What’s next in networking and security?

Savill: Regulatory and data sovereignty pressures, technical debt and fragmented operations are the three forces widening the gaps. As the 2025 Kyndryl Readiness Report illustrates, enterprises still have major pain points to overcome. 

Data residency and privacy regulations have become critical components of IT strategy. But most architectures weren’t built to accommodate the infusion of AI to handle these workloads. Compounding the challenge, we’re rapidly approaching the time when bad actors will have quantum computing tools that can break into files that are currently encrypted. We’re already seeing instances of “steal now/decrypt later” data theft in anticipation of these new technologies. This creates significant long-term risk for organizations that delay addressing these gaps. 

We also see lots of enterprise environments where components have been “bolted on,” built from years of incremental upgrades and acquisitions. While organizations have invested in automation to reduce manual configuration, monitoring and troubleshooting, basic automation is no longer enough as AI workloads scale. These fragmented networks lack unified observability, creating blind spots that slow response times and make it difficult to understand root causes, validate AI performance or trace security incidents. As workloads become more distributed, these limitations are amplified — making modernization essential for AI-native, real-time operations.  

Savill: Compliance by design is the only way to stay current with the constantly evolving regulations that impact every global industry — including supply chains and materials sourcing. Together with ecosystem partners such as Nokia, Kyndryl helps customers modernize their approaches to compliance. We develop network architectures that incorporate policy-aware routing to move data only where regulations allow; in-network geo-boundary enforcement to help prevent off-region data movement; and continuous, real-time auditability delivered through the Kyndryl Bridge AI-native platform.

Using these approaches, organizations can innovate with AI while maintaining control over data movement to help meet today’s and tomorrow’s regulatory requirements. And by shortening redesign cycles and reducing compliance-related operational risk, the compliance-by-design approach can help organizations realize positive ROI on their AI spend.

Savill: Most enterprises are still in the early stages of this journey. Today, AI is being used primarily as an operational assistant rather than a fully autonomous controller. We’re seeing early adoption in areas such as automated network design, rapid provisioning and the reduction of Tier-1 troubleshooting workloads. However, true cross-stack, zero-touch operations remain years away. Network changes affect mainframes, operational technology, cloud platforms and applications, and must follow strict governance and change-control processes.

That complexity makes achieving full autonomy a gradual, evolutionary process. At the same time, we’re seeing Kyndryl’s technology partners including Cisco, Nokia, NVIDIA, and Hewlett-Packard Enterprise, embedding AI deeply across routing, security and data center platforms. And we expect major advances this year. A key next step will be interoperability — enabling  AI agents from different vendors and platforms to communicate and operate cohesively across complex environments.