No 6G network will carry a commercial call before 2030, and nothing about 6G is standardized yet. But four separate things are being built anyway: 3GPP locked in a fixed 6G specification timeline at a June 2026 plenary, regulators in the US and China are actively clearing the spectrum 6G will run on, and Nokia, Ericsson, and Samsung are already shipping — and openly disagreeing over — the AI-native radio infrastructure meant to carry it. Most coverage treats all of this as one countdown to a single launch date. It's four projects moving on four different clocks, and only one of them has a date fixed in writing.

The confusion starts with what "building" means here

Most 6G writing still leans on the 2020-era pitch: holographic calls, tactile internet, terabit downloads to a contact lens. None of that is being built in any concrete sense in 2026 — it's a usage-scenario wish list that predates any working system. What's actually happening this year sits one level down from that pitch: standards bodies locking in dates, regulators moving frequency bands out from under incumbent users, and equipment vendors shipping software that will run on the same radios 6G eventually uses. None of it is 6G yet. All of it is the specific, checkable work that has to finish before 6G can exist. That distinction is why "is 6G real yet" is the wrong question. The better one is which of the four workstreams below is actually on schedule.

The standards clock has fixed dates now

3GPP spent 2025 studying 6G use cases; it spent the first half of 2026 turning that study into a locked calendar. At a June 2026 plenary in Singapore, 3GPP agreed that Release 21 — the release that writes the actual 6G specifications — will run from March 2027 to the end of 2028, with a functional freeze targeted for December 2028 and the first formal 6G specifications available in early 2029. Ericsson, which supports the timeline, expects the first commercial 6G systems in 2030. The groundwork for that release is already measurably underway. 3GPP's TR 22.870, the Stage 1 study defining 6G use cases and service requirements, was approved at TSG SA's 111th plenary, and the RAN working group's parallel study on 6G scenarios and requirements was 60% complete as of March 2026. That's a hard date on paper, not a network. Whether Release 21 actually freezes in December 2028 is the single most falsifiable claim in this entire story — check back then.

Two governments are clearing spectrum in opposite ways

6G needs wide, contiguous channels in what's called the upper mid-band, and the US and China are getting there through very different processes. In the US, the FCC voted to move forward with an Upper C-Band auction covering 3.98–4.14 GHz — 160 MHz that, combined with existing lower C-Band holdings, creates a contiguous 440 MHz block. The auction itself won't complete until no later than July 2027, and winners have to help fund a multi-year relocation of the satellite operators and aircraft altimeters currently using that spectrum. Separately, NTIA is studying the 7.125–7.4 GHz band as the leading US candidate for licensed 6G capacity, with a final report due by the end of 2026, according to remarks from NTIA chief administrator Arielle Roth at May's CTIA Summit. China moved faster by regulating differently. In May 2026, the Ministry of Industry and Information Technology granted the IMT-2030 (6G) Promotion Group a trial frequency permit for the 6,425–7,125 MHz band — making China the first country to dedicate that specific range to 6G field testing rather than route it through a multi-year auction and relocation process. Neither country has actually allocated commercial 6G spectrum yet. What's real is that both governments spent 2026 doing the specific regulatory work — auctions, studies, trial permits — that has to happen before either one can.

The infrastructure shipping today is AI-RAN, and the vendors don't agree on how to build it

The one piece of "6G-adjacent" technology actually running on live networks in 2026 isn't 6G radio at all — it's AI-RAN, software that puts machine-learning models directly into the radio access network to improve spectral efficiency ahead of 6G's arrival. Ericsson launched its AI in RAN product commercially in June 2026 as a software-only subscription that runs on hardware already in the field, reporting up to 20% higher downlink throughput and 10% better spectral efficiency across more than 15 deployments and trials. Nokia's competing platform went commercial in July 2026; nine operators are engaged, but only T-Mobile has a confirmed public timeline — pilots by the end of 2026, commercial service in 2027, mass deployment by 2028. Where the vendors genuinely split is on what chip should run the AI models. Nokia is betting on Nvidia GPUs, targeting a 100% spectral-efficiency gain by 2028, backed by a $1 billion investment Nvidia made in the partnership in October 2025. Ericsson argues that radio physics caps useful AI models at a size where purpose-built ASICs beat GPUs on cost and power, and has started embedding its own "neural network accelerator" chip directly into its Massive MIMO radios. Samsung has taken a third position, running a hardware-agnostic, CPU-first approach — though its own AI-RAN work so far consists of a multi-cell test inside Samsung's own R&D center, not a live deployment. All three vendors have a direct financial stake in their own answer being correct. That's worth weighing against their performance numbers specifically, since none of them are disinterested. The one figure from a party with nothing to sell either way comes from Dell'Oro Group, which projects $35 billion in cumulative AI-RAN revenue between 2026 and 2030 — while explicitly not expecting AI-RAN to grow the overall RAN market, and pricing the GPU-specific slice of that market at just over $1 billion.

China's field trials are running a published three-stage script

China's 6G program isn't a single announcement — it's a documented sequence. The first stage, focused on validating core technologies, is complete. The second stage, testing full technical solutions in real environments, is underway, enabled by MIIT's May 2026 trial frequency permit to the IMT-2030 (6G) Promotion Group, a CAICT-led body whose membership includes Huawei, ZTE, Xiaomi, Ericsson, and Nokia. A third stage, testing full system networking, is still to come. The clearest physical marker of that second stage is in Nanjing, where Purple Mountain Laboratories put what it describes as China's first Pre-6G outdoor trial network into live operation in April 2026, running alongside — not replacing — existing 5G infrastructure. A separate pilot program launched in June 2026 extends testing into specific verticals: industrial manufacturing, smart maritime operations, embodied intelligence, and the low-altitude drone economy. This work sits inside China's current Five-Year Plan, and CAICT has projected a $147 billion domestic 6G market by 2035. The trials are real and running outdoors. Whether they translate into a standards or deployment lead by 2030 is a separate, unresolved question.

The one 6G use case already tested outside a lab

Integrated Sensing and Communication, or ISAC, turns a cellular network into a distributed sensor — detecting motion, position, and objects through the radio signal itself, without a separate camera or radar. Verizon's 6G Innovation Forum, whose founding members include Ericsson, Samsung, Nokia, Meta, and Qualcomm, added nine new members this month, including AWS, Cisco, Intel, and Nvidia. Verizon also disclosed successful real-world ISAC field trials conducted with founding members Samsung and Qualcomm. That's a meaningful distinction from the AI-RAN chip disagreement above. Samsung's AI-RAN silicon work is still confined to its own R&D center; its ISAC work with Verizon has actually run in the field. Not every 6G-adjacent claim from the same vendor carries the same level of evidence.

What none of this gets you yet

There's no 6G phone in 2026, and there won't be one before commercial networks exist around 2030, on 3GPP's own timeline. Most of what gets reported as "6G news" this year — the AI-RAN launches, the spectrum auctions, the field trials — is precursor infrastructure and regulatory groundwork, not 6G itself. The 6G that eventually ships in 2030 will very likely run on radios that carriers are installing and marketing as 5G-Advanced upgrades right now. If there's a single number worth tracking over the next two years, it's whether Release 21 actually hits its December 2028 freeze date — and whether any operator besides T-Mobile puts a confirmed date on an AI-RAN deployment.

FAQ

When will 6G actually launch commercially?

3GPP's Release 21, which writes the formal 6G specifications, is scheduled to run from March 2027 to the end of 2028 with a December 2028 functional freeze, producing first specifications by early 2029. Ericsson, working within that timeline, expects the first commercial 6G systems around 2030.

What spectrum will 6G actually use?

The leading global candidate is the upper mid-band. In the US, NTIA is studying 7.125–7.4 GHz as the frontrunner, with a final report due by the end of 2026, while the FCC separately moved forward with a 440 MHz Upper C-Band auction to complete by July 2027. China took a faster regulatory path, granting a trial permit for the 6,425–7,125 MHz band to its IMT-2030 Promotion Group in May 2026.

What is AI-RAN, and is it the same thing as 6G?

No — it's software that adds machine-learning models to today's radio networks to improve efficiency ahead of 6G. Ericsson launched a commercial AI-RAN product in June 2026 that runs on existing hardware, while Nokia launched a competing GPU-based platform in July 2026 with nine operators engaged, though only T-Mobile has a confirmed deployment timeline.

Why do Nokia, Ericsson, and Samsung disagree about AI-RAN hardware?

Nokia is betting on Nvidia GPUs, backed by a $1 billion Nvidia investment, targeting a 100% spectral-efficiency gain by 2028. Ericsson argues physics limits useful AI models to a size where custom ASICs beat GPUs, and has built its own AI accelerator chip into its radios. Samsung is pursuing a hardware-agnostic, CPU-first approach, though its testing so far has been confined to its own R&D center rather than a live network.

Is China ahead of the US in building 6G?

China's field-trial program is running a documented three-stage process, with the second stage now underway and a live outdoor Pre-6G trial network operating in Nanjing since April 2026. China also moved faster to grant dedicated 6 GHz trial spectrum than the US has moved on its own upper mid-band process. Whether that translates into a lasting standards or deployment lead by 2030 remains an open question.

What is ISAC, and has it actually been tested?

Integrated Sensing and Communication uses the cellular signal itself to detect motion and objects, functioning like a distributed radar without a physical camera. Verizon's 6G Innovation Forum has already run real-world ISAC field trials with founding members Samsung and Qualcomm, making it one of the few 6G-adjacent capabilities validated outside a lab setting.