Generative AI is gaining significant traction, with 65% of companies now employing it, as highlighted by a recent McKinsey study. A partnership between Verizon and Nvidia aims to address the challenge of integrating generative AI in industrial settings through private 5G networks and edge computing. This collaboration promises to facilitate real-time AI applications on-site, catering to sectors like logistics and healthcare while ensuring high performance, security, and control over data.
Generative AI Takes Center Stage
Two years after the debut of ChatGPT, generative AI is capturing widespread attention. Few technologies have sparked such interest and witnessed rapid adoption like this one. Numerous projects are emerging within organizations aiming to harness the potential of large language models (LLMs). According to a recent McKinsey study, a staggering 65% of companies are now regularly utilizing generative AI, a significant increase from just ten months ago.
Enhancing Industrial Applications with Private 5G
However, a key challenge lies in integrating and securing generative AI as close to users as possible, especially within industrial environments or logistics hubs. Training and inferring LLMs requires substantial computational resources, which are powered by cutting-edge graphics processors (GPUs).
This need has led to a strategic partnership between two American giants: Verizon, a telecom operator, and Nvidia, a leading chipmaker. Verizon contributes its knowledge in private 5G networks and mobile edge computing (MEC), while Nvidia offers its edge computing platform from the VAI Enterprise suite, along with specialized micro-services for inference.
Although we will have to wait until 2025 for the first demonstrations, this collaboration appears promising. It enables real-time execution of workloads and AI applications directly on-site, eliminating the necessity to rely on cloud resources or transfer data outside the organization.
This innovative approach is likely to attract interest from industries like logistics, distribution, and healthcare. The AI-enhanced private 5G platform is designed for immediate deployment in a plug-and-play format, allowing third-party developers to innovate rapidly while responding to the evolving demands of AI and connectivity.
Beyond just large language models and visual processing, this platform is equipped to manage high-performance applications, including computer vision, augmented reality (AR), virtual reality (VR), extended reality (XR), autonomous mobile robots (AMR), automated guided vehicles (AGV), and IoT solutions.
The capabilities of private 5G are essential for these applications, as they provide guaranteed service quality regarding bandwidth and latency, coupled with heightened security due to an independent infrastructure from public networks.
With dedicated bandwidth, ultra-low latency, robust security, and greater control over data and applications, a private 5G network creates an ideal environment for executing demanding and sensitive AI workloads.
In France, the adoption of private 5G has been relatively slow. Notable examples include Alcatel Submarine Networks and ArcelorMittal, which have shared their initial experiences with process digitalization and predictive maintenance. The introduction of standalone 5G and features like network slicing—allowing the division of virtual networks with tailored service levels—should further enhance the attractiveness of private networks.