Published: November 10, 2026 Author: Saf Malik Location: Capacity Europe
Executive Overview
The rapid, relentless integration of Artificial Intelligence (AI) across global industries has triggered a fundamental paradigm shift in telecommunications and enterprise IT. No longer confined to localized servers or experimental testbeds, AI workloads are now heavily distributed across complex ecosystems encompassing hyperscale data centers, multi-cloud environments, and geographically dispersed edge regions. This architectural decentralization has introduced unprecedented performance demands on global network infrastructure.
At the prestigious Capacity Europe conference, Saf Malik sat down with Michael Wheeler, Executive Vice President (EVP) at NTT DATA, to examine the profound ways AI is reshaping the underlying fabric of international connectivity. Their comprehensive discussion peered deep into the engine room of modern telecommunications, addressing how high-capacity IP networks must evolve to support seamless data transmission, surging regional traffic, and the increasingly sophisticated threat landscape targeting critical infrastructure.
As enterprises stake their futures on real-time data processing and autonomous digital operations, expectations around network resilience, redundancy, and availability have reached an all-time high. Wheeler offered critical insights into how network operators can balance these aggressive performance demands with robust cybersecurity postures, leveraging intelligent automation to secure, manage, and scale the next generation of global networks.
Detailed Chronology: The Evolution of AI-Driven Infrastructure
To understand the current pressures facing global networks, one must examine the rapid trajectory of digital transformation over recent years. The journey from traditional cloud computing architectures to AI-native, hyperscale networks has unfolded in rapid phases, each introducing unique engineering challenges.
Phase 1: The Cloud Consolidation Era (2018–2022)
For years, global networks were optimized primarily for enterprise cloud migration, SaaS adoption, and streaming media distribution. Traffic patterns were largely predictable, flowing on north-south vectors from end-users to centralized data center hubs. During this period, infrastructure investments focused heavily on raw bandwidth expansion, submarine cable deployment, and content delivery network (CDN) edge caching. Network resilience was measured by uptime SLAs and traditional path redundancy, with failover mechanisms operating on time scales of seconds or minutes.
Phase 2: The Generative AI Explosion (2023–2025)
The public debut and hyper-scaling of Large Language Models (LLMs) and generative AI applications shattered traditional traffic assumptions. AI training and inference demand massive, parallelized data movement. Unlike traditional web traffic, AI workloads require continuous, high-throughput, low-latency east-west communication between thousands of specialized GPUs spread across multiple facilities. Network architectures strained under the weight of hyper-bursty data flows, forcing operators to rethink backbone capacity, latency budgets, and interface speeds.
Phase 3: The Distributed AI and Multi-Region Reality (2026 and Beyond)
As Wheeler highlighted during his discussion at Capacity Europe, the industry has now entered the era of distributed AI. Organizations no longer rely on a single data center for their AI pipelines; instead, workloads are fragmented and distributed dynamically based on power availability, regional compliance mandates, and compute efficiency. This distributed model requires high-capacity IP networks that act as an invisible, highly responsive nervous system. High-performance computing clusters must communicate instantaneously across oceans and continents, making network agility a primary competitive differentiator for global service providers like NTT DATA.
Supporting Context & Metrics: The Scale of the Challenge
The transition toward AI-dominated network traffic is underpinned by staggering macroeconomic and technical metrics that illustrate why traditional network paradigms are no longer sufficient.
+-------------------------------------------------------------------+
| THE AI NETWORK TRAFFIC MULTIPLIER |
+-------------------------------------------------------------------+
| Traditional Enterprise Traffic | [======] 1x |
| Standard Cloud Migration | [===========] 2.5x |
| Hyperscale Generative AI (Train) | [=======================] 6x |
| Distributed Multi-Region AI | [===========================] |
+-------------------------------------------------------------------+
1. Exponential Bandwidth Consumption
Industry analyses indicate that inter-data center traffic generated by AI workloads is growing at a compound annual growth rate (CAGR) exceeding 45%. This growth outpaces standard internet traffic expansion by a factor of three. Backbone networks that previously required generational upgrades every five to seven years are now experiencing capacity exhaustion cycles in as little as 18 to 24 months.
2. The Resiliency Imperative and the Cost of Downtime
In an AI-driven operational environment, network downtime is vastly more expensive than in the traditional web era. When an e-commerce storefront goes offline for ten minutes, revenue is paused. When a distributed AI model training run—costing millions of dollars in compute resources—is abruptly interrupted by a network blip or latency spike, entire checkpoints can be corrupted, forcing hours or days of computational rollback. Customer expectations have consequently shifted from "five-nines" (99.999%) availability to demand-driven, zero-packet-loss guarantees tailored specifically for high-performance computing (HPC) environments.
3. The Escalating Threat of Advanced DDoS Attacks
The expansion of high-capacity infrastructure has unfortunately coincided with a dramatic evolution in cyber threats. During his conversation with Saf Malik, Wheeler pointed to the growing sophistication of Distributed Denial of Service (DDoS) attacks. Modern threat actors leverage automated, AI-assisted botnets capable of launching volumetric and application-layer attacks exceeding multi-terabits per second. These attacks are designed not merely to crash websites, but to intelligently probe, exhaust, and destabilize core routing infrastructure and peering links, making automated mitigation a mandatory pillar of modern network defense.
Official Insights & Perspectives: Michael Wheeler on the Future of NTT DATA
Providing a strategic glimpse into how a tier-one global player navigates these complexities, Michael Wheeler emphasized that modern infrastructure providers can no longer afford to be mere "dumb pipes." The convergence of AI and mission-critical enterprise workloads demands a proactive, intelligent approach to network architecture.
"As AI workloads become increasingly distributed across data centres, cloud environments and regions, the role of high-capacity IP networks is no longer just about moving data—it is about enabling seamless connectivity and acting as the foundational bedrock for future global growth." — Michael Wheeler, EVP at NTT DATA
Wheeler underscored that the definition of network resilience has expanded. It is no longer just about physical redundancy—such as dual fiber paths or diverse subsea cable landings. True modern resilience requires dynamic, software-defined resilience that can anticipate congestion points, reroute traffic around potential failure domains in milliseconds, and maintain absolute security integrity without sacrificing performance.
Furthermore, Wheeler addressed the rising tide of sophisticated cyber threats, noting that manual security intervention is mathematically obsolete against modern volumetric attacks. Operators are now forced to integrate automated defense mechanisms deep within the routing fabric. By deploying machine learning algorithms at network edges, operators can identify anomalous traffic signatures and scrub malicious payloads at line rate, ensuring that legitimate AI and enterprise data flows remain uninterrupted.
Future Outlook: The Next Generation of Global Networks
Looking ahead, the roadmap for global telecommunications and network infrastructure is inextricably tied to the continued maturation of artificial intelligence and edge computing. Several critical trends will dictate industry success over the coming decade:
1. Hyper-Automation and Intent-Based Networking
Human operators can no longer keep pace with the sheer volume of telemetry generated by modern IP backbones. The future belongs to closed-loop, intent-based automation systems. These platforms use localized AI models to continuously monitor network health, predict hardware failures before they occur, optimize traffic engineering paths in real time, and execute self-healing protocols without human intervention.
2. Sustainable Infrastructure and Energy Efficiency
The exponential power demands of AI data centers are well documented, but network infrastructure is also facing intense scrutiny regarding its carbon footprint. Future network deployments will prioritize energy-efficient optical transport gear, dynamic port power-scaling during off-peak hours, and tighter architectural integration with renewable energy microgrids powering colocation facilities.
3. The Deepening Synergy Between Security and Connectivity
As network slicing, edge computing, and multi-cloud architectures blur the lines between private enterprise networks and the public internet, security must be baked into every layer of the connectivity stack. Secure Access Service Edge (SASE) frameworks and automated DDoS mitigation will evolve from value-added services into foundational prerequisites for any global IP transit provider.
Conclusion
As articulated by NTT DATA’s Michael Wheeler at Capacity Europe, the telecommunications industry stands at a defining crossroads. The digital economy’s insatiable appetite for AI capability requires a total reinvention of global network infrastructure. By embracing ultra-high-capacity architectures, uncompromising resilience standards, and intelligent automation, forward-thinking operators are building the robust digital highways that will power the innovation of tomorrow.