At its press conference preceding the Consumer Electronics Show in Las Vegas, NVIDIA CEO Jensen Huang stood on stage and declared the company’s dominance in the exciting and growing autonomous driving market. With a flurry of new partner announcements with key players like Volkswagen and Uber, along with new details on upcoming hardware and platform changes that enable customers to integrate these solutions, NVIDIA has the appearance of an organization firing on all cylinders.
Because of that execution, NVIDIA believes it has a two-year lead over any competition in autonomous driving and artificial intelligence platforms. Though a bold statement for any technology company to make in an era of rapid advancement in software and silicon, this confidence stems from NVIDIA’s significant investment in hardware, software, and partnerships.
The engineering advantage NVIDIA enjoys started with its experience in graphics and gaming chips but now spans into autonomous machine specific products like Drive Xavier. This new processor embeds an impressive array of hardware including high-performance graphics, 8 custom processor cores, image and video processors, networking, a deep learning accelerator, and a programmable vision accelerator. All of this adds up to one of the most complex combinations of machine learning capability ever announced requiring 8,000 engineering years and $2B to develop.
There is no shortage of companies that are attempting to gain traction in autonomous driving. Intel has made noise for years and continues to do so with a shifting focus on compute, security, and connectivity. It has purchased several technology providers to accelerate its development but hasn’t shown significant impact. Qualcomm and Arm have been engaged in the self-driving vehicle space for longer, but the compute capability provided by their mobile-first processors falls well behind what NVIDIA can provide.
AMD announced its intent to enter the autonomous driving space at CES this week. It has the graphics processor to enable a competitive solution, but it has a steep hill to climb because of its late entry.
Tesla CEO Elon Musk confirmed late last year that it was developing its own artificial intelligence processor for its cars, an interesting direction considering Tesla currently integrates NVIDIA hardware for its AutoPilot features.
There is surgical precision associated with NVIDIA’s proclamation of a two-year lead and the extreme cost of developing a chip capable of providing this feature set. Of course NVIDIA wants to enhance its public profile and bring in new customers from all areas of the automotive ecosystem. That includes car manufacturers but also service providers (like Uber, trucking and shipping) and vehicle system providers like ZF.
Maintaining the current customer base is also a key driver. Tesla in particular is of concern as it was the flagship partner that helped put NVIDIA on the map as the leader in autonomous driving and assistance services. If Tesla decides it is better off moving to its own platform, either because of technological advantages or cost, it could have a waterfall effect.
Declaring the cost of keeping up with NVIDIA is also a tactic that can dissuade other technology companies from attempting to enter the race at all. Boards of directors may question the direction of investment if they understand the high price and complexity of trying to compete with an entrenched player. The fewer competitors that NVIDIA needs to face, the better its roadmap and outlook appears.
There are clearly risks to NVIDIA’s position in autonomous driving and the entirety of the machine learning space. New technology advancements could commoditize the artificial intelligence requirements for computing capability and we have seen NVIDIA adapt and pivot already with the integration of Tensor Cores in its most recent architecture. For now, NVIDIA appears confident in its ability to sustain this dominance and the market apparently agrees.