largely irrelevant" because they're lower than PC sales (206.8 million, says IDC), and even if you include tablet PCs running Windows, declining even faster? https://www.idc.com/getdoc.jsp?containerId=prUS40990116
There are roughly 4x as many iOS, Android (and runners-up) sales and growing. The graphs continue to trend in that direction. IDC is infamous for continuing to make useless comparisons which show PC sales to be better than they are. Possibly (if cynically) it's because that's who still pays for the research.
PC sales, even as narrowly defined by IDC, are now a small minority of total personal computing volume and profit. This shouldn't be surprising to anyone
Regardless of people buying smartphones and owning multiple devices, PCs are still selling well. In fact, they are selling in greater volumes than in most previous years.
Smartphones also benefit from their high junkability factor. A decent PC now has a useful life of around 5 years, so sales should be half what they were when the average life was 2.5 years.
Smartphone sales will also halve when they last 3 years instead of 18 months.
> PC sales, even as narrowly defined by IDC
IDC's PC numbers are artificially depressed because it doesn't include tablets and 2-in-1s. So when Microsoft sells a Surface Pro 3, IDC doesn't count that as a PC.
If Intel was the leader of the ARM mobile gang, TSMC would have been denied a lot of money and maybe even couldn't compete with Intel, Quallcom wouldn't have so strong and posed a possible future threat to Intel, and maybe Apple would have been dependent on Intel(due to better process), enabling Intel to get decent margins, And Intel would have been in a better position to influence the Android market towards Atom or some close extension of the ARM.
> I disagree that it was simply an economic choice, and I think anyone who worked in the ARM embedded industry around that time would also disagree.
I agree. Anyone watching the PDA/smartphone market at the time could see that something big would come behind the corner and shake things up. Yes that was the time of the HP iPAQ and Dell Axim, but also dozens of HTC made phones under different brands and tens of millions Nokia smartphones, all powered by TI's OMAP and not Intel.
The original iPhone sold around 6 million units, at the time Nokia S60/S80 devices, Blackberries, UIQ and any other smartphones driven by ARM CPUs had sales of over 100 million.
So yes, Intel just missed a chance and didn't care.
This was an interesting time for embedded microprocessors. I was a big fan of DEC/Intel StrongARM and XScale, and the team I worked with preferred using them in designs.
I disagree that it was simply an economic choice, and I think anyone who worked in the ARM embedded industry around that time would also disagree. The move to sell off XScale was after years of neglect. The most obvious defects were the PXA250, which could only cache write-thru due to a bug (Wikipedia does not mention this), and the lack of USB2.0 high speed, despite it being a key feature of every competing SoC.
It was obvious to everyone in my industry that ARM was going to take off in a big way, and exponentially -- "microcontrollers" had grown up, and reasonable compute power could be put everywhere at home, and even carried around with you.
Intel, on the other hand, was seen by everyone I knew, to be more concerned with cannibalization of its low-end x86 market. Stealey and Atom were coming down the pipeline (Silicon Valley secrecy wasn't what it is today), and Intel was looking to make a play in the market. Except, Intel's designs were off the mark by orders of magnitude in power consumption, and could never fill that niche. Still, the word in the street was Intel had bad internal politics at play, and its ARM offerings had to go.
So yes, it's a great shame Intel missed that incredible opportunity at a pivotal moment. But no, it was not an economic decision.