NVIDIA (NASDAQ: NVDA) was one of the biggest winners on the stock market in 2017, rising 82% as demand for its graphics processing units (GPUs) spiked thanks to their usage in emerging applications such as machine learning and artificial intelligence. Not surprisingly, Wall Street expects the graphics specialist to extend its run this year.
Investment bank firm Evercore ISI expects the chipmaker’s stock price to hit $250 in the near term, a 30% increase from where it ended 2017. It wouldn’t be surprising if NVIDIA stock hits that price, but investors need to keep a close watch on a couple of threats that could knock the wind out of its sails.
Artificial intelligence will get more competitive
NVIDIA has been one of the pioneers in the field of artificial intelligence, stealing a march over its rivals thanks to its GPU expertise. The company’s GPUs have been widely adopted in data centers and other applications because of their massive computational power, which is derived from the thousands of cores present inside.
This makes GPUs suitable for running multiple heavy workloads simultaneously, while keeping costs under control. For instance, an NVIDIA GPU has close to 4,000 cores as compared to less than 30 cores on flagship Intel (NASDAQ: INTC) server CPUs.
So, GPUs are the preferred choice to run AI workloads in data centers. NVIDIA has capitalized impressively on this trend, as evident from the massive 108% year-over-year revenue growth in its data center segment last quarter. But the graphics specialist’s juggernaut could be halted by the advent of new technologies and competitors.
Field-programmable gate arrays (FPGAs), for instance, could pose a serious challenge to GPUs in the AI space. An FPGA can be reprogrammed to perform a variety of tasks after it is manufactured, and they are more energy-efficient. This makes them ideal for deployment in large-scale server settings.
Moreover, the flexible architecture of an FPGA will allow developers to explore different AI training models, which isn’t possible on a fixed-architecture GPU that’s programmed to perform specific tasks. In fact, Allied Market Research forecasts that the demand for FPGA chips will grow at a faster pace than GPUs for powering AI applications over the next five years.
So, it is not surprising to see Intel is betting big on FPGAs, posing a potential threat to NVIDIA. On the other hand, Alphabet has taken potshots at NVIDIA, claiming that its second-generation Tensor Processing Unit (TPU) AI chip is 15 to 20 times faster than existing GPUs, and is available for use by its cloud customers. It is important to note that Alphabet’s Google Cloud Platform has traditionally used NVIDIA’s Tesla GPU accelerators.
All this means that the competition for AI chips is going to get more intense.
NVIDIA will feel the pain in automotive
NVIDIA has built a lot of hype around its self-driving car work, boasting of numerous partners and an early contract at electric-vehicle maker Tesla. But it’s been all show and no go for the chipmaker as its automotive business has hit a roadblock, growing just over 13% in the last-reported quarter.
NVIDIA isn’t the go-to stock to take advantage of connected and self-driving cars anymore. Intel, for instance, has cut NVIDIA’s first-mover advantage in self-driving cars quite spectacularly. Chipzilla has built a really strong ecosystem of clients and partners after its Mobileye acquisition, and seems ready to move into lucrative markets such as ride-sharing services. Moreover, Intel could start selling an off-the-shelf driverless car system as soon as next year thanks to its Mobileye acquisition.
Mobileye has partnered with Delphi to create a Level 4 autonomous driving system, which needs human intervention only in certain circumstances and is just one level below full automation. This self-driving car platform is expected to be ready by next year, indicating that the competition in the automotive space is going to intensify.
Moreover, Mobileye is planning to open up its design architecture. The company has abstained from doing this so far, and this has kept some automakers and component suppliers from joining its ecosystem. But all this could change once Mobileye allows automakers access to its designs.
So, NVIDIA needs to be wary of the competition in these two fast-growing areas. The company’s terrific growth in the gaming and data center markets has made it a stock market darling over the past year or so, but savvy investors shouldn’t forget the potential challenges that it faces.