The executive takeaway
- AI initiatives operating across distributed environments may place greater demands on data movement, workload placement and network capacity.
- Many traditional network architectures were designed around relatively predictable traffic patterns. Distributed AI workloads can introduce additional complexity that may require wider business oversight.
- One practical response is a programmable network: software-defined connectivity that gives enterprises more control over where data moves, how quickly it gets there and how easily capacity can adapt.
CIOs leading enterprises across Asia Pacific (APAC) are balancing several demands at once: cloud platforms that do not naturally work the same way, AI workloads that may be sensitive to network latency, data centres that still matter and regional policies that must be applied consistently across markets. Without a coordinated approach, operational complexity can make the business ambition harder to deliver.
Many organisations may want faster AI adoption, better cloud flexibility or expansion across APAC. But the CIO has to answer a more practical question: can the network support that ambition without adding more manual work, fragmented visibility and environment-by-environment complexity?
Today, the AI debate often begins with GPUs, data centres and energy. Network performance and data movement can also affect how efficiently distributed AI workloads operate. Whether data can reach the right place at the right time without workloads waiting for the network to catch up.
For the CIO, the issue is no longer just connectivity. It is control. Can the enterprise reduce environment-by-environment network management while retaining appropriate control across its supported clouds and data centres?
That is where a programmable network becomes relevant. It gives enterprises a way to move from fragmented connectivity towards a more controlled, software-driven network model that can adapt as AI and cloud demand changes.
| “AI is reshaping how businesses operate, creating unprecedented demand for intelligent, programmable networks that can move data securely and seamlessly across cloud, edge, and data center environments.”1 — Kate Johnson, CEO, Lumen Technologies |
Why static networks may struggle to support AI workloads
The problem starts with an old assumption. Many established enterprise networks were designed around comparatively predictable traffic patterns: people connecting to applications, sites connecting to sites and traffic moving in patterns teams could plan for.
AI changes that rhythm. Distributed AI architectures may move data between clouds and place some inference workloads closer to users. Each additional environment can introduce further connectivity, security and management requirements. A network issue can quickly become an operating model issue.
| “Everybody right now keeps talking about the training of the models. I think what a lot of people really haven’t absorbed yet is that the speed of the transaction really matters.”² — Jim Fowler, EVP Chief Technology and Product Officer, Lumen Technologies |
How programmable networks work
A programmable network uses a software control layer to reduce manual, environment-specific configuration and support more centralised management of connectivity, policy and routing.
A simple way to understand the shift is to imagine the difference between pilots individually negotiating flight paths over the radio and a single control tower setting rules that every aircraft follows automatically. One model depends on constant, manual coordination. The other sets the conditions once, so the wider system can adapt more consistently.
Programmable networks bring a similar shift to enterprise cloud connectivity: from managing each environment separately to using a software control layer designed to coordinate connectivity, policy and routing across supported environments.
Lumen acquired Alkira to add cloud-native orchestration capabilities that are designed to help customers manage connectivity and network services across compatible hybrid and multi-cloud environments. In practical terms, this strengthens the idea of the network as a control layer for distributed cloud and AI environments, not only as the infrastructure that connects them.
This evolution expands the role networks play in supporting digital operations. Beyond connecting people and devices, networks are increasingly expected to support interactions among applications, systems and automated technologies. In distributed AI environments, that includes communication between AI agents, applications and systems, helping data and operational information move where they are needed.
The implication for CIOs is clear: as more digital activity becomes automated and distributed, the network needs to become easier to control, not harder to manage.
Traditional network vs programmable network
| Traditional network | Programmable network | |
|---|---|---|
| Management model | Often managed through provider- or environment-specific processes | Software-defined management across compatible services and environments |
| Visibility | Fragmented, one view per environment | More unified visibility across supported network environments |
| Cloud connectivity | Custom integration for every cloud or partner | Any-to-any connectivity through supported platform capabilities |
| Security and policy | Configured separately in each environment | Consistent policy applied across supported environments |
Disclaimer: Programmable network capabilities depend on service design, deployment configuration and platform availability. “Supported environments” refers to clouds, carriers and platforms compatible with the relevant programmable network service at the time of deployment. Actual outcomes may vary based on your existing network architecture and the environments you operate in.
What this changes for APAC enterprises
For enterprises operating across APAC, the value is not another network feature: it is the complexity that starts to disappear from an already stretched operating model.
APAC is not one market, and that is the point. Enterprises expanding across Singapore, Japan, Australia and Hong Kong may need to account for different regulatory, operational and data-location requirements. Without a more coordinated approach, each new market, cloud or partner can become another integration project.
In a programmable model, a software-based management layer can help reduce the operational impact of managing connectivity across multiple supported environments. This does not remove the need for reliable underlying network infrastructure; rather, it helps enterprises reshape connectivity, policy and routing faster as the business expands across APAC, while local carrier, regulatory and deployment requirements continue to apply.
Business outcome
More connectivity is not just a network benefit – it is a strategic one. When enterprises can move data, they can choose the cloud or AI provider that suits the workload, rather than staying within the limits of static connections. That flexibility supports a stronger partner ecosystem.
Control shapes how fast, how securely and how intelligently that happens. Greater network control can help enterprises respond more effectively as AI and cloud requirements change.
For example, Lumen and Alkira capabilities can help organisations to:
- Manage east-west traffic between supported clouds, data centres and AI infrastructure.
- Enforce consistent security policies, gain better visibility and monitoring.
- Ease operational complexity and manual cloud networking tasks through a centralised control plane.
- Preserve your existing network environments – Alkira is carrier and cloud-agnostic, so you keep the benefit of what you already have.
Related reading: Managing complex networks with Lumen + Alkira
The future of programmable network
What a programmable network can support today is only part of the picture. As automated systems become more capable, network management may increasingly use policy-driven orchestration and automation.
The Alkira acquisition moves Lumen towards that future – building a platform designed to enable and orchestrate connectivity for the AI era, not only for the workloads enterprises run today, but for whatever form AI takes as it continues to evolve.
A practical next step
If your organisation manages policy separately across multiple clouds, it may be useful to assess where greater consistency is possible.
Ask yourself:
- Does a new cloud or AI workload mean another integration project?
- Do you manage policy separately in every environment rather than from one place?
- Is visibility fragmented across your clouds and data centres rather than unified?
If any of this sounds familiar, understanding where that complexity already sits is the first step.
Talk to a Lumen network specialist about reviewing your current connectivity, policy and visibility requirements.
Sources
¹Lumen Technologies, Lumen completes Alkira acquisition, accelerating its unified digital platform for AI-era networking, press release, July 2026.
2“What happens when AI hits the network wall,” Lumen Technologies, YouTube, 2026.
Disclaimer: Services not available everywhere. Business customers only. Lumen may change, cancel or substitute products and services, or vary them by service area at its sole discretion without notice. This blog contains forward-looking statements. These forward-looking statements are not promises nor guarantees of future results, are based on our current expectations only, and are subject to various risks and uncertainties. This content is provided for general informational purposes only and does not constitute legal, regulatory or technical advice. ©2026 Lumen Technologies. All Rights Reserved.
