These days, just about every industry feels the presence of artificial intelligence (AI), and networking is no exception. It’s hard to imagine a network—whether it’s a small office setup or a vast global telecom system—where AI wouldn’t bring improvements.
Mark Düsener, the chief technical officer at Swisscom, sums it up well. He points to his partnership with Cisco-owned Outshift to introduce agentic AI into their operations, aiming to cut service disruptions, minimize downtime, and enhance customer experiences. Essentially, AI boosts efficiency, reliability, and user satisfaction. But implementing AI isn’t as simple as flipping a switch. To truly harness its benefits, you have to rethink networking strategies.
Let’s talk about Nvidia. It stands out in the AI landscape, playing a critical role in shaping technology for businesses. CEO Jensen Huang recently pointed out that we’re entering a new age of generative AI. He emphasizes that companies now need to tap into their wealth of data to create smart, data-driven operations. Huang believes that every business will need to become an intelligence manufacturer, digitizing their knowledge into AI systems to improve productivity and service delivery.
Swisscom is taking this approach to heart. With over six million mobile users and thousands of antenna sites, they face the challenge of upgrading their infrastructure without interrupting service. By collaborating with Outshift, they’re using AI to reshape customer experiences and meet user demands.
But you don’t have to be a giant corporation to benefit from AI in networking. Smaller businesses can now access tools that level the playing field, allowing them to scale without the burden of complex IT setups.
Now, let’s look at a broader view. Swisscom and Outshift exemplify that making AI work requires a robust infrastructure for secure communication. David Hughes from HPE Aruba Networking highlights the significance of two facets: “AI for networking” and “networking for AI.” He points out that while AI can empower IT staff to manage their growing workloads, networks also need to be built to support AI’s demands—especially as businesses increasingly require capabilities like real-time data processing.
Bastien Aerni, from GTT, emphasizes that companies need to rethink their networks to facilitate rapid data processing associated with AI. For AI investments to pay off, the network architecture must handle significant data loads swiftly and efficiently. This means aiming for low latency and high bandwidth, particularly as AI processes become more decentralized and edge computing grows.
BT’s recent challenges with network overload during high-traffic events underline the urgency of these issues. CTO Colin Bannon stresses that as AI reshapes various workflows, a strong network infrastructure becomes essential. He believes networks must be not only scalable but also agile, with features that allow for real-time monitoring and quick adjustments to manage traffic.
As AI models evolve, the focus is shifting from generic large models to smaller, more specialized ones. For instance, NTT Data has developed Tsuzumi, a compact language model designed for edge deployment. Tom Winstanley, their CTO, explains that this model aims to minimize network strain, tackling issues like privacy and sustainability while still providing powerful performance for specific tasks.
But even though the tech community is geared up to solve these challenges, many businesses are still hesitant. A recent study by Expereo found that while a vast majority of UK leaders see AI as vital for future success, there are significant barriers—like poor infrastructure and unrealistic expectations—that could hinder progress. Many organizations feel unprepared to support AI initiatives, with nearly half reporting that their current connectivity isn’t up to the task.
Expereo CEO Ben Elms reinforces the need for careful planning and realistic goal-setting. Successful AI integration hinges on robust network infrastructure, ensuring that performance remains consistent.
In essence, simply activating an AI system won’t yield guaranteed results. Without the right network support, your AI ambitions could be stunted.