The big picture has been the same all year - NVIDIA is about artificial intelligence - a technology market that demands high levels of performance and is simultaneously a very large market touching nearly every economic activity.
At the European GTC in Munich, the focus was on self-driving cars. At the Japanese GTC in Tokyo, the focus was robotics. Both areas are booming due to NVIDIA's work in artificial intelligence.
NVIDIA states clearly that the company is about accelerated computing. They invented the Graphics Processing Unit, GPU, in 1999 and this technology has transformed the world of parallel computing. One after another, parallel computing applications leveraged this new parallel-processing powerhouse: simulations, genetics, finance, ray tracing, and more. Over the last several years, the processing power of GPUs has been applied to training neural networks for artificial intelligence. This latter area blows open NVIDIA's market for GPUs. The combination of high-performance computing demands in AI and a massive market size has the NVIDIA CEO aligning 150% of the company's resources behind succeeding in this technology.
The company's advantage is that the AI market is also aligning itself around GPU computing. This mutual market-hug means that NVIDIA's most powerful GPU architecture ever released, the Volta architecture, is being embraced by AI researchers, startups and all of the CSPs providing computing platforms for AI development. Team Green is pushing solutions that deliver GPUs in desk-side workstations - NVIDIA labels the DGX Station a "Personal Super-computer" and the system prices can run into six-digits.
Essentially every workstation system builder worldwide is offering the DGX Station to the market. The system has been custom-designed by NVIDIA to address Deep Learning (DL) applications. It is tuned with 4 NVIDIA Volta GPUs using the fast NV-Link interconnect that provides 200 GB/S transfer per GPU and is five times faster than the current PCIe Gen-3 interconnections. The workstation has 256 GB of main memory, a 20 core Intel Xeon CPU, and an entire software stack targeting DL applications.
On the server side, essentially every server OEM is offering Volta-based computing systems for artificial intelligence. And the same is true with every cloud-service provider. Given NVIDIA's GPU coverage with Volta in desk-side systems, servers, and the cloud, the company is covering every avenue through which AI companies might access these high-performance GPUs.
Yes, the company is aligned around artificial intelligence. Yet, NVIDIA is more than an AI company today. As a company, they eat and sleep artificial intelligence, however, they are more than just an AI company as they deliver fully integrated platforms for computing - here the DGX Station. But likewise, they deliver fully integrated solutions for Automotive OEMs to develop A-based self-driving vehicles. The company provides an interactive VR environment for immersive design named the NVIDIA Holodeck - and the Holodeck is a tool used for training Robots in in a virtual world before they venture out into the real world.
The PW Perspective
More than graphics. More than GPU computing. More than AI. More than even the company's own tagline of being a leader in accelerated computing. NVIDIA, as is evident at any of the worldwide GTC events, is building solutions to accelerate the businesses of their clients. The hardware and software solutions from NVIDIA make is easier for customers to leverage the power of NVIDIA GPUs whether that is in Deep Learning, HPC, virtualized graphics, or virtual reality. Their market is booming, and their level of innovation is booming, too.
