Networking for AI Workloads & AI for Networking

The Future Takes Shape at The AI Networking Summit

In an era where Artificial Intelligence (AI) is not just a buzzword but a foundational technology reshaping industries, the AI Networking Summit at ONUG stands as a pivotal event. This year, the Summit delves deep into the symbiotic relationship between AI and networking, focusing on the crucial components of Chips and Infrastructure within the AI ecosystem.

The AI Ecosystem: A Comprehensive Landscape

The AI ecosystem is a complex web that includes Chips, Infrastructure, Frameworks, and AI Applications. Each component plays a vital role in the seamless functioning and evolution of AI technologies. Chips, the cornerstone of processing power, are continuously advancing to handle complex algorithms and data-intensive tasks; Nvidia dominates this space, especially with the Grace Blackwell GB 100 superchips, but there are many others entering such as AMD, Intel, Broadcom, etc. Infrastructure, the backbone of AI, provides the necessary hardware and network capabilities to support these demanding workloads. Every networking, security and computer company will launch AI infrastructure products this year. Frameworks offer the tools and libraries needed for developing AI models. This part of the ecosystem is quickly moving to commodity status thanks to open-source solutions. The AI Applications represent the end-use cases that bring tangible benefits to businesses and society and will be the largest number of providers with the biggest amount of investment. 

At the AI Networking Summit, our focus is on the Infrastructure and Chips segments, which are critical for supporting the burgeoning needs of AI workloads and the development of AI-driven networking solutions for corporations. Within AI infrastructure there are two main categories: 1) AI infrastructure that is being built to support AI workloads; this is the province of many hyperscalers, large enterprises and governments, and 2) compute infrastructure that is becoming AI enabled; this infrastructure supports legacy and AI workloads. 

AI Infrastructure: Powering Private AI Deployments

In the realm of private AI deployments, the infrastructure must be robust, scalable, and flexible. Enterprises are looking for solutions that can handle the intensive requirements of AI workloads, from training complex models to deploying real-time analytics. This necessitates a shift towards specialized hardware, including GPUs, TPUs, VPUs, DPUs, and advanced storage solutions that can manage the velocity and volume of AI data. Networking infrastructure, too, must evolve to ensure low latency and high bandwidth, enabling seamless data flow across components. Data centers are being reimagined and built for special AI workloads in the cloud providers as well as large corporations such as every large financial, healthcare provider, manufacturer, etc.   

The Summit will explore these infrastructure needs in detail, discussing how organizations can build or adapt their IT environments to harness the full potential of AI. This includes insights into the latest trends in data center technologies and edge computing in facilitating AI deployments.

AI’s Impact on Networking and Security

As AI becomes increasingly embedded in networking and security protocols, the relationship between applications and infrastructure is undergoing a significant transformation. Networking will not only provide a connectivity service, but a wide range of value-added services to applications. AI-driven networking promises to optimize traffic flow, predict and mitigate network anomalies, and enhance security protocols through intelligent threat detection and response.  Envision an infrastructure under the direction of an AI policy manager that presents connections for users and hosts that include security posture, latency, cost, etc., that address stated user experience goals.  

The integration of AI in networking and security infrastructure not only promises enhanced efficiency but also introduces adaptive mechanisms that can anticipate and react to changing network conditions and security threats. This dynamic interplay between AI and networking is paving the way for more resilient, efficient, and secure IT ecosystems or a more trusted infrastructure with a simpler life-cycle management model.  

The Dawn of AI-Native Products

The Summit will also shine a light on the wave of AI-native products set to revolutionize the market. From PCs to routers, switches, firewalls, load balancers, cloud services, and new product categories. These products are designed with AI at their core. Embedding AI directly into these devices and services promises unprecedented levels of automation, performance optimization, productivity, and security. For example, Juniper Networks’ launch of its AI-native networking products is a great example of new products entering the market. At the AI Networking Summit there will be many new products announced that embed AI at some level.  Be on the lookout for a new AI firewall that eliminates rules and rule management, that is all the toil associated with legacy firewalls.

The AI networking and security products I’ve reviewed are falling into three categories or stages of products. 

  • Generative AI Networking and Security Products. Stage 1: Product utilizes an LLM for operational support
  • Pre AGI networking and security products. Stage 2: Embeds AI within the product, detailing its unique value proposition usually in life-cycle management
  • AGI networking and security products. Stage 3: Part of a suite of, or connected products sharing AI information and data, offering unique life-cycle management value and presenting a new service to applications it serves; this is when infrastructure transforms into AI infrastructure 

AI will be embedded into every single networking and security category of products. Think of it like when the internet started to become part of enterprise infrastructure; every product needed to have internet capability in some way. The same is happening with AI. All the new SASE/SSE/NaaS products will be enhanced with AI as well as their services too. Cloud provider infrastructure services will be AI enabled. Security products from SIEMs, SOARs, Data Lakes, Firewalls, Intrusion Detection Systems (IDS) and Intrusion Prevention Systems (IPS), Cloud Security, Secure Web Gateways, etc., all will be AI enabled.  Network routers, switches, load balancers, etc will be AI enabled.  Look for AI routers, AI firewalls, AI SIEMS, AI switches this year.  

The value proposition of AI-native products lies in their ability to learn and adapt. These products can enhance operational experiences through personalization, improve operational efficiency by automating routine tasks, and bolster security through proactive threat intelligence. The summit will delve into these value additions, offering a glimpse into the future of AI in networking and beyond.

Looking Ahead

As we stand on the cusp of a new era in technology, the AI Networking Summit at ONUG is more than just an event; it’s a confluence of ideas, innovations, and insights that will define the future of AI and networking. Join us as we explore the evolving landscape of AI infrastructure, witness the transformative impact of AI on networking and security, and unveil the next generation of AI-native products. Together, we are shaping a future where AI and networking converge to create a smarter, safer, and more connected world. See you at the dawn of the new world at the AI Networking Summit at ONUG Spring, hosted by FedEx, in wonderful downtown Dallas at the Wolfgang Puck Union Station. Get your pass here.  

Author's Bio

Nick Lippis