Matt Swinden
Director Digital Connectivity, Business, BT

Since the launch of ChatGPT, the AI hype cycle has switched to overdrive. According to some, it will either save the world or end it. In reality, AI has been evolving for decades. The difference now is that a perfect storm of application development, hardware, and the cloud will lead to widespread adoption. A new wave of digital transformation is imminent. The question is: how will an organisation’s network cope?

AI may be making headlines, but it’s not the only digital application organisations are looking to the cloud for. Successive waves of digital transformation mean multinational organisations now typically use 50 Software-as-a-Service (SAAS) solutions. These are applications hosted in the public cloud or a SaaS provider’s own data centre, then delivered to users via an organisation’s network.

Furthermore, 50% of organisations expect to add more SaaS within a year. With SaaS on the rise, it is important to remember that the productivity gains it brings will be dependent on predictable, secure, and resilient connectivity. Like AI, SaaS demands real-time data processing and contextualisation to enable quicker decision-making, which can accelerate productivity and efficiency. It requires authenticated and trusted connections from the organisation to wherever the application is hosted. As the number of SaaS solutions an organisation uses grows, the complexity of managing all these connections becomes a significant challenge.

Returning to the subject of AI, hyperscalers today offer cloud infrastructure that can scale with Generative AI (GenAI) applications and deal with unpredictability and pace of change. But the same flexibility and scale are also required from the network. This can be broken down into three areas that organisations should consider when developing their AI strategies:

• Data diversity—AI relies on diverse data inputs of various data types, sources, and sizes, which are driven by an organisation’s ‘AI data governance framework.’ However, that diversity creates unpredictable, often ‘bursty’ network traffic. Cloud computing offers flexibility and pay-as-you-use models, yet most organisations’ networks have not kept pace. This can hinder GenAI adoption and transformation efforts.
• Data quality—GenAI strategies involve balancing quality and quantity. Large datasets are used initially to train GenAI models quickly, but to achieve higher quality outputs, GenAI models require smaller datasets with greater fidelity for quicker, more accurate insights. These datasets are in many different locations across many organisations and are better referred to as ‘wide data.’ In an AI-enabled world, wide data needs to interoperate seamlessly.
• Data security—LLMs (Large Language Models) sit at the core of GenAI applications. They process sensitive and confidential data, making data security crucial. The need to prevent unauthorised access and data breaches is essential. Security measures such as encryption and access controls are vital for maintaining data integrity and reliability.

In the digital world, where sharing data between organisations is the new norm, traditional networks lack flexibility and scalability and struggle to overcome the challenges of complex configuration, interoperability, latency, and performance. Moving forward, future networks must offer the same pay-as-you-use commercial flexibility as the hyperscalers.

To access diverse datasets at scale, dense and secure global interconnectivity is essential. The telco edge is the ‘Goldilocks location.’

Goldilocks locations are ‘just right’ for harnessing AI’s power. In the context of networking, this means locations that allow the best balance between factors such as:

• Latency and bandwidth—ensuring data travels quickly;
• Security and accessibility—providing robust security while maintaining ease of access; and
• Cost and performance—achieving high performance without excessive costs.

Deploying cloud services that host AI or SaaS at the edge of a network means interconnectivity is much easier with partners, and there is good access to third-party data sources, making it a perfect location to unlock digital potential.

At BT, our transformative network-as-a-service (NaaS) platform, Global Fabric, will make it easier and quicker for organisations to securely connect their people, partners, and devices to apps and digital services—including AI and SaaS—hosted across multiple clouds. Organisations benefit from scalable, secure, high-capacity, and resilient connectivity pre-integrated into the world’s leading cloud locations and SaaS providers, ready to meet the growing and complex demands of breakthrough AI technologies. Global Fabric’s flexibility will be unprecedented. With legacy networks, setting up connectivity or making changes can take weeks. With Global Fabric, it happens in an instant, helping manage unpredictable, AI-driven spikes in traffic.

Global Fabric’s design is based on the idea of ‘interconnectivity’—connecting to third-party clouds, solutions, and partners. This contrasts with the internally focused thinking of the past, where networks primarily ‘intra-connected’ an organisation’s users and systems. In the digital world, the ‘intra’ model is no longer fit for purpose.

To thrive in a digital world, multinational organisations like NATO need a seamless, secure, and interconnected ecosystem. The more interconnections, the more it drives the organisation forward. For organisations, flexible, robust, and secure interconnectivity will be vital to onboarding AI and other cloud-hosted digital solutions such as SaaS. This, in turn, will help the empowered leaders to make faster, better-informed decisions.