5G is the fifth generation of mobile networks, the 3GPP system that began with Release 15 (2017 to 2019) and now carries about half the world's mobile data traffic. 6G, which ITU calls IMT-2030, is still being standardized: 3GPP plans its first 6G specifications in Release 21, frozen by March 2029, and first commercial services are expected around 2030.
This guide explains what 5G is in standards terms (New Radio, non-standalone and standalone, the frequency ranges, network slicing, URLLC and massive IoT), what 5G-Advanced adds, where 6G stands as of September 2026 and how it differs from 5G, and what 5G means for a business today: fixed wireless access, private 5G networks and edge computing.
What 5G is: the standards behind the name
A mobile generation is defined in two places. ITU-R, the radio sector of the International Telecommunication Union, sets the requirements a technology must meet to count as that generation; for 5G the label is IMT-2020. 3GPP, the standards project that has written the mobile specifications since 3G, then specifies a system that meets them: not only the radio, but also the core network, the protocols and the interfaces that let equipment from different vendors and operators work together.
ITU's IMT-2020 vision, Recommendation ITU-R M.2083 (September 2015), described three usage scenarios. Its minimum technical requirements, Report ITU-R M.2410 (November 2017), put numbers on each:
- Enhanced mobile broadband (eMBB): more capacity and speed for phones and computers. The minimum peak data rate is 20 Gbit/s downlink and 10 Gbit/s uplink, and the radio's user plane latency is at most 4 ms.
- Ultra-reliable low-latency communication (URLLC): traffic that must arrive on time, such as machine control. The latency requirement is 1 ms, and a 32-byte packet must be delivered within 1 ms with a success probability of 99.999% at the edge of the cell.
- Massive machine-type communication (mMTC): very large numbers of simple devices, at a minimum connection density of 1 million devices per square kilometre.
These are requirements for evaluating a radio technology under defined test conditions, such as a single user in an unloaded cell. They are not the speeds or delays a phone sees: those depend on the spectrum an operator holds, the distance to the mast, the load on the cell and the device.
3GPP delivered 5G in phases, one release at a time:
- Release 15: the first New Radio specifications, for non-standalone operation, arrived at the end of 2017. Release 15 then grew into the first full set of 5G standards, including standalone 5G with a new core network. Its protocols were stable in June 2019, according to 3GPP's release timeline.
- Release 16: completed on July 3, 2020, and with it 3GPP's submission to ITU for IMT-2020. Its release description aims 5G at industry: enhancements to URLLC, network slicing, edge computing and cellular IoT, plus non-public (private) networks, positioning, vehicle-to-everything (V2X) communication over NR sidelink, NR in unlicensed spectrum and integration with IEEE time-sensitive networking.
- Release 17: functionally frozen in March 2022. It added reduced-capability (RedCap) NR devices, NR up to 71 GHz, NR over non-terrestrial networks (satellites), enhanced support for non-public networks and support for edge computing in the 5G Core.
ITU published the IMT-2020 specifications as Recommendation ITU-R M.2150 in February 2021.
How 5G New Radio works, and its frequency ranges
5G's radio interface is called New Radio (NR), and its base station is the gNB. For the downlink, NR uses OFDM with a cyclic prefix, as LTE does. The uplink can use the same, or DFT-s-OFDM, which improves uplink coverage but carries a single layer only. A gNB can also be split into a central unit and one or more distributed units, linked by the F1 interface.
NR is specified for two frequency ranges. TS 38.104, the base station radio specification, defines them as follows (version 19.5.0, August 2026):
| Range | Frequencies | What it is good at |
|---|---|---|
| FR1 | 410–7,125 MHz | Includes the bands below 1 GHz that cover large rural areas and the 1 to 6 GHz bands used in cities and suburbs. Carriers up to 100 MHz wide below 6 GHz |
| FR2-1 | 24.25–52.6 GHz | Millimetre wave: short range, weaker propagation, carriers up to 400 MHz wide, for dense hot spots |
| FR2-2 | 52.6–71 GHz | The millimetre-wave extension added in Release 17 |
The trade-off is physics. Lower frequencies travel further and reach deeper indoors; higher frequencies carry wider channels over shorter distances. This is why the same "5G" logo can mean very different performance from one network, or one street, to the next.
Non-standalone vs standalone 5G
3GPP defines two ways to deploy 5G:
- Non-standalone (NSA), also called EN-DC or Architecture Option 3. NR radios are added to an existing 4G LTE network and its core, the Evolved Packet Core (EPC). The LTE base station is the master node and the 5G base station the secondary node, so a phone holds two radio links at once. Operators get NR capacity without replacing the network, but only 4G services are supported.
- Standalone (SA). NR connects to the new 5G Core (5GC), and 3GPP notes that only in this configuration is the full set of 5G services supported. The 5GC is a service-based architecture of network functions: the AMF handles access and mobility, the SMF manages sessions and the UPF carries user data.
The difference matters because the features businesses ask about live in the 5GC: network slicing, private networks integrated with a public one, and native support for edge computing. According to Ericsson's June 2026 subscriptions outlook, around 390 service providers have launched commercial 5G and more than 90 of them have launched 5G standalone (GSA and Ericsson data, May 2026). The others offer non-standalone 5G only.
Important
A 5G icon on a device does not tell you whether the network is standalone. If a use case depends on slicing, a private network run by an operator or edge computing, ask the operator whether it is delivered on 5G standalone at your sites, and put that in the contract.
Network slicing, URLLC and massive IoT
Network slicing is the ability to run several logical networks over one physical network, each with its own capabilities and characteristics. In TS 23.501, the 5G system architecture, each slice is identified by an S-NSSAI: a slice/service type (SST) plus an optional slice differentiator that separates slices of the same type. Version 19.8.0 (July 2026) standardizes seven SST values: eMBB (1), URLLC (2), massive IoT (3), V2X (4), high-performance machine-type communication (5), high data rate and low latency (6) and guaranteed bit rate streaming (7). An operator does not have to support all of them.
3GPP's own example: one slice for the operator's ordinary subscribers, one for a virtual operator's, and one for a single service such as tracking cargo containers. Commercial offers are growing: Ericsson counted 84 commercial 5G standalone slicing offerings in June 2026, up from 65 six months earlier.
URLLC is the part of 5G designed for control traffic. 3GPP's own service requirements give an example: remote control of process automation expects 99.9999% reliability with an end-to-end latency of 50 ms, which 3GPP says is provided in particular through edge computing. The radio can be fast; the rest of the path has to be short too. Running AI models at that edge, and inside the radio network itself, is covered in our guide to AI and 5G.
Massive IoT covers large numbers of low-cost, low-power devices. Release 17 added RedCap, a reduced-capability device type with lower complexity and power use, aimed at industrial wireless sensors, video surveillance and wearables, and Release 18 extended IoT support, including coverage by satellite.
5G-Advanced: Releases 18, 19 and 20
5G-Advanced is the name 3GPP gives to 5G from Release 18 onward, with its own logo, adopted in April 2021. It is not a new generation: Release 18 enhances features from earlier releases and introduces new topics.
| Release | Status as of September 2026 | Main themes |
|---|---|---|
| Release 18 | Complete: stage 3 frozen March 2024, protocols stable June 2024 | AI/ML, extended reality (XR), energy efficiency; more satellite access, IoT, positioning, industrial features, slicing and drone support |
| Release 19 | Complete: fully implementable specifications at the end of December 2025 | The second 5G-Advanced release, consolidating earlier work |
| Release 20 | Open: stage 3 freeze targeted for March 2027, ASN.1 and OpenAPI freeze for June 2027 | The last "5G-Advanced only" release: energy efficiency, AI/ML, satellite access, mobility, coverage, public safety, sensing with communication |
Release 20 runs two tracks at once: the last 5G-Advanced features, with 126 work items and 74 study items under way in March 2026, and the first 6G studies, covering use cases, radio, architecture, security, management and more.
For a business, 5G-Advanced arrives as operator and equipment upgrades, not as a new network to buy.
What is 6G? The IMT-2030 framework
6G is the next generation, and IMT-2030 is its official name: ITU's Radiocommunication Assembly confirmed it in November 2023 and approved Recommendation ITU-R M.2160, the "IMT-2030 Framework". It plays the role M.2083 played for 5G: the goals that standards bodies then turn into specifications.
M.2160 describes six usage scenarios. Three extend 5G's, and three are new:
| IMT-2030 usage scenario | Extends | Typical use cases in M.2160 |
|---|---|---|
| Immersive communication | eMBB | Immersive XR, remote multi-sensory telepresence, holographic communication |
| Hyper reliable and low-latency communication (HRLLC) | URLLC | Full automation and control in industry, emergency services, telemedicine, power grid monitoring |
| Massive communication | mMTC | Smart cities, logistics, agriculture, environmental monitoring, devices without batteries or with long-life batteries |
| Ubiquitous connectivity | New | Coverage for rural, remote and sparsely populated areas |
| Artificial intelligence and communication | New | Distributed AI training and inference, offloading heavy computation, digital twins, assisted driving |
| Integrated sensing and communication (ISAC) | New | Navigation, movement tracking such as fall detection, vehicle and pedestrian detection, environmental sensing |
Four overarching aspects apply to all six: sustainability, security and resilience, connecting the unconnected, and ubiquitous intelligence.
3GPP has started turning this into requirements. TR 22.870, "6G Use Cases and Service Requirements", was approved in March 2026 and anchors the first 6G release. TR 38.914, "6G Scenarios and Requirements", was approved in June 2026, and at the same June plenaries 3GPP agreed its basic terms: 6GR for the 6G radio, 6GC for the 6G core network, 6GS for the 6G system. Studies on the radio itself, including waveform, numerology, channel coding and the migration from 5G, continue at working group level.
When will 6G come out? The timeline as of September 2026
As with earlier generations, ITU sets the 6G calendar and 3GPP develops the technology it submits. 3GPP decided that 6G needs two releases: Release 20 for studies and Release 21 for the specifications. The Release 21 timeline was agreed at the June 2026 plenary meetings.
| Milestone | Date | Status |
|---|---|---|
| IMT-2030 framework, Recommendation ITU-R M.2160 | November 2023 | Approved |
| 3GPP 6G use cases and service requirements (TR 22.870) | March 2026 | Approved |
| IMT-2030 minimum technical performance requirements | Draft completed February 2026 | Up for approval in December 2026 |
| 3GPP 6G scenarios and requirements (TR 38.914) | June 2026 | Approved |
| ITU window for candidate IMT-2030 radio technologies | February 2027 to February 2029 | Planned |
| Release 21 stage 1 freeze (service requirements) | March 2027 | Planned |
| Release 21 stage 2 freeze (architecture) | June 2028 | Planned |
| Release 21 stage 3 freeze (protocols): the first 6G specifications | December 2028 | Planned |
| Release 21 ASN.1 and OpenAPI freeze | March 2029 | Planned |
| Specifications submitted to ITU | By mid-2030 at the latest | ITU target |
| First commercial 6G services | Around 2030, varying by country | Ericsson forecast |
The stages are 3GPP's way of building a release. Stage 1 fixes the service requirements, stage 2 the architecture and stage 3 the concrete protocols. The ASN.1 and OpenAPI freeze, which fixes the formal definitions of messages and interfaces, is the date 3GPP lists as protocols stable. The dates for the ITU window come from ITU-R Working Party 5D, and the Release 20 page gives ITU's target of complete specifications by mid-2030.
Note
Everything after September 2026 in this table is a plan. 3GPP's own release table says future dates are proposals, subject to change, and Release 16 slipped by three months in 2020. Treat any product sold as "6G" before the Release 21 specifications freeze as pre-standard.
On the market side, Ericsson's June 2026 report expects the first commercial 6G services around 2030, with the US, China, Japan, South Korea and the Gulf Cooperation Council countries as likely early adopters, and forecasts 180 million 6G subscriptions by the end of 2031. That is a forecast from an equipment vendor, not a standard.
What 6G research is exploring
The IMT-2030 framework says what 6G should do, not how. These are the directions research and 3GPP studies are taking as of September 2026. None of them is a decided 6G feature yet.
New spectrum, from 7 GHz upward
Each generation has added spectrum, and 6G is expected to keep today's low and mid bands. In a June 2026 article on 6G radio requirements, the chair of RAN4, the 3GPP group that sets radio frequency and performance requirements, describes the range around 7 GHz as one of the leading 6G spectrum candidates, with wider channels than today's mid-band and manageable propagation loss. The extended upper mid-band, around 10 to 15 GHz, remains under active study with open questions, and millimetre-wave bands continue from 5G. Further out, ITU-R Report M.2541 (May 2024) examined the technical feasibility of IMT in bands above 100 GHz, sometimes called sub-terahertz. Feasibility studies are not allocations: which bands 6G gets is decided by regulators.
Integrated sensing and communication
ISAC would let the network use its own radio signals to sense the world: range, velocity and angle estimation, object detection, localization, imaging and mapping. M.2160 lists navigation, fall detection, vehicle and pedestrian detection and environmental monitoring as typical uses, and gives 1 to 10 cm as a research target for positioning accuracy. ITU-R's draft evaluation guidelines, completed in June 2026, add ISAC test environments for factories and cities, and sensing with communication is already a Release 20 work area for 5G-Advanced.
AI built into the network
5G-Advanced already applies AI and machine learning to the network from Release 18. For 6G, M.2160's "AI and communication" scenario goes further: collecting and processing data, distributed model training and inference across the network, and orchestrating computing resources. TR 22.870 lists AI among the 6G system requirements. For the application side of the same shift, from on-device to cloud inference, see our guide to building intelligent apps.
The difference between 5G and 6G
Read this table as direction, not promise. The 5G column holds ITU's minimum requirements for IMT-2020; the 6G column holds the ranges in M.2160, which the recommendation itself calls estimated targets for research, not necessarily reachable at the same time. The binding 6G numbers will be the minimum technical performance requirements, 20 of them, completed in draft in February 2026 and awaiting approval.
| Aspect | 5G (IMT-2020) | 6G (IMT-2030) |
|---|---|---|
| Status, September 2026 | Deployed; close to 3.3 billion subscriptions | Being specified; first services expected around 2030 |
| First 3GPP release | Release 15 (2017 to 2019) | Release 21 (freezes planned December 2028 to March 2029) |
| ITU framework | M.2083 (2015) | M.2160 (2023) |
| Usage scenarios | eMBB, URLLC, mMTC | Six, including AI and communication and ISAC |
| Peak data rate | 20 Gbit/s downlink (minimum requirement) | 50, 100 or 200 Gbit/s given as examples (research target) |
| Latency over the radio | 1 ms for URLLC, 4 ms for eMBB | 0.1 to 1 ms |
| Reliability | 99.999% | 99.999% to 99.99999% |
| Connection density | 1 million devices per km² | 1 million to 100 million devices per km² |
| Mobility | Up to 500 km/h | 500 to 1,000 km/h |
| Positioning accuracy | Not an IMT-2020 capability | 1 to 10 cm (new capability) |
| Sensing and AI | Not IMT-2020 capabilities | New capabilities, alongside sustainability |
| Spectrum | FR1 and FR2, up to 71 GHz | Existing bands plus candidates around 7 GHz; 10 to 15 GHz and above 100 GHz under study |
The practical difference is compatibility. Ericsson expects 6G to have a core network built on the architectural principles of 5G standalone, but a new radio access network with a new radio interface, so using 6G will take new radio equipment and new devices. How operators migrate from 5G to 6G is part of 3GPP's study work. Standalone 5G, in other words, is also the road toward 6G.
What 5G means for businesses today
6G is years away. The decisions a business can make now are 5G decisions, and three uses are worth understanding.
Fixed wireless access
Fixed wireless access (FWA) delivers broadband to a fixed site, a home, an office or a branch, over the mobile network through a receiver instead of a fibre or cable line. Ericsson calls it a clear way for operators to earn money from 5G. In its June 2026 FWA outlook, 71% of FWA service providers in its study offer the service over 5G, and FWA connections are forecast to grow from 185 million at the end of 2025 to 350 million at the end of 2031, about 17% of all fixed broadband connections, close to 85% of them over 5G.
For a business, FWA is worth pricing for sites where fibre is slow or expensive to reach, for temporary sites, and as a second path beside a wired line. Ask two questions: is the plan speed-based or a monthly data bucket (57% of FWA providers now offer speed-based plans, per the same report), and does the operator offer it on 5G standalone with differentiated connectivity, which some now sell to businesses with service-level commitments?
Private 5G networks
In 3GPP terms a private network is a non-public network (NPN): a 5G system deployed for non-public use, introduced in Release 16 and enhanced in Release 17. TS 23.501 defines two kinds:
| Type | Who runs it | How it is kept separate |
|---|---|---|
| Stand-alone non-public network (SNPN) | An NPN operator, not relying on a public network's functions | Its own network identity (a PLMN ID plus a network identifier, NID) |
| Public network integrated NPN (PNI-NPN) | A public operator, through a dedicated data network or network slice | Slicing, optionally closed access groups that restrict which cells a device may use |
An SNPN gives the most control: the core, the data and the subscriber list stay with you, and so does the work of running them. A PNI-NPN leans on an operator's network and coverage, and devices need a subscription with that operator. For an SNPN, the first question is spectrum, which depends on your country's regulator; for a PNI-NPN, it is what coverage and service levels the operator will commit to.
Edge computing near the network
Latency has a floor set by the speed of light, so the reliable way to cut it is to shorten the path. That is the idea behind multi-access edge computing (MEC), defined by ETSI as cloud computing capabilities at the edge of an access network, close to its users, whatever the access technology. The 5G Core supports it natively: the user plane function can sit at a local site, and traffic reaches it in one of three ways. A distributed anchor sends all of a device's traffic to the local UPF; a session breakout diverts only the edge application's traffic there; or the device uses separate sessions for local and central traffic. In 3GPP's illustrative example, edge computing cuts latency by a factor of 2 to 10, depending on assumptions.
The edge can be an operator's site near the radio network or servers on your own premises. 3GPP's edge work began in Release 15, and Release 17 added normative support for edge computing in the 5G Core and for edge applications. Two details decide whether it works in practice: where the operator actually places the UPF for your traffic, and whether your application can move its session to another edge server when a device moves, which 3GPP calls application context relocation. 3GPP's own examples of applications that need the shorter response time are virtual reality, factories of the future and autonomous driving; for the application side of immersive apps, see our guide to augmented reality programming.
How to plan for 5G now and 6G later
- List the use cases, then the network features they need. Slicing and operator-run private networks rely on the 5G Core, and edge computing is natively supported only there, so confirm standalone coverage at your sites and write the features you need into the contract.
- Pilot FWA at one site against the wired alternative, on a speed-based plan, and measure it at the busiest hour.
- For a private network, settle spectrum and operations first: an SNPN or a PNI-NPN, who holds the spectrum and who runs the core.
- For edge computing, measure the latency budget end to end before paying for proximity: if most of the delay is in the application or its database, a closer server will not fix it.
- Keep workloads portable between a central cloud, an edge site and your premises, using containers and infrastructure as code, so the network decision does not lock in the application.
- Track Release 21, not marketing. The next real 6G milestones are the Release 21 stage 1 freeze in March 2027 and the stage 3 freeze in December 2028.
The radio side of these decisions belongs to your operator; where applications run and how they connect is yours. Our cloud transformation service covers the cloud part of that work: an assessment of your applications and their dependencies, a landing zone with identity, hub-and-spoke networking and private endpoints, and containerized workloads on OpenShift or Kubernetes, moved in rehearsed waves. More on the same themes is in our cloud articles.
