At MWC 2026, Gurpreet Singh Kohli (GSK), EVP and Global Business Head – Telco and Enterprise Networks at HCLTech, explored why networks are no longer passive infrastructure and instead, the foundation of the autonomous era.

AI cannot succeed without robust connectivity

While much of the AI conversation focuses on compute, models and data, Kohli explained that “if AI is to become successful, the underlying connectivity layer has to be robust.”

Today’s enterprises operate in a federated, distributed data environment. Applications reside across hyperscalers, colocation facilities and the edge. Devices connect across multiple protocols. Decision-making is increasingly decentralized.

In that context, maintaining seamless, secure connectivity across all data sources becomes critical. Without it, AI cannot deliver the right results.

Kohli distinguished between “AI in network operations” and “networks for AI,” noting that in a distributed world, it is networks for AI that are becoming “omnipotent.”

Real-time intelligence depends on network resilience

The implications become clearest in high-stakes physical environments.

Kohli described a scenario in which a surgeon, assisted by Agentic AI, performs a delicate procedure. If a potentially catastrophic incision is detected, an edge application responds within milliseconds, flagging the risk instantly. Additional layers of intelligence may escalate to colocation facilities or hyperscaler environments within seconds to validate next steps.

That responsiveness depends not just on compute power, but on a connectivity layer that is “super reliable, super secure and super available.”

Physical AI, whether in healthcare, manufacturing or robotics, relies on millisecond-level decision loops across distributed architectures. The network becomes the invisible enabler of lifesaving and defect-reducing outcomes.

“The underlying network is super important,” he continued.

Private 5G, micro-segmentation and contentionless bandwidth

With autonomy comes heightened demand for control and security.

Private 5G is emerging as a cornerstone of this transformation. Unlike traditional Wi-Fi architectures, which operate in contention-based environments, private 5G enables what “contentionless bandwidth,” delivering predictable performance for mission-critical applications.

Enterprises are increasingly seeking full control over SIM management, spectrum leasing and packet core infrastructure. In some cases, no application data is permitted to leave the campus under any circumstances.

Security models are evolving accordingly. Micro-segmentation reduces attack surfaces, particularly as sophisticated threats increasingly originate from inside networks rather than at perimeter firewalls.

Yet transformation is gradual. “This is a journey,” noted Kohli. Adoption may begin with a small footprint before scaling over several years. The challenge lies in managing heterogeneity and integrating legacy technologies such as LoRa, Zigbee, wired and earlier Wi-Fi generations alongside emerging private 5G architectures.

Heterogeneity also extends to vendor ecosystems. No single OEM delivers a complete stack across SD-WAN, SASE, Wi-Fi and private 5G. Integration requires orchestration across multiple stakeholders, and cross-skilled talent capable of navigating complex environments.

Solving the right problems first

A broader industry pattern is emerging: solutions are often deployed before problems are fully understood.

“Organizations [can fall into the trap of providing a] solution without understanding what the problem is.”

To address this, he described a recent engagement with a large energy enterprise facing SD-WAN challenges. Rather than replacing vendors blindly, the team conducted a comprehensive assessment, developed extensive use cases and evaluated multiple stacks before integrating a tailored solution.

Architecture must precede deployment. Use cases must precede procurement.

Integration as an art

As OEM stacks diversify, integration complexity increases. Enterprises may now rely on five different vendors across network layers.

Coordinating those environments, particularly during critical incidents, requires high-touch technical services and deep ecosystem relationships. Interim fixes may need to be implemented while working with multiple OEMs to resolve root causes embedded in code or firmware.

Kohli described managing such scenarios across multi-vendor environments as “a piece of art.” It is a discipline built over decades.

Networks are no longer “dumb pipes”

Looking ahead, Kohli sees a structural shift in how networks are perceived.

“It is going to make us shine like never before, because networks are becoming to become omnipotent. There are no more dumb pipes.”

In the autonomous era, networks carry decision loops, not just data packets. They underpin Physical AI, secure distributed compute and enable real-time industrial automation.

For telecom providers, system integrators and OEMs, this represents a defining moment. As AI becomes embedded in every enterprise function, connectivity moves from background utility to strategic differentiator.

The autonomous era will not be powered by algorithms alone. It will be powered by the resilience, intelligence and security of the networks beneath them.