A report from Bloomberg this week has made public something that should already have been apparent to tech industry observers: Apple is planning to replace Intel processors in Mac computers with its own chips starting sometime around 2020. The two California companies have enjoyed a long and fruitful partnership ever since Apple made the switch to Intel CPUs with the 2006 MacBook Pro and iMac, but recent trends have made the breakup between them inevitable. Intel’s chip improvements have stagnated at the same time as Apple’s have accelerated, and now iPhone systems-on-chip are outperforming laptop-class silicon from Intel’s Core line. Even if Intel never cedes its performance crown, the future that Apple is building will invariably be better served by its own chip designs.
Apple’s decision to ditch the world’s most popular CPU line for laptop and desktop computers may seem radical, but there are a number of key factors that actually make it obvious and unavoidable.
Attend any major tech exhibition and you’ll find Intel announcing or reannouncing mildly improved processors. Whether you’re at IFA in Berlin, CES in Las Vegas, or Computex in Taipei, the spiel is always the same: the future is wireless, battery life matters to everyone, and there are a lot of people with five-year-old PCs who might notice a difference if they buy a new Intel-powered computer. It’s all painfully incremental and out of sync with Apple’s product cadence. Apple will give you, at most, two years with an iPhone before enticing you into upgrading, whereas Intel is trying to convince people with PCs that are half a decade old to do the same.
In the past, Intel could rely on microarchitecture changes one year and production process shrinkage another year to maintain its momentum of improvement. But the infamous Moore’s Law sputtered to an end back in 2015. Intel is approaching the limits of what’s possible to achieve with silicon, and it hasn’t yet figured out its next step. The chart below, compiled byAnandTech, illustrates Intel’s predicament well. Notice how long the 14nm process node has endured, the question marks next to the release window for 10nm chips, and the almost total absence of a future road map. In previous years, Intel’s ambitious plans would be known well in advance. (The company hasn’t grown more secretive; it just doesn’t seem to have any secrets left.) And without the power efficiency gains that come from building smaller chips, Intel just can’t compete with ARM processors designed for efficiency first.