Intel lost the war for mobile silicon before they even realized it had started. With their purchase of Mobileye they're fighting to not miss out on self-driving cars too.
Odds are good your computer has an Intel processor inside. Intel dominated the early years of the personal computer boom and has dominated ever since, despite chief rival AMD's best efforts. But the PC is dead, or at least has one foot in the grave. Sales are falling year after year as more and more people turn to mobile devices powered by ARM-based chips made by Qualcomm and Huawei and Samsung and Apple. ARM chips are even starting to encroach on the PC business and servers, threatening to pull Intel's bottom line out from under them.
Intel missed the mobile boat. Big time. Their chips were never able to match Qualcomm's Snapdragon processors or Apple's in-house A-series chips in terms of performance, efficiency, or price. There's no chance left for Intel in mobile — they tried, they failed, and they've acknowledged as much. But Intel's a big company with huge assets and many shareholders that would prefer the company continue to operate, be relevant, and turn profits. So what's Intel to do?
They're turning to the next big computing revolution: autonomous cars, through their purchase of Mobileye.
Autonomous driving is just starting to heat up, and it's going to be a driving force in growth for both the automotive sector and in technology for years to come.
There's a reason that everybody from Apple and Google to Tesla and Ford is investing billions in autonomy: there are billions and billions more to be made. But we're in the earliest days of it right now, and that's where Intel's opportunity lies, and where they've been working relatively quietly for the past few years.
No company produces the components for self-driving cars entirely in house. The cameras, radars, lidars, sensors, processors, software, and controllers are all sourced from a variety of vendors. Some manufacturers, like Tesla, have demonstrated an aptitude for pulling together these technologies in innovative and exciting ways that push the boundaries of technology, comfort, and the law. Tesla advertises their new cars as being fully equipped to handle full self-driving someday, once the software is complete. The eight cameras, 12 ultrasonic sensors, and radar all feed into an NVIDIA Drive PX 2 — a computing system designed and built by NVIDIA for the express purpose of autonomous driving.
Intel missed the mobile boat. Big time. They'll be damned if they miss autonomous driving too.
NVIDIA has separately partnered with automotive parts supplier Bosch to build their own "AI car computer system" capable of self-driving. This system will be sold to car manufacturers; it won't quite be plug-and-play, but more capable and powerful than what the typical automotive manufacturer will be able to produce.
Tesla's first generation of Autopilot hardware was simpler, with just one camera looking forward. It was capable of basic lane and distance keeping as well as slightly more advanced lane changing. That system was designed and built in conjunction with Mobileye, though the partnership fell apart on Tesla's desire for more data from the camera that Mobileye could not or would not provide. So while Tesla is going their own way and building a new Autopilot software suite and dataset from scratch, Intel is following the route of NVIDIA and Bosch to eventually offer their own comprehensive self-driving hardware system with their acquisition of Mobileye.
Self-driving cars require a few things. One is full and redundant environmental awareness. Tesla's cars are equipped with multiple overlapping cameras, sensors, and radar. Google's self-driving test cars are equipped with cameras and lidar. The self-driving cars that Uber is testing are decked out with a veritable cornucopia of sensors.
All of that means there's a huge amount of data flowing to the computer to the processed, and it needs to be processed accurately, smartly, and quickly. This means you need massive computing power. While power efficiency in that computer would be appreciated, in the end it's the raw processing power that's far more important, and if there's anything that Intel is good at, it's making processors that are powerful. With enormous power reserves on tap, the self-driving car plays to Intel's strengths far more than mobile ever could.
The $15 billion purchase of Mobileye pulls together a company with expertise in the hardware and software of self driving and a Silicon Valley behemoth with the chips that can power it and the coffers to accelerate those efforts. Smartphones have become a $500-billion-a-year business in less than a decade. Automotive has been an enormous industry for decades and its stability is about to be rocked by the advent of autonomy. By snapping up a leader in the nascent autonomous driving tech industry, Intel is positioning themselves to take advantage of the coming changes.
Intel missed out on mobile. They'll be damned if they miss out on self-driving cars too.