On April 12-16, NVIDIA virtually held its annual GPU Technology Conference (GTC), showcasing the latest advancements in natural language processing, robotics, accelerated computing, smart networking, collaboration, and much more. GTC was a mix of industry and vision, with artificial intelligence (AI) as the overarching theme. NVIDA is among the vendors developing products and software stacks aimed at democratizing AI, so it’s accessible to companies of all sizes.
While NVIDIA hosts this event several times a year, the April 2021 version of GTC will be remembered as one of change. Historically, NVIDIA has been thought of as a GPU company that appeals to developers, gamers and the automotive industry. Over the past few years, NVIDIA has been innovative and acquiring at a furious pace and the results of that activity was highlight at this year’s event.
One of the big shifts for NVIDIA is that it’s not just GPUs any more. While the company is the de facto standard in graphics processing units, it’s made signficant strides in the area of data processing units (DPUs) and CPUs. This will create broader appeal to enterprises and during his keynote, Founder and CEO Jensen Huang outlined NVIDIA’s vision, sharing new products and enterprise-focused solutions launching this year.
To help understand what these announcements mean for enterprises, I recently chatted with Maribel Lopez, technology industry analyst and strategic advisor at Lopez Research. Highlights of my ZKast video with Lopez, done in conjunction with eWEEK eSPEAKS, are below:
Omniverse is Not Just for Gaming
NVIDIA is finding practical use cases for its Omniverse virtual platform, which will be available as an enterprise product later this year. The vendor has dubbed Omniverse as a “metaverse for engineers” that enables photorealistic 3D simulation and collaboration. In his keynote, Huang demoed how teams can work together to create a scene in an Omniverse virtual environment.
Such advancements will help move us closer to simulating and capturing information that can then be used to collaborate virtually. In present-day context of COVID-19, organizations are looking for ways to reconstruct buildings and safely return to work. Adding intelligence, automation, and robotics to spaces can address those challenge. Omniverse will be an important technology, as organizations consider new ways of designing physical structures.
Conversational AI is Evolving
Chatbots can communicate with humans using facial expressions and contextual awareness thanks to innovations in conversational AI. NVIDIA’s GPU-accelerated application framework Jarvis, for example, allows organizations across different industries to use video and speech data to build conversational AI services like 3D chatbots. Developers can employ the Jarvis GPU-accelerated software stack and tools to create conversational AI apps and services, instead of building from the ground up.
The big news NVIDIA announced at GTC is general availability of Jarvis, which means the technology can be widely used to automate customer support and provide supplemental services in industries like healthcare, retail, and financial services. NVIDIA also announced a graphical user interface (GUI)-based framework, called TAO (train, adapt, optimize). TAO encapsulates the entire workflow of adapting and lets multiple parties collectively train a global model through data diversity.
Chatbots have come a long way, but preserving context is still problematic. We’ve all gotten frustrated at one point or another talking to automated systems that don’t know what we want. By combining tools like Jarvis with personalized data, organizations can create customized lexicon. Additionally, the ability to translate between different languages and control pitch or tempo—features offered in the latest version of Jarvis—will greatly improve the quality of interactions between humans and digital assistants.
NVIDIA is Making Enterprise IT Push
Although Nvidia isn’t well known in enterprise IT, the vendor is making a major enterprise move this year through its partnership with VMware. NVIDIA is bringing AI to organizations that want to run AI workloads on the same infrastructure they use for traditional business applications. Several high-volume enterprise servers from top manufacturers are now certified to run the NVIDIA AI Enterprise software suite, which is exclusively certified for VMware vSphere 7, a leading virtualization platform.
Together with VMware, which has a deep enterprise IT background, NVIDIA recognizes there’s a base-level computing challenge that needs to be addressed and it’s pushing down the stack. NVIDIA is coming up with ways to create a better data center GPU, which could be used not only by hyperscalers like Amazon and Google, but also by enterprise IT.
AI is Needed to Combat AI Threats
Many organizations are stuck in the past when it comes to security, trying to use traditional ways to solve problems that are originating from AI-based systems. In fact, security is the number one issue that every IT leader is focused on, according to a Lopez Research study of top concerns. Security is like a layer cake that more products are being added to constantly, but it’s still not working. What organizations want to see is an AI-based approach to security.
NVIDIA is determined to inject AI into every aspect of the enterprise, hence, it’s bringing AI-driven automation to cybersecurity. Morpheus is NVIDIA’s new cloud-native cybersecurity framework that uses AI and machine learning (ML) to detect threats. Morpheus deals with security breaches as they happen by inspecting all packets in real time.
AI is Revolutionizing Data Centers
Complex systems are difficult to put together and no one has done a better job of simplifying this task than NVIDIA. The vendor is venturing into new territory and that’s CPUs. NVIDIA has developed its first data center CPU, an Arm-based processor that promises to deliver ten times the performance of today’s fastest servers. The Grace CPU leverages NVIDIA’s NVLink multi-processing technology to meet the computing needs of advanced apps like natural language processing, recommender systems, and AI supercomputing.
NVIDIA also unveiled the BlueField-3 DPU for AI and analytics workloads. BlueField-3 is designed specifically for organizations that run massive data centers, offering 10 times greater offload capability than the company’s BlueField-2 DPU.
Coupled with Grace’s launch, NVIDIA now has a third foundational technology for computing. As a three-chip company, NVIDIA plans to re-architect the data center using AI. No matter how an organization chooses to put a computer together—whether it’s workstations or supercomputers—NVIDIA has a solution for them.