Your AI pilot works perfectly: The model performs as expected, and the business case makes sense. So, you decide to scale it up throughout your organization — and that’s when things start to unravel. Latency creeps in, data transfers slow to a crawl and your security team sounds the alarm.
This scenario is playing out in many organizations right now as AI investment accelerates. There’s enough budget to convert pilot projects into measurable results, but a critical constraint is emerging. It’s not the AI model, and it’s not the cloud.
It’s your network.
Long treated as invisible infrastructure, the network is now a decisive factor in AI success. Because it directly affects AI performance, security and ability to scale, it will determine whether your AI project moves forward or stalls.
AI is changing what the network is expected to do
AI workloads don’t behave like legacy enterprise applications. They are distributed across cloud and edge environments and around the world, generating unpredictable traffic patterns. They also depend on real-time secure and sovereign data flows. Security, latency and throughput now directly influence business outcomes.
This means the network is playing a more important role than simply transporting data between systems.
In late 2025, NTT DATA conducted wide-ranging, independently sourced AI and technology architecture research among more than 7,000 decision-makers — including CEOs and CAIOs — in 35 countries and 15 industries. We classified organizations as AI leaders if their AI strategies were well-defined or in progress, their AI maturity was at a mature or evolved stage, and they realized significantly higher profit from AI than their peers.
We found that 7 in 10 AI leaders are already making substantial investments in AI, fueling exponential growth in data and workloads and putting immediate pressure on networks to become intelligent, autonomous systems.
Is AI ambition outpacing network readiness?
Leaders are very aware of how critical the network has become. Our data shows that 98% of all organizations recognize that AI is increasing the need for network investment, yet fewer than half are fully satisfied with their IT modernization progress.
As I noted earlier, getting an AI pilot project off the ground is no longer the hard part, and issues with performance and security gaps often trace back to the network. That helps explain why 96% of organizations believe it’s important to have a dedicated network modernization and transformation strategy.
However, only 14% expect to have a fully modernized network in the next 12 months. This implies that most are trying to run next-generation AI workloads on infrastructure designed for a different era.
Their AI ambition is outpacing their network readiness.
3 ways the network is moving to the center
AI thrives on intelligent, autonomous networks, which have become the new foundation for competitive advantage.
1. From connectivity to autonomous intelligence
Static, deterministic networks can’t keep up with the requirements of AI-driven environments. But software-defined architectures can adapt in real time to optimize traffic, prioritize high-value workloads and respond autonomously to changing demands. As a result, the network shifts from passive infrastructure to a strategic control point that directly affects performance, risk and business outcomes.
In our survey, 96% of all organizations said a modern, software-defined network, incorporating network APIs and agentic AI interfaces, is essential to modernizing cloud and applications and achieving network goals.
2. From perimeter security to embedded trust
Our data shows the number-one priority for AI-ready network transformation is to deploy more security infrastructure from the start instead of bolting it on later.
As AI moves data across clouds, edge environments and regions, the old security perimeter starts to dissolve, and risk and scrutiny emerge just as quickly. Nearly all organizations (97%) say they’re worried about AI-related risks, and more than half of AI leaders (59%) cite cross-geography data privacy and sovereignty as a top governance concern.
Trust can no longer live at the edge. It needs to be built into the network itself, constantly verifying every interaction, enforcing policies in real time and protecting data wherever it moves. In this model, the network becomes the first line of defense for managing risk at scale.
3. From visibility gaps to real-time observability
According to our research, 97% of AI leaders say maintaining performance and security requires real-time observability and a unified view of cloud, network and security. The network is becoming a real-time decision-making engine that provides the telemetry and intelligence needed to manage increasingly complex environments across your business.
Additionally, 96% of organizations say the ability to dynamically scale and segment network traffic is critical when supporting evolving AI-driven architectures.
However, maintaining performance and security in dynamic AI systems requires full, continuous visibility of cloud, network and security layers, and many organizations aren’t there yet: 50% of CIOs and CTOs say they haven’t fully integrated these layers, leaving them with operational blind spots.
This is not your everyday network-modernization effort
Incremental change won’t be enough.
Organizations that upgrade components, add capacity and improve network performance incrementally will always fall short. This is because AI introduces volatility — the rapid scaling of workloads, shifting traffic patterns and the need for continuous policy enforcement.
Piecemeal network upgrades end up adding complexity rather than providing a solution. What’s required is a redefinition of the network’s role in your enterprise architecture.
The case for a platform approach
As the network takes on a more strategic role, fragmentation will be a liability. Disconnected cloud, network and security tools create blind spots and increase operational risk.
Platform-based approaches address this by bringing infrastructure, applications, security and governance into a unified, AI-powered operating model. The result is quicker, better-informed decisions, consistent policy enforcement and the ability to scale AI with confidence instead of complexity.
In fact, organizations expect a threefold increase in fully managed and automated cloud platforms over the next 12 to 18 months. The goal is not simplification for its own sake, but the ability to scale AI workloads with control and consistency.
Already, 32% of AI leaders are prioritizing scalable, secure platforms and technology stacks to speed up compliant scaling, compared with 22.1% of AI laggards — the opposite cohort in our research.
Partnerships help sidestep execution challenges
Even with the right strategy, execution remains a barrier. Our research shows 89% of organizations believe they lack the in-house capability to integrate agentic AI into existing systems.
This is also where the conversation shifts from technology choices to operating models. Organizations are recognizing that success depends on integrating cloud, network and security into a cohesive whole. To do that, they often need trusted partners with expertise in all three domains who can assist with performance improvements and alignment with business outcomes.
Most organizations now say it’s extremely important to work with a managed service provider that also owns the core infrastructure — data centers, networks and more — to ensure integration, performance and accountability.
Additionally, organizations identify working with a qualified infrastructure and services partner as the top contributor to achieving AI ROI.
It’s time for a leadership reframe
Senior leaders need to acknowledge that it’s not only models, data and computing that drive AI success. Without the right network foundation, those investments won’t hold up at scale. The network now sits alongside them as a core driver of performance, security and growth, directly influencing how quickly AI initiatives turn into real results.
If you recognize this in your organization early on, you’ll be able to treat your network as a strategic asset to design, govern and optimize in a way that helps transform your business into an AI-driven enterprise.