TRAI's August 5, 2026 consultation paper does something India's net neutrality framework has never explicitly done before: it writes differentiated 5G service tiers into a binding quality-of-service regulation, rather than treating them as a licensing gray zone. The draft amends the Standards of Quality of Service of Access and Broadband Service Regulations, 2024, and stakeholders have until August 26 to comment, with counter-comments due September 7.
What the draft actually permits
Network slicing lets an operator partition a single physical 5G network into virtual segments — one tuned for low latency, another for guaranteed throughput, another for best-effort browsing — and sell access to those segments separately. TRAI's proposal does not ban this. It conditions it. Per the paper, operators must keep cells with daily Physical Resource Block (PRB) utilisation above 80% to under 1% of their total network; if a cell's PRB load sits above 80% for five days in a month, the operator must add capacity, and if it can't, the slice comes off that cell (Communications Today; MediaNama). New slices must be filed with TRAI, with technical parameters, at least 21 days before launch. Each slice is treated as its own tariff product with its own QoS obligations — meaning a premium slice's speed promises are as enforceable as a standard plan's.
Why this revives the debate
This isn't happening in a vacuum. In July 2026, the Department of Telecommunications formally asked TRAI to revisit India's net neutrality framework in light of "emerging technologies and network management practices, including network slicing and differentiated service offerings" (Whalesbook). That request sits uneasily next to India's own history: TRAI's 2016 order banned differential data pricing outright (the rule that killed Facebook's Free Basics), and the 2018 net neutrality license amendments enshrined non-discriminatory traffic handling as a licensing condition. Both were built for an internet where "fast lane" meant charging content providers for preferential delivery. Network slicing is a different animal — it differentiates by service class (say, a remote-surgery link needing guaranteed sub-10ms latency versus a video stream that can tolerate buffering), not by which website or app a user visits. But critics are right that the line is not self-enforcing: if a carrier can reserve capacity for a premium slice, that capacity is definitionally unavailable to the general-purpose internet during congestion, and ordinary users bear that cost first (Business Standard).
The steelman for TRAI's approach
The strongest case for treating slicing as a QoS question rather than a neutrality violation is that undifferentiated best-effort internet was never neutral to begin with — it already privileges latency-tolerant traffic over latency-sensitive traffic by default. Telemedicine, industrial automation, and connected-vehicle applications genuinely need guaranteed performance characteristics that a single best-effort pipe cannot deliver at scale. Refusing to let operators build for that need doesn't preserve neutrality; it just forces those use cases onto private networks or foreign infrastructure, and starves Indian operators of a viable 5G monetisation path at a moment when capital-intensive network buildout badly needs new revenue. A capacity-triggered cap — congestion, not content, decides when a slice contracts — is a coherent way to reconcile differentiation with a baseline service guarantee, and it's more precise than a blanket prohibition that would have applied equally to a hospital's dedicated slice and a gaming company's paid speed boost.
Where the draft is still thin
The problem is enforcement architecture, not the underlying principle. The 80%/1% threshold is self-reported by operators through the same coverage-map and outage-disclosure channels TRAI is simultaneously trying to tighten elsewhere in this same consultation paper — a tacit admission that operator self-reporting has been unreliable. There's no independent audit mechanism specified for PRB data, no public dashboard, and no penalty schedule attached to the slicing provisions themselves (compensation clauses in the draft apply to outages, not to slice-driven degradation of the general network). A 21-day filing window with TRAI is a paperwork checkpoint, not a public-interest review; nothing in the draft requires operators to disclose what capacity a new slice will draw from the shared pool before it launches. If TRAI wants proportionate regulation instead of prohibition, the fix is not more restriction on slicing itself — it's making the 1% figure independently verifiable, the way SEBI or RBI compliance reporting increasingly requires audited submissions rather than self-certification.
The bottom line
TRAI has picked a defensible middle path: differentiation permitted, degradation capped, transparency assumed. The August 26 comment window is where that last assumption gets tested — civil society groups and consumer advocates should push for audit and disclosure requirements, not for killing network slicing outright. A congestion-based ceiling that nobody can verify is not meaningfully different from no ceiling at all.