What Is 6G? Targets, Timeline, and What's Actually Different

Every generation gets oversold in its research phase, and 6G is no exception. Terabit peak rates and microsecond latency circulate in slide decks well ahead of anything resembling a specification.

So it's worth separating three different things: what ITU-R has formally agreed as the framework, what 3GPP is actually working on, and what remains research aspiration.

The formal framework exists. ITU-R Recommendation M.2160, approved in late 2023, defines the IMT-2030 vision. That's the closest thing to an authoritative statement of what 6G is meant to be.

---

Why another generation

5G delivered enhanced mobile broadband, low-latency communication, and massive IoT connectivity. The honest question is what genuinely needs more.

The credible drivers aren't peak data rate. They're structural:

Sensing as a network function. 5G networks move data. 6G is expected to also perceive the physical environment using the same radio infrastructure — a genuinely new capability rather than an increment.

AI as architecture rather than add-on. 5G-Advanced bolts machine learning onto existing functions. 6G is being designed with data pipelines, model lifecycle, and inference placement as architectural elements.

Coverage gaps. Non-terrestrial networks integrating satellite and terrestrial access into one system, rather than the parallel systems we have today.

Energy. RAN energy is a major operational cost and an increasingly binding sustainability constraint. Efficiency per bit is a first-class 6G design target, not a nice-to-have.

The less credible drivers are the ones that lead with holographic telepresence. Those applications have been promised since 5G research and haven't materialised — not because networks were too slow, but because the rest of the stack, the content, and the economics weren't there.

---

The IMT-2030 capability targets

M.2160 defines fifteen capabilities, nine inherited from IMT-2020 and six new. The headline figures:

Two things stand out.

Peak rate is 50–200 Gbps, not terabits. The terabit figure that circulates comes from early academic papers, not the agreed framework. Anyone quoting it is working from research aspiration rather than the standard.

The genuinely new capabilities aren't speed. M.2160 adds integrated sensing and communication, integrated AI and communication, ubiquitous connectivity, sustainability, security and resilience, and interoperability as formal capability dimensions. That's the real signal — 6G is being framed as a change in what the network does, not how fast it does it.

The usage scenarios follow the same shape: immersive communication, massive communication, and hyper-reliable low-latency communication extend 5G's three. Ubiquitous connectivity, AI and communication, and integrated sensing and communication are new.

---

The four technology pillars

AI-native design

The distinction that matters: an AI-assisted network adds models to selected functions. An AI-native network treats data, models, compute, and governance as architectural elements from the start.

In practice this means learned components in the air interface — channel estimation, beam selection, CSI compression — alongside cross-domain optimisation and closed-loop automation with policy guardrails. The hard part isn't the models. It's the data pipeline, model lifecycle management, and explainability. (Covered in more depth in AI-native networks.)

Integrated sensing and communication

Using the same waveform and infrastructure for both data transmission and environmental sensing — detecting objects, tracking movement, and building a spatial picture from radio reflections.

The physics is well understood; the difficulty is doing it without sacrificing the throughput you were selling. Privacy is likely to be as decisive as the technology, since the network would detect people who aren't subscribers and never consented. (ISAC in detail.)

New spectrum

Three ranges are in play, and the interesting one isn't the highest.

Upper mid-band (7–24 GHz), often called FR3 or the "golden band," is where most serious deployment attention sits. It offers meaningfully more bandwidth than FR1 with propagation far better than mmWave — a genuine sweet spot, and the reason WRC-27 spectrum decisions matter so much.

Sub-THz (above 100 GHz) offers enormous bandwidth over very short range. Realistic applications are fixed wireless backhaul, in-room coverage, and sensing — not mobile coverage.

Existing spectrum remains the backbone. 6G will refarm sub-7 GHz spectrum, as every generation has.

Sustainability

Energy efficiency as a design target rather than an afterthought: network-level sleep modes, AI-driven energy optimisation, efficiency measured per bit rather than as absolute consumption, and lifecycle considerations for equipment. Given RAN energy costs and regulatory pressure, this is one of the more commercially grounded pillars.

Two more deserve mention. Non-terrestrial networks integrate satellite access natively rather than as an overlay — building on the NTN work already in 5G Releases 17 and 18. And network as a compute platform extends edge computing so the network exposes compute and inference as services, not just connectivity.

---

The timeline

This is where expectations most often run ahead of reality.

2021–2023 — ITU-R vision development, concluding with M.2160 approval in November 2023.

2024–2025 — 3GPP Release 20 study phase. 6G studies formally began in 2025, running alongside continued 5G-Advanced work.

2026–2028 — Release 21 is expected to carry the first 6G normative specifications. This is the release to watch.

2028–2030 — specification completion, IMT-2030 submission and evaluation, early trials.

2030 and beyond — first commercial deployments, following the roughly ten-year generational cadence.

Two caveats worth holding onto. Spectrum decisions at WRC-27 will shape what's actually deployable, particularly in the upper mid-band — technology timelines depend on regulatory ones. And first commercial 6G will look modest. First-release 5G was NSA with limited features; the interesting capabilities arrived years later. Expect the same shape.

---

How 6G builds on 5G-Advanced

The framing of a clean generational break is misleading. Much of what gets labelled 6G is already progressing through 5G-Advanced.

Release 18 introduced AI/ML for the NR air interface, network energy saving, and NTN enhancements. Releases 19 and 20 extend AI/ML, add ISAC study work, and continue positioning and energy work.

The pattern is consistent: capabilities are studied in 5G-Advanced, mature there, and get architected properly in 6G. It's evolution with a rebranding milestone, not a discontinuity — which is also how 4G became 5G.

---

What to be sceptical about

Terabit data rates. Not in the agreed framework. 50–200 Gbps peak, and peak rate is a lab figure regardless of generation.

Microsecond latency. M.2160 says 0.1–1 ms. Sub-100µs over a radio link with real processing isn't a serious near-term target.

Holographic communication as a driver. Promised for 5G, didn't happen. The bottleneck was never the network.

"6G will replace fibre." Every generation attracts this claim. It hasn't happened yet.

Firm dates before 2030. Any deployment claim earlier than that is marketing ahead of the standardisation calendar.

---

The takeaway

IMT-2030 (M.2160) is the agreed framework. 50–200 Gbps peak, 0.1–1 ms latency, 1.5–3× spectral efficiency.

The new capabilities are sensing, AI, ubiquitous connectivity, and sustainability — not speed.

Upper mid-band (7–24 GHz) is the spectrum that matters most, not sub-THz.

Release 21 (2026–2028) carries the first normative 6G specifications. Commercial deployment around 2030.

Much of 6G is already in 5G-Advanced, being matured before it gets architected properly.

The useful way to think about 6G is as the generation where the network stops being purely a data pipe. Whether that vision survives contact with deployment economics is the genuinely open question — and it's worth remembering that 5G's most-hyped capabilities were not the ones that ended up mattering.

---

Further reading

• ITU-R Recommendation M.2160 — IMT-2030 framework and overall objectives
• 3GPP Release 20 — 6G study items
• Integrated sensing and communication
• AI-native networks
• Hexa-X and Hexa-X-II — European 6G flagship projects