In early August 2025, Numaligarh Refinery Limited (NRL) initiated a significant development for both the refinery and telecom industries – it signed an MoU with Bharat Sanchar Nigam Limited (BSNL) to deploy a dedicated private 5G network within its Assam refinery. The brief was simple yet demanding: deliver secure, high-reliability, low-latency connectivity for industrial workloads. By integrating a captive non-public network (CNPN), the official term for private 5G networks in India, the refinery aims to enhance operational efficiency and cybersecurity while enabling AR/VR-based training, digital twins and real-time IoT applications.
NRL’s plans are being supported by the Department of Telecommunications (DoT). On June 30, 2025, DoT launched a module on the Saral Sanchar portal to map enterprise demand and identify frequency bands for direct spectrum assignment, a reset prompted by a maturing device ecosystem in the International Mobile Telecommunications (IMT) bands and rising interest from system integrators.
Mapping India’s CNPN landscape
A CNPN is a private 5G network built for exclusive use by an enterprise within its premises such as plants, ports, campuses and hospitals. Traffic on these networks remains isolated from the public network. Meanwhile, policies, quality of service (QoS) and security can be tailored to operational needs. The concept of CNPNs was introduced in India in June 2022 with DoT’s guidelines and related licence changes. Uptake since then has been modest, but momentum is building as the authorisation framework becomes clearer and device support in IMT bands improves.
The four deployment routes that shape enterprise choices are:
Public slice (TSP-managed): A licensed TSP creates a logical slice on its public 5G network and aligns it with the enterprise’s policies and QoS. Under this model, patching, upgrades and change windows remain the operator’s responsibility.
TSP on-premises (isolated): The operator deploys a physically segregated radio and core network on the enterprise’s premises using its IMT spectrum. This model ensures predictable performance while compliance obligations lie primarily with the operator across the lifecycle.
Enterprise-run on leased spectrum: The enterprise, typically in partnership with a system integrator, installs and operates the on-premises stack while leasing IMT spectrum from a telecom operator. This model provides tighter control over policies and coverage while ensuring data remains local, but it requires higher capital outlay and stronger in-house skills.
Enterprise-run on directly assigned spectrum: The enterprise seeks spectrum directly from DoT and runs a fully self-managed network. This model offers maximum sovereignty over the radio layer and data, but also imposes the highest obligations in terms of audits, security and round-the-clock operations.
Limited adoption
While the CNPN framework, launched in 2022, introduced multiple routes for implementation – leasing, slicing and direct assignment – its adoption has remained modest for practical reasons. The device ecosystem for the bands initially “identified” for direct CNPNs was limited, which slowed procurement and certification. By mid-2025, DoT itself noted that support was “mostly available” in IMT bands, prompting the launch of its Saral Sanchar demand exercise to map real interest and shortlist workable spectrum options. In addition, authorisation ambiguity deterred buyers. Until 2025, enterprises and system integrators (SIs) had limited clarity on who could operate private networks, under which licence and with what obligations. Integrators have also stated that factories are unlikely to invest without a strong business case tied to outcomes and lifecycle management, hence they are pivoting to bundled offers that pair private 5G with edge computing artificial intelligence (AI) under service level agreements (SLAs).
Meanwhile, the Cellular Operators Association of India has stated that direct spectrum allocation for CNPNs to enterprises is “not tenable in India”, citing fiscal and security concerns, and has promoted TSP-delivered enterprise solutions via leasing or slicing. For buyers, the near-term translation is simple – plan around operator-anchored deployments while viewing direct assignment as a parallel policy track that may take longer to mature.
In practice, the lack of direct enterprise spectrum assignment does not stall adoption; it changes design choices. Refineries or ports seeking on-premises isolation today can opt for TSP-run isolated CNPNs, which meet compliance requirements, deliver predictable performance and can be structured with outcome-linked SLAs. Once direct licensing stabilises, enterprises that prioritise sovereignty can reassess their strategy.
Why 2025 seems different?
India now has a real industrial reference: the BSNL and Numaligarh Refinery MoU for a refinery-grade 5G CNPN. Meanwhile, Tech Mahindra is pitching private 5G for Industry 4.0 in India alongside operator partners, highlighting deployments in automotive manufacturing that pair private 5G with real-time analytics and connected robotics. The company has collaborated with Bharti Airtel to roll out its Chakan facility, India’s first 5G-enabled automobile manufacturing unit.
Recently, BSNL’s Kerala circle also announced India’s first standalone, on-premises 5G private network (CNPN) in the mining sector, reportedly deployed at the Amlohri Opencast Coal Mines, emerging as an industrial automation use case.
Overall, activity in 2025 has been centred around telcos and enterprise collaborations that package private 5G as managed solutions. The logic is straightforward – factories buy outcomes (predictable latency, reliable uplink, secure segmentation), not stacks. This has led to designs that combine radio access network (RAN), on-premises core, multi-access edge computing (MEC) and analytics under a single SLA. Tech Mahindra’s public narrative aligns with this approach, tying private 5G with real-time analytics and robotics in Indian manufacturing.
On the demand side, boards remain disciplined. They do not commit without a clear return on investment strategy, preferring lifecycle-managed offers that simplify skills, spares and change windows. Infosys’ leadership has also articulated this recently, echoing the demands of procurement teams in requests for proposals.
Where will CNPNs gain traction first? The most promising starting points are high-stakes environments where outages are costly and safety critical. These include refineries, steel and cement plants, ports, automotive body shops and logistics yards. These settings have consistent workload profiles – condition monitoring with on-premises analytics, machine vision for quality/safety, AR/VR-assisted maintenance, automated guided vehicles (AGVs) and robotics. NRL’s refinery represents a high-value use case for CNPNs.
The way forward
Looking ahead, four factors will shape India’s CNPN trajectory. First, post-survey band clarity and fees (which IMT bands DoT shortlists and under what prices and conditions) will define device portfolios and total cost of ownership. Final band choices determine which radios, devices and modules vendors bring to India and how quickly enterprises can scale. Second, the authorisation fine print – how the government implements the Telecom Regulatory Authority of India’s (TRAI) recommendations – will differentiate operator and third-party roles for lifecycle management, security obligations and reporting. Third, deployments in oil and gas, ports and steel will determine whether NRL is a template or a one-off. The trade press already points to active pipelines including operator-system integrator bundles in manufacturing. Fourth, SLAs are likely to become outcome-centric, formalising availability, maintenance windows and per-use-case latency/throughput targets in line with global factory benchmarks.
In the near term, scale will follow strong economics. As per industry estimates, India’s private 5G network market stood at $30.6 million in 2023 and is projected to reach $674.6 million by 2030 at a CAGR of 55.5 per cent.
Meanwhile, policy will still decide the long-term game. If direct enterprise licensing emerges with workable bands and fees, sovereignty-seeking sites will take that route. If not, slicing and operator-built isolated CNPNs on customer premises remain a compliant path to scale up at speed. Either way, CNPNs now appear less like trials and more like a repeatable playbook, starting with process-heavy environments and extending into adjacent use cases as device maturity and in-house expertise deepen.
