The Internet of Things (IoT) includes many things outside of computers. It certainly includes robots and drones, and both categories could bring untold billions of dollars for those companies that capture the trends just right. NVIDIA Corp. (NASDAQ: NVDA) hopes to be one of those winners.
On Tuesday, NVIDIA showed just how it wants to win in drones and robots. The company unveiled the NVIDIA Jetson TX1 module, a credit-card sized module that uses machine learning so that future smart machines and autonomous machines can learn. The company specifically noted that the tiny supercomputers would bring artificial intelligence (AI) for robots and drones.
NVIDIA said in its release that this aims to address the challenge of creating a new wave of millions of smart devices. This would include drones that self-navigate rather than being flown by remote control. The company also pointed to compact security surveillance systems that can identify suspicious activity while they scan crowds. For robots, that was said to include tailoring tasks to individuals’ habits rather than just performing preset tasks.
All of this will come from machine learning, computer vision, navigation and more. NVIDIA’s selection of the name “Jetson” pretty much says it all. Maybe the name “Skynet” just wasn’t what it wanted to communicate to the public.
The company indicated that its Jetson TX1 module is the first embedded computer designed to process deep neural networks. This was described as computer software that can learn to recognize objects or interpret information (machine learning) that can be used to perform complex tasks like recognizing images, processing conversational speech, or self-navigation.
The company noted that the Jetson module has 1 teraflop of performance and will draw much less power than comparable modules and systems on a chip elsewhere. This Jetson TX1 is also built into a developer kit to enable hobbyists and professionals to develop and test highly advanced autonomous devices — to ease the transition from development to manufacturing and production.
So, what is made of? NVIDIA showed that the graphics processing unit for 1 teraflops is from a 256-core Maxwell architecture-based, and the central processing unit is a 64-bit ARM A57 CPUs. The video capability is 4K and the camera is said to support 1,400 megapixels per second. The size of 50X87 millimeters is said to be slightly smaller than a credit card.
The NVIDIA Jetson TX1 Developer Kit can be pre-ordered on November 12 for $599 in the United States, and this module will be available in early 2016 at a suggested price of $299 for quantities of 1,000 or more.
Is there any confusion over why so many people are worried about the power and growth in capabilities of robotics, drones and other fixed or mobile systems?
NVIDIA already has a $16.7 billion market capitalization. At $31.00, its stock has a consensus price target of $30.65 and a 52-week range of $18.94 to $31.94. The Thomson Reuters consensus estimates project revenue growth of 5% in 2015 to $4.92 billion and growth of 6% to $5.21 billion next year.