A rollout that got ahead of the rulebook
At India Mobile Congress 2026, Reliance Jio's Sanjay Thakur said the company had defined a "mission critical slice" for the Maharashtra Police when 35 lakh people, about 3.5 million, gathered at Pandharpur in a single day. He said a similar slice was used at Tirupati, where about five lakh people gathered. Jio's Aayush Bhatnagar added that six commercial slices already run in its production network, MediaNama reported on 9 October. TRAI has not finalised any rules for slicing.
The gap is not an oversight Jio exploited. It is the normal state of a young technology. The question for regulators is whether the rules they write next will protect users or freeze a useful capability before its uses are understood.
The strongest case for TRAI's approach
The regulator has a serious concern. A network slice is, by design, a way of giving some traffic guaranteed or prioritised treatment on shared radio spectrum. The radio resources in a cell are finite, so a premium slice can be paid for out of the capacity left to everyone else. India has strong reasons to be wary. TRAI's Prohibition of Discriminatory Tariffs for Data Services Regulations, 2016, issued in February 2016, barred content-based pricing, and the country has since treated net neutrality as a settled public commitment. A regulator that asks operators to show ordinary subscribers are not degraded by slices is doing its job.
TRAI's August 2026 consultation paper, summarised by Indian Infrastructure, contains several sound ideas. Slices offering the same service, such as internet access, would follow the same prioritisation standards. Each enhanced mobile broadband slice would be a separate tariff offering with measured download and upload speeds disclosed. Operators would disclose the capacity provisioned in each radio cell. Transparency of this kind is cheap for operators, and it lets researchers and users check claims.
Where the draft goes wrong: utilisation is not harm
The same summary reports a benchmark of 80% radio resource utilisation during peak hours, with no more than 1% of cells in an area exceeding it. Jio opposes this and argues that high utilisation does not necessarily mean congestion or slower speeds, according to MediaNama. On that point Jio is right.
A cell running at 90% utilisation while every user gets an acceptable experience is an efficient network. A cell at 60% can still serve a video call badly because of interference. Utilisation is an input to quality, not a measure of it. A hard threshold would push operators toward building capacity ahead of demand or throttling slices to stay under the line. That raises costs in a market where Indian consumers are used to some of the world's cheapest data. It also treats a sign of efficient spectrum use as evidence of a fault.
The regulator already has the better instrument. The draft's own proposals on disclosed per-slice speeds and a non-compliance finding when speeds fall below typical advertised levels measure what users actually experience. Outcome metrics such as throughput, latency and drop rates should anchor enforcement, with utilisation kept as diagnostic data operators report and TRAI monitors.
The net neutrality question is real but narrower than critics claim
The Broadband India Forum warns that application-specific slicing could breach net neutrality by giving different apps different speeds. MediaNama has argued that neutrality rules should apply to slices. Jio and the GSMA, by contrast, urge TRAI to allow service-based slicing.
The distinction between those two phrases is where policy should focus. A slice defined by a service class, such as a mission-critical public-safety link or a low-latency industrial connection, is a different product bought for a different purpose. A slice that makes one video platform fast and a competitor slow, because the first paid, is the discrimination the 2016 regulations targeted. Airtel's submission to TRAI argues that network slicing is a distinct 5G capability that should not be treated like ordinary technology-agnostic quality-of-service, as its filing on TRAI's site reflects. That argument is correct about the technology but it does not answer the neutrality objection.
The Pandharpur example shows why a blanket ban would be a mistake. A police slice at an event of 3.5 million people is not an attack on the open internet. It is a managed service for an emergency function, on a network where ordinary users keep ordinary access. Rules that cannot tell this from paid prioritisation of a commercial app will either block the former or permit the latter.
A proportionate framework
Three principles would let India protect users without suppressing the technology.
- Test discrimination, not utilisation. Define the violation as degrading or favouring lawful content or applications on commercial grounds in the general internet-access slice. Enforce through measured user outcomes.
- Carve out and disclose specialised services. Public-safety, enterprise and machine-to-machine slices should be permitted if they do not impair general internet access, with operators registering them and publishing the capacity they reserve.
- Publish per-slice reporting. TRAI's disclosure proposals are the strongest part of the draft. Adopt them, and let independent measurement and stakeholder submissions, such as those from The Dialogue, test whether operators comply.
Why timing matters
A Reliance Industries executive said on an earnings call that the network is ready but regulations and the market are not. Capital waits on clarity. If TRAI finalises a utilisation cap, investment in slicing may stall or move to private-network deployments outside the consumer rules. If it finalises an outcome-based neutrality test with disclosure, operators know the line and users can see whether it is crossed.
India can lead on this. It already has a rulebook on discriminatory tariffs and now has live slicing deployments to learn from. The sensible course is to regulate what slices do to users and competitors, and leave the efficiency of the radio network to the operators who pay for it.