Consistent execution is key to India’s AI build-out

India’s data centre and AI infrastructure build-out has progressed quickly. Capacity crossed roughly 1,700 MW at the end of 2025 and is set to grow by about 30% in 2026, with close to 500 MW of new supply coming online. Investment commitments tell the same story, reaching approximately $126 billion by the end of 2025 and are projected to exceed $180 billion in 2026.

For the operators who build and run the networks underpinning this build-out, the question is no longer whether to build or even how to secure network capacity. It is how quickly that network capacity can be turned up, and how it performs once live, across a country that is expanding well beyond its established hubs.

While four cities—Mumbai, Chennai, Delhi NCR, and Bengaluru—still account for close to 90% of India’s established data centre capacity, new hubs such as Ahmedabad, Visakhapatnam, Patna, and Bhopal are emerging. Each new site and route between sites must be planned, built, and brought into service quickly, often in places where skilled talent is harder to find.

Additionally, AI workloads rarely sit in one place. Workloads are spread across clusters and data centres, tuned both for training models and, increasingly, for running AI services close to where people use them. More sites and more links mean more networks to deploy, with little room for rework when timelines are tight.

In India, demand for engineers and technicians who build and operate networks and data centres is outpacing supply. That is not a reason to slow down. It is a reason to change how the work gets done. As IDC observes, the deployment of new technologies has turned the in-house skills gap into a strategic priority, one that operators increasingly address by partnering with network infrastructure vendors that bring expertise and operational experience to deploy and run new networks and the technologies essential to their success.

Automated, repeatable deployments are essential for scale

For years, deploying a network meant hands-on, site-by-site work: survey, plan, configure, test, commission. The job now is shifting towards making sure new capacity reaches users quickly, works the first time, and delivers high performance consistently, whether it goes live in Mumbai or in a city which is lighting up its first facility.

Software plays a critical role. Networking vendors like Ciena have embedded more advanced software instrumentation into network equipment and intelligent automation into network operations.

Such networking vendors work closely with operators to create powerful software-driven deployment models. Pre-validated designs, automated configuration workflows, and factory-based staging allow networks to be built and tested before they even reach the field. Instead of configuring each site from scratch, operators can deploy infrastructure using standardised templates and repeatable processes, reducing on-site effort and ensuring every rollout meets the same performance baseline from day one. Operators can also adopt staging-in-factory models in which equipment is assembled, configured, and tested before deployment, helping accelerate turn-up and improve consistency across sites.

This shifts deployment from a site-by-site activity to a scalable production model. Automation reduces dependency on scarce field expertise, minimises configuration errors, and compresses the time from equipment delivery to service activation. In high-growth environments, that consistency matters as much as speed.

AI brings yet another higher level of operational efficiency. A recent Omdia study of operators found that 12% already use AI routinely in planning and operations, another 24% expect to within a year, and a further 45% within one to three years. The same research ranked faster fault-finding, performance monitoring, and prediction as the top areas where operators expect AI to help. In plain terms, that means catching a problem before a technician is sent out, not after.

Deployment in a country as varied as India has long leaned on a handful of experts and local workarounds, with quality that could shift from one rollout to the next. A more repeatable approach changes that: the same checks, the same standard of work, the same definition of a job done well, applied from a large metro build to a smaller one. That is how an operator adds capacity while ensuring an adequate workforce at the same rate, which counts for a lot when skilled people are scarce. The operators who succeed will treat deployment as a scaled, automated process rather than a series of individual projects.

The job is not complete when the network goes live

The same approach extends into day-to-day operations. With greater use of network automation, remote monitoring, and AI-driven assurance, many operators are moving toward models where deployment and operations are linked through shared data and standardised workflows, enabling faster issue resolution and more predictable service outcomes.

In practice, this is the logic of a managed operations model: a single point of contact running a network operations centre across a multi-vendor footprint, freeing an operator’s engineers for design and planning rather than watching dashboards. IDC puts the operating expense savings from managed services at 15-20%, but the larger prize is faster deployment and repair.

Predictive tools have also been helpful: by Omdia’s count, only about one in five operators run them routinely today, but more than nine in 10 expect to within three years. For an Indian operator running a network across many cities, often in price-sensitive markets, this is the difference between spending scarce engineering time watching dashboards and spending it on design, planning, and assurance, the work that actually adds value.

From building to outcomes

With India’s AI infrastructure rapidly expanding, this ability to deploy, monitor, and optimise networks consistently will become just as important as adding raw capacity. The next phase of growth will depend not only on what operators build, but on how reliably they can turn new infrastructure into measurable service outcomes.

Put AI tools in the hands of the teams that build and run networks first, because that is where speed and consistency are won or lost. And measure success by the customer experience: how quickly a network goes live, how often it works the first time, and how soon operators can realise monetisation.

Finally, choose execution partners based on domain expertise, operational experience, and a track record of outcomes, not solely on cost.

India has scaled 5G connectivity to hundreds of millions of people faster than almost anyone expected, because ambition, partnerships between operators and vendors, and execution lined up. Applying the same execution discipline to how AI infrastructure is built and operated will determine whether operators simply keep pace with demand or turn it into a lasting advantage. Deployment and operations used to be the last things anyone talked about. They are becoming the competitive edge.

(The author is Vice President, Customer Operations (APAC), Ciena. Views are personal.)