Edge computing is hot. In fact, there isn’t one edge but several. There’s the connected device edge (mobile, PCs, and IoT). The trend in edge computing is to analyze and act on data closer to the source where it was created. The edge computing concept expanded as more devices were connected, 5G network deployments started to roll out and companies wanted faster access to insights.
Depending on whom you’re speaking with, there are three to four different types of edge computing. In short, the edge spans from a small sensor to large regional data centers. The first edge includes any connected device computing (mobile, IoT, PCs and connected stationary equipment). The second edge is the wide-area network device from a company’s building to the outside world. A third edge represents the first network operator point of presence (POP), be that a cell tower or data center. The POP edge is increasing as providers build more facilities closer to the user. Some even consider the first large regional points of presence an operator as an edge.
Regardless of which edge we’re discussing, device makers and enterprise IT buyers are looking for platforms to support delivering AI, IoT, cloud gaming, augmented reality and 5G capabilities at the edge. AI at the edge is a new market requirement. Computing companies such as NVIDIA are racing to drive innovation, performance and AI support at every level of edge computing.
As a precursor to the Mobile World Congress Los Angeles show, NVIDIA took the stage to discuss its role in the ever-expanding edge computing marketplace. Given the buzz around 5G, it’s no surprise that NVIDIA would be talking about how the company would support edge computing from software through carrier networks. CEO Jensen Huang started his keynote with an Infinity Wars reference. Like the six gem stones in Infinity Wars, Nvidia noted that there were also six critical technology factors for success that include items its known for such as GPUs and AI. However, he added 5G, Smart NIC, Cloud Native, EGX and a software-defined 5G Radio Access Network (RAN). CEO Jensen Huang’s keynote highlighted the following:
• A platform that compliments GPU servers with software. Huang said the world needs an edge supercomputer The NVIDIA EGX is an accelerated computing platform that enables companies to perform low-latency AI at the edge to process continuous streaming data. NVIDIA’s EGX platform combines NVIDIA CUDA-X software with NVIDIA-certified GPU servers, and the Mellanox Smart NIC. NVIDIA described how EGX supports a wide range of compute-intensive applications for running 5G networks, intelligent video analytics, automated optical inspection, cloud-based gaming services, and augmented and virtual reality services.
• Software to support specific use cases. Success in IoT and 5G requires a horizontal platform that offers strong focus on vertical software stack to use cases. NVIDIA plans to drive support for it’s solutions by offering industry-specific software stacks. These include NVIDIA Aerial, a CUDA-X software developer kit that enables telecommunications companies to develop high performance, efficient and completely virtualized 5G radio access networks. For example, NTT is using EGX in its data centers to develop new AI-powered services in remote areas through its broadband access network. oy a wide range of AI applications at the edge.
Additionally, NVIDIA’s GeForce NOW cloud gaming platform, allows telcos and others to deliver PC-class gaming across a variety of devices. NVIDIA Metropolis is an IoT application framework. For example, BMW is using with EGX edge servers with Metropolis to enable intelligent video analytics and automated inspections by analyzing data from cameras and sensors in its manufacturing line. BMW’s use case is for fit and finish which is a highly detailed and complex AI application. NVIDIA also spoke of Jarvis, which is a multi-modal AI SDK that runs on EGX. Jarvis comes with pre-trained models that will help developer kickstart efforts to create domain specific applications.
• Hallmark customers across verticals to highlight momentum. At this event, the company spoke of its progress and early adopters of the platform. Companies such as Walmart, Procter & Gamble, FoxCon, and the cities of San Francisco and Las Vegas are using the platform for a wide range of industry-specific applications. For example, Samsung Electronics is using AI at the edge for highly sophisticated semiconductor design and manufacturing processes.
• Big partner announcements to show support for its vision. Telecom industry stalwart Ericsson and Open Source Guru Red Hat (now part of IBM) each announced new collaborations with NVIDIA to bring GPU acceleration to the telco edge. Ericsson took the stage to announce it was working with NVIDIA to create a software-defined 5G RAN.
Microsoft’s also been making a significant play at the edge with products such as Azure Internet of Things, a microcontroller (MCU) called Azure Sphere and a device-specific OS called Windows 10 IoT Core. Today Microsoft announced that it has integrated EGX into its Azure Data Box Edge appliance and is working closely with NVIDIA to integrate EGX with Microsoft Azure IoT. To round out the portfolio, NVIDIA shared that it’s collaborating on EGX, including Cisco, Dell Technologies, Hewlett Packard Enterprise and VMware.
Technology moving us closer to edge AI
After years of discussion about companies deploying IoT, we’re finally seeing companies connecting sensors and collecting data in areas such as manufacturing, retail, and smart city applications. These new data sources are fueling the need for next-generation analytics and AI. Meanwhile, 5G promises to unlock a wide range of new use cases by delivering low latency and high bandwidth. But to deliver on the promise of services at scale with low latency, companies need robust computing and software at all layers of the stack.
NVIDIA has a tremendous opportunity to participate and drive innovation at every level of the edge. It already has a healthy EGX ecosystem that includes more than 100 technology companies worldwide, from startups to established software vendors, cloud service providers, and global server and device manufacturers. Today’s announcements helped define what edge markets NVIDIA will participate in, how it will address the market and its success thus far. NVIDIA made it clear that is plans to dominate the next generation of edge computing. From the myriad of products and logos that graced the stage today, it’s clear NVIDIA is off to a good start.