The 6G calendar is locked.
The calendar isn't the hard part.
By: Steve Douglas
For two years, 6G timelines were the kind of thing you could nod along to without committing to anything. Everyone said "around 2030," the roadmaps had soft edges, and nobody had to put their name next to a date. That ended in June, when 3GPP approved the Release 21 schedule at its TSG#112 plenary in Singapore.
Release 21 is the first official 6G release for commercial implementation, so this is the point the dates stopped being conference-panel material and became something the working groups have to resource and hit. It's worth laying out what got agreed, and then the part I think matters more: whether the ecosystem can actually build and validate a new generation of radio and core inside the window it just signed up for.
The dates
The service requirements (Stage 1) are frozen by March 2027. Stage 2, covering the System and Core Architecture, has an 80% checkpoint in March 2028, before the overall freeze in June 2028. A “checkpoint” signifies most of what you need is completed, whereas a “freeze” signifies everything needed is complete. With such tight schedules, the distinction is not trivial. Stage 3, Protocol Detail will be frozen by December 2028. The Final ASN.1 and Open API Freeze will occur in March 2029; at which time, the specifications are truly code complete. Demonstrations of pre-commercial systems should start around 2028, while commercially viable systems (globally interoperable) can be expected starting sometime after 2029 through 2030.
None of it is arbitrary. All of the framework that exists today has been developed backward from the International Telecommunications Union (ITU) schedule for the IMT-2030 calendar. Candidate technologies need to be submitted to the ITU by early 2029; therefore, 3GPP needed to finalize its standards process as soon as possible after 3GPP allowed for. As for the engineering timeframes, they become somewhat compresses. RAN1 will run from Spring 2027 through mid-2028, and RAN2, 3, and 4 will run from Fall 2027 through the end of 2028. That is six quarterly periods total to define an entire generation of radios and an evolved and/or new core. Those who have attended a standardization process for 5G know what it feels like to be under a very tight timeframe. This one is even tighter.
One detail shows how little room there is. The March 2027 "package approval" for 6G is really sign-off on a placeholder, with the actual RAN content finalized three months later. That's not an oversight, it's how 3GPP runs concurrency on purpose, the equivalent of pouring the foundation while the upper floors are still being drawn. It works. But there's almost no slack in it, and slack is what absorbs the problems nobody saw coming.
What hasn't been decided
A locked schedule isn't a finished design, and some of the open questions are large. The biggest, if you operate a network, is migration: how you actually move from 5G to 6G, and what it costs you in hardware. Operators have been asking for a decision on the migration architecture for more than a year. At the same meeting that approved the timeline, the chair said plainly that no decision was possible yet on the migration options. A few people I trust called that the most important thing said all week, and I'd agree. The deadline is fixed, the destination is fixed, and the route between them is still an open argument. If your job is running a network, that's the gap you're staring into.
A smaller decision is more telling than it sounds. The group set the baseline transmit power for coverage modeling at 23 dBm; some vendors pushed for 26. The operators won, and the reason they gave is the line I'd pin above the whole exercise: if you model 6G coverage on a stronger transmitter than 5G used, every coverage "gain" turns into marketing nobody can disprove. That's operators refusing to let the comparison quietly drift in their favor, because they're the ones who'll have to deploy against whatever the standard assumes. The specs are being written by people who will personally have to live with the distance between what's claimed and what holds up in the field.
For this readership, that distance is most of the job. OSS, BSS, the assurance stack, day-two operations, none of it runs against a slide. It runs against a live network. And the live 6G network is going to be harder to run than anything we've operated before.
"AI-native" is an operations problem, not a feature
6G is being built as the first AI-native mobile network, and I want to be careful with the phrase, because it gets thrown around like a sticker. It doesn't mean the network ships with a few AI features bolted on. It means the network is being designed so AI can become part of the way it senses conditions, predicts what will happen next, and optimizes radio, compute, energy, and service resources. In 3GPP this direction is already emerging through 5G-Advanced work on AI/ML for the NR air interface, including channel-state information, beam management, positioning, and mobility.
