Put a Hold on Building a Position with Nvidia
AI in general and autonomous car, in particular, provide an excellent potential for Nvidia’s GPUs, which are well-suited for parallel computing and training of AI algorithms. However, Nvidia’s shares dropped by 16.9% in the last month, compared to the 2.5% drop of Nasdaq Composite Index. Even after beating the earning expectation slightly for 2020Q1, Nvidia’s stock still lagged behind the market. While I believe Nvidia has a considerable growth prospect in the next 5 to 10 years, there is no hurries to build a position with Nvidia due to the near term pressure and the uncertainties about Nvidia’s advantages in AI.
In the latest quarter, the gaming segment, which accounted for about half of Nvidia’s revenue, seems to have started to recover from the crypto hangover. However, all other segments, such as autonomous car and data centers, are either did not change much or drop modestly. Nvidia also walked back from earlier guidance regarding the full-year revenue and refrained from offering a full-year revenue. Reports from other chip makers also point to a cyclical downturn this year.
On the other hand, there are considerable uncertainties about the of Nvidia’s advantages in the future built on AI and cloud. Game streaming, such as those by Google’s Stadia, and ProBeat where Microsoft and Sony join forces, may increase Nvidia’s gaming GPUs in the short run. However, as gamers move to the cloud, the demand for each gaming to equip a high-end GPUs may subsume.
Many have recognized that Nvidia GPUs’ technological advantages in training machine learning algorithms. But some have also pointed out that Nvidia’s dominance in inference, where users and software applied the new data with the trained AI algorithms, is less clear. For AI to provide competitive advantages to business, it must be combined with a tremendous amount of customer or organizational data. These data are mostly machine-generated and more and more stored in the cloud because the applications and software are hosted on the cloud. The AI’s tie to the cloud provides a double-edged sword for Nvidia. On the one hand, Nvidia’s open source libraries that are empowered by Nvidia chips for fast applications of machine learning have created a large number of ML programmers using Nvidia’s technologies. The large user based may make cloud providers such as AWS hesitate with switching to alternative technologies.
On the other hand, if being on the cloud means that cloud users could use whatever off-the-shelf ML applications and need not worries so much about the training algorithms themselves, then there are not network effects for sticking with Nvidia. For example, Google has developed its own chips specifically for the training AI algorithms. Being the AI giant, Google’s route to catch up AWS and Microsoft in the cloud IaaS market would depend on their ability to translate the technological edges into customer attractions. If Google manages to do it, the AI future with the cloud may be less dependent on Nvidia chips.
Alphabet’s Waymo also makes use of field-programmable data array (FPGA) chips for their autonomous cars. Tesla also developed its own chip. Elon Musk claimed that Tesla’s own chip is superior, which is yet to be proven though. Therefore, it is far from certain that the future of autonomous cars depends on Nvidia chips. The good news, however, is that leading auto-makers such as Toyota bet on collaboration with Nvidia for their development of autonomous vehicles. It remains uncertain, though when the mass production and adoption of self-driving cars would materialize. Therefore, while I will keep paying close attention to NVDA, I am not in a hurry to build a position.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.