Has Moore’s Law gotten away from Intel?
The democratization of Moore’s Law, the revolution led by ARM
I think it’s pretty obvious the PC market is transforming from the conventional two form factors we have been using for the past 15 years into a three or four form factor future. Transforms, not disappears.
The “new PC” is many things: tablet/convertible/hybrids; conventional clam shells with bigger processors, optical drives, more I/O, and bigger screens; notebooks; conventional desktop/desk-side PCs; and All-in-One desktop versions make up the new PC model.
As processors, memory, and, sensors got smaller, and better, and people began carrying small powerful portable computing devices, cleverly disguised as mobile phones, new usage paradigms emerged. You could do things with a smartphone you couldn’t do with a PC.
For example, accelerometers used to cost several hundred dollars each. With the introduction of MEMS accelerometers to trigger air bags in cars, the massive purchases in the automotive industry drove the cost down. The size and power demands shrank, too—Moore’s Law in action. They were adopted in phones for screen rotation initially and later in dozens of applications including AR and games—but no one could have, or did, predict that.
Smaller cheaper sensors, smaller and cheaper processors, led to smaller and cheaper, but very capable devices—the revolution started without Intel, even though Intel fueled it. And although no one could, or did, forecast the social or business impact of mobile devices, when it was realized suddenly everyone was a forecaster of the demise of the PC—crowd wisdom of the worst kind.
Nonetheless, there is no denying ARM-based devices are selling well. The price/power/performance efficiency of the RISC-based heterogeneous SoCs, and their multi I/O is fabulous and is satisfying a great many needs for hundreds of millions of people worldwide. So far, Intel has largely missed out on the revolution it helped ignite. The world’s biggest, and best semiconductor company is catching up however, and with the next generation of 22nm parts will offer the first real alternative to the ubiquitous ARM-based SoCs. And Intel’s SoCs will offer something else, that ARM currently can’ do—multi OS compatibility. The next gen (and to a certain extent this generation) of Atom processors will run Android and Windows, as well as IOS, and Linux. That means if Intel can hit the same power/price/performance points as ARM-based SoCs, it has a bigger TAM than ARM does. Surely, this hasn’t escaped Intel’s notice, but no one else seems very interested in the idea. Am I the one-eyed man here?
It is true Intel and AMD ignored the ARM revolution in their own imprudent way and two CEOs are the casualty of such missteps. It’s hard to give up fat margins on the gamble you’ll make it up in volume. Nvidia is trying to mitigate this very same issue. Qualcomm, Samsung, and even Apple, started out with much smaller semiconductor margins and so didn’t have to go to a withdrawal rehab process. TI got stuck somewhere in the middle and is on its twelve-step program.
Revolutions always catch someone off guard and displace the incumbents. Andy Grove knew this and introduced a catchier description to the process; he called it a strategic inflection point. Later he called it a disruptive innovation crediting Clayton Christensen at COMDEX 1997 as Intel was bringing out the wildly successful Celeron—a smaller, cheaper, more powerful (Israeli designed) processor.
But the Pentium and its successors all the way to Ivy Bridge have been such astoundingly good products with such sweet fat margins, there was little appetite within Intel to abandon that cash cow. It was those cows that provided the positive feedback mechanism to fund the fabulous fabs Intel has built. Now, finally, Intel is applying those super guns to the cheap and tiny and preparing to meet ARM on its home ground.
ARM may have started the revolution, and even won the first round, that’s the good news. The bad news is they’ve woken up the bear and its cousin in Austin and now ARM is going to have a new set of problems to deal with, and one I’m not sure they’re prepared to take on.
Not that ARM is going to suffer much with Intel and AMD in its traditional market space. ARM is a company that was built on tiny processors with penny transactions; it’s a lean mean fighting machine and won’t roll over easily to mighty Intel or pugnacious AMD.
The revolution is on, boys.
Next entry: Nvidia winner in Q1, AMD flat, Intel down