Chip darling Nvidia just got a major boost from search engine giant Alphabet Inc., which has announced plans to integrate Nvidia Inc.’s Tesla P4 graphic cards into the Google Cloud Services to accelerate artificial intelligence inference tasks.
AI inference is the process of taking large data models obtained from AI training modules and compressing them into a format that lower-power and lower-cost hardware can deploy.
NVIDIA introduced two Tesla GPUs (graphics processing unit) in 2016 based on the company’s next-generation Pascal architecture that delivers 45 times faster response than Intel Inc.’s top-of-the-range CPUs and GPUs.
Their impressive processing speeds makes them ideal for AI-related tasks such as video-content scanning, image matching, voice-recognition algorithms and autonomous driving.
AI tasks tend to be resource-hungry and demand chip performance beyond what general purpose x86 CPUs and lower-powered GPUs can deliver.
Nvidia’s AI Segment Expanding Rapidly
This will, however, not be the first Cloud-GPU partnership between the two tech giants.
Google Cloud has already incorporated Nvidia’s Tesla that it rents out to the public via virtual machines for the Compute Engine and Cloud Machine Learning.
Although Nvidia’s Gaming GPUs still account for the lion’s share of the company’s top line, its AI chip business is gradually catching up.
During the company’s Q1 2018 earnings call, Gaming revenue clocked in at $1.72B (53.6 percent of the total), while revenues from sale of AI-optimized data center hardware were $701M (21.8 percent of the total). That’s impressive growth by the segment when you consider that it accounted for just 13 percent of Nvidia’s revenue two years ago.
Nvidia is reputed to own nearly 90 percent of the enterprise-grade AI chip business. And the segment is likely to only get bigger and better with time.
There are literally hundreds of applications where cloud-based AI-inference can be deployed. Microsoft Inc. uses Nvidia’s AI chips to power Bing Visual search capability that allows users of the search engine to search for pictures with similar imagery or text. SAP Inc. uses the smart GPUs to create visually appealing ads during sporting events in real-time that can be adjusted on the fly. In fact, Nvidia has revealed that more than 1,200 companies are using its inference platform, including Facebook Inc., Amazon Inc., Alibaba Group, Tencent Holdings, Hi Vision and Baidu Inc. among others.
Last month, UBS raised its Nvidia price target from $266 to $285 citing a boom in mass facial recognition project by the Chinese government.
The firm sees the trend as a $5-billion opportunity for Nvidia by 2020 as the Beijing government doubles down on the drive to monitor its citizens. Nvidia is well-placed to capitalize with its Metroplis Platform, which allows law enforcement agencies to analyze video feeds and cities to provide traffic monitoring.
Gaming Business Going Strong
Nvidia’s gaming business, however, continues to hog the limelight due to its sheer size and significance for the company.
Nvidia ‘s major rival in the gaming console business is AMD. Nvidia supplies gaming APUs for Nintendo’s wildly successful 7th generation Switch console while AMD does the same for Sony’s PS4 and Microsoft’s Xbox.
Nvidia has been doing a great job persuading gamers to use its higher-end GPUs such as GeForce graphics. The result: the company’s gaming GPU sales are not only increasing in volumes but in price points as well.
Wall Street’s Raymond James has encouraged investors to buy NVDA stockbefore the company launches its next generation of gaming chips in October.