Dr Badri Gomatam, Group Chief Technology Officer, STL
Finance Minister Nirmala Sitharaman’s Union Budget 2026 announcement of a tax holiday until 2047 for cloud providers using Indian data centres signals a significant policy push to attract global artificial intelligence (AI) infrastructure investment. It is a bold move that positions India to compete as a hub for cloud computing and AI. But the real story is not just about tax incentives. It is about whether the optical fibre networks connecting these facilities can handle future demand.
India’s data centre capacity is projected to reach 8 GW by 2030, requiring investments of about $30 billion. The AI market is expected to grow at 25-35 per cent annually through 2027. Behind these numbers lies a fundamental question: Can the network infrastructure scale at the same pace?
Why AI rewrites the network playbook
Think of AI infrastructure as a living organism. The data centre is the brain where graphics processing units (GPUs) perform the thinking. The fibre network is the circulatory system, moving data at the speed of light between servers, clusters and users. A powerful brain without blood flow is inert. Similarly, advanced AI models are ineffective without dense, high-capacity fibre networks to sustain them.
Traditional data centres and AI-enabled facilities are fundamentally different. An AI data centre, built around thousands of interconnected GPUs rather than conventional CPUs, requires 36 times more fibre infrastructure than its traditional equivalent.
This is not about adding capacity to existing networks. AI workloads generate massive internal traffic within data centres as tightly coordinated GPU clusters exchange enormous volumes of data continuously. The fibre infrastructure needs to support these parallel flows while maintaining ultra-low latency – both at the same time.
The energy equation nobody talks about
Here’s what often gets overlooked in discussions about AI infrastructure: by 2030, data centre power consumption in India will hit 8 GW. Globally, data centre capital expenditure surpassed $450 billion in 2024 and is projected to exceed $1.2 trillion by 2029. As these facilities scale, energy efficiency becomes as critical as computational performance.
Network design directly impacts this equation. Fibre infrastructure that minimises signal loss reduces amplification needs and the power consumption that comes with it. At this scale, these are not incremental improvements. They determine whether AI infrastructure is economically sustainable over the long term.
The innovations happening in optical technology are about building networks that can operate efficiently for decades, not just meet today’s requirements. That is the difference between a sustainable AI economy and one that hits an energy wall.
Getting the timing right
Budget 2026 provides valuable policy support, and the industry is responding with capital commitments. However, the challenge is execution. Data centre construction cycles are relatively short. Major facilities can be operational within 12-18 months. Fibre network deployment, especially when building diverse routes with enterprise-grade redundancy, takes longer. The networks need to be ready in parallel, not playing catch-up.
Global data creation is expected to more than triple in the coming decade. Meeting this requires massive investment in fibre infrastructure, not only inside data centres but also between them. Modern data centres already rely on thousands of fibre links to connect clusters and campuses. The scale required for India’s ambitions is substantial.
What success looks like
India has demonstrated strong capabilities in executing large-scale infrastructure projects across sectors. The technical expertise exists within the optical networking ecosystem. Budget 2026’s policy framework, combined with industry investment, creates the conditions for establishing a competitive position in global AI infrastructure.
The foundation lies in recognising optical networks not as supporting infrastructure, but as core enablers. The future of AI will not be written in code alone. It will be carried by fibre. With the right focus on fibre deployment alongside data centre construction, India can translate policy incentives into meaningful outcomes. The tax holiday creates the opportunity. The fibre networks will determine whether it translates into reality.
