Skip to main content
6G · 3GPP Quarterly March 2026

6G locks its radio numbers while AI priorities stay a fight

400 MHz channels, a 410 MHz–52.6 GHz spectrum band and 60 bit/s/Hz peak efficiency are set; OPPO and Samsung blocked Ericsson and Huawei's push to rank AI use cases now.

30 decisions10 work areas3,056 contributions

Plenary cycle

SA#111 · RAN#111 · CT#111

Fukuoka · 9 Mar 2026 – 13 Mar 2026

Source meetings for this article (12)

Meetings in this article’s source material. Inclusion does not mean that every meeting approved every decision.

3GPP QuarterlyThe next issue by email, at most eight a year.

OPPO, Samsung, LG and Lenovo blocked an attempt by Futurewei, Ericsson, ZTE, Huawei and AT&T to have 3GPP's plenary pick winners among 6G's AI/ML use cases this quarter. OPPO's objection was blunt: without a non-AI benchmark to compare against, there's no evidence yet that any AI use case delivers a performance gain worth standardizing. The result is that each working group keeps managing its own AI priorities rather than following a top-down list — a deferral that keeps dozens of competing AI proposals alive rather than culling them.

The concrete numbers that did land are large. 6G base stations operating around 7 GHz will support a single 400 MHz downlink channel and 200 MHz uplink — double 5G NR's 200 MHz maximum — on one component carrier, without needing carrier aggregation. RAN also approved a spectrum range of 410 MHz to 52.6 GHz for terrestrial 6G, covering existing 5G bands, plus support for satellite (NTN) access from the outset. Peak spectral efficiency targets were set at 60 bit/s/Hz downlink and 30 bit/s/Hz uplink, alongside a control-plane latency ceiling of 20 ms and a connection density target of 10^6 devices per square kilometer.

Bandwidth for the cheapest 6G devices remains unresolved. Apple, moderating the discussion, floated a compromise between companies wanting a flat 20 MHz for both downlink and uplink and others wanting as little as 5 MHz for FDD or 10 MHz for TDD, with half-duplex FDD proposed to cut cost for the lowest-tier IoT devices. No alternative was chosen; RAN working groups were instead told to analyze complexity and cost further. Separately, MediaTek led sixteen companies — including DOCOMO, Nokia and CMCC — in cataloguing why 5G's Bandwidth Part framework should not be copied into 6G, citing excessive signaling overhead and switching latency; Samsung, Qualcomm, Xiaomi and Apple pushed back on adopting that critique wholesale.

On AI specifically, 3GPP picked its first three study cases: network-side digital post-distortion to fix power-amplifier signal errors, SRS power-imbalance compensation, and beam-level prediction for handovers. Ericsson, LG and Huawei separately pushed to deprioritize any AI model requiring coordination between network and device ('two-sided' models) in favor of simpler 'one-sided' ones — a preference InterDigital echoed, while InterDigital also warned that AI's hidden costs in signaling, hardware and energy could outweigh its radio gains. None of this was adopted as binding; it remains company opinion feeding into later study.

Watch RAN's September 2026 meeting, where working groups revisit coordination on overlapping AI mobility-prediction studies — the outcome will determine whether RAN1's signal-prediction work and RAN4's beam/cell-prediction work stay aligned or start producing incompatible assumptions. Also unresolved: whether 6G's minimum device bandwidth lands at 20 MHz or a narrower tier, a choice that will set the cost floor for the simplest 6G devices.

The decisions behind this

Deferred
Study on 6G Radio
R1-2600951

6G bandwidth operation will be studied to avoid 5G's complexity and latency issues

This document is a summary of proposals for the 6G bandwidth operation study. It compiles views from many companies on what should be studied, based on lessons learned from the 5G NR Bandwidth Part (BWP) framework. A BWP is a contiguous set of physical resource blocks configured for a user equipment (UE) within a carrier's bandwidth, used for power saving and supporting devices with different capabilities.

The summary notes a strong consensus that the 5G NR BWP framework, while functional, has significant drawbacks. The primary issues identified are: 1) Excessive configuration overhead due to too many parameters set per BWP, leading to heavy signaling. 2) Considerable switching latency and service interruption when changing BWPs. 3) Reliability issues with DCI-based BWP switching, where a missed command can cause misalignment between the UE and network. 4) Restrictive constraints, such as requiring the downlink and uplink BWP to share the same center frequency in Time Division Duplex (TDD) mode.

A group of companies, led by the moderator MediaTek, proposed a list of these 'lessons learned' from 5G to guide the 6G study. The proposal, labeled 'Proposal 3.1-1a', stated: 'Lessons learned from NR BWP framework include: Excessive configuration overhead due to unnecessary flexibility; Complicated numerology changes under BWP framework; Considerable switching latency and service interruption; Ambiguity issues of DCI-based BWP switching; Restrictive constraints about DL/UL BWP relation such as DL/UL BWP coupling and center frequency alignment in TDD.'

The meeting did not adopt this specific proposal as a formal agreement. The document's status is 'noted', meaning it was taken for information. The moderator indicated they would close discussion on this specific list to focus on more important proposals about use cases and design considerations for 6G.

Support listing these specific 5G lessons to guide 6G designOppose or seek changes to the proposed list of lessons
MediaTek · Tejas Networks · DOCOMO · Spreadtrum · OPPO · vivo · NEC · CMCC · ETRI · Nokia · CEWiT · LG Electronics · Sharp · Panasonic · Fujitsu · Futurewei · TCL17 companiesSamsung · Qualcomm · Xiaomi · Apple4 companies
Why it matters

The working group acknowledges widespread industry criticism of the 5G BWP design but has not yet formally agreed on a specific list of flaws. The discussion will move forward to defining what 6G should achieve, with the clear intent to create a simpler, faster, and more reliable bandwidth adaptation framework.

Deferred
Study on 6G Radio
R1-2600951

Study proposed for 6G bandwidth operation on fragmented spectrum and multi-carrier cells

A key challenge for future networks is efficiently using fragmented and irregular blocks of spectrum owned by operators. In 5G NR, a Bandwidth Part (BWP) must be a contiguous set of frequencies within a single carrier, which can limit flexible use of non-adjacent spectrum blocks.

A coalition of companies proposed that the 6G study should analyze bandwidth operation for non-contiguous frequency resources and multi-carrier cells. The specific proposal, 'Proposal 7.3-1a', stated: 'Study bandwidth operation with non-contiguous frequency resource allocation and multi-carrier operation. The analysis should compare the signaling efficiency and complexity of a single non-contiguous BWP versus multiple simultaneous active BWPs.'

The meeting did not decide on this proposal. The document's status is 'noted'. Some supporters, like Spreadtrum, suggested rewording it to 'Study whether/how to support...'. Opponents like Samsung argued there is 'no reason to discuss multiple active BWPs per cell' and that the fundamental concept of a BWP needs to be agreed first. Nokia and DOCOMO suggested the topic should be coordinated with another agenda item discussing multi-carrier frameworks.

Support studying non-contiguous and multi-carrier bandwidth operationOppose or believe the proposal is premature
MediaTek · Tejas Networks · Spreadtrum · OPPO · NEC · CMCC · CEWiT · Panasonic · Futurewei9 companiesSamsung · Xiaomi · Nokia · DOCOMO4 companies
Why it matters

There is industry interest in designing 6G to natively support patchwork spectrum, but fundamental architectural questions remain open. The core debate is whether to allow one complex BWP that spans disjoint frequencies or to let a device operate multiple simpler BWPs at once, each on a different carrier.

  • DeferredStudy proposed for 6G bandwidth features to support Sub-band Full DuplexR1-2600951
  • Deferred6G device type study confirms minimum 3MHz operation and pushes bandwidth decision to JuneRP-260720
  • Deferred6G base stations will support 400MHz downlink and 200MHz uplink as a baseline for smartphonesRP-260746
  • DeferredCompanies clash over whether to prioritize AI use cases now, plenary defers decisionRP-260747
  • DeferredSA4 defers decision on normative work for AI data formatsS4-260323
  • Deferred6G architecture study to examine using 3GPP NTN access for satellite feeder linksS2-2601605
Quotations are verbatim from moderator summaries in the 3GPP archive. Document numbers link to the original files. Back to the full report.