Table of Contents
For a deep dive into 6G, register for the upcoming 6G Forum virtual event.
With IMT-2030 requirements defined and 3GPP resolving foundational design choices, the industry is moving from 6G vision to specifying an implementable system
The 6G conversation is moving from vision-setting into engineering. After years of research projects, vendor white papers and debates about what the next generation should accomplish, the standards process is beginning to put hard boundaries around what 6G will actually be.
At the ITU, the IMT-2030 framework now includes 20 minimum technical performance requirements spanning six usage scenarios: immersive communication, hyper-reliable low-latency communication, massive communication, ubiquitous connectivity, AI and communication, and integrated sensing and communication. In June, ITU-R Working Party 5D also completed draft evaluation guidelines that define how candidate 6G radio technologies will be tested, including new factory and urban-macro environments for sensing. Candidate radio-interface submissions are expected between February 2027 and February 2029.
In parallel, 3GPP is turning those high-level requirements into an implementable system. Release 20 is the study phase for foundational 6G technologies; Release 21 will contain the first normative 6G specifications. Based on the timeline agreed in June 2026, technical studies should conclude across 3GPP groups during the first half of 2027, with specification work continuing through late 2028 and final review extending into early 2029. That puts first commercial systems broadly in the 2029-2030 window; the 2028 Summer Olympics in Los Angeles, set for July 14-30, will likely serve as a first-look at 6G trials.
More interesting than the dates, however, is what 3GPP has already decided. The emerging 6G radio is looking less like a clean-sheet break from 5G than early generational hype sometimes implied. 3GPP has selected CP-OFDM for the downlink, with CP-OFDM and DFT-s-OFDM supported in the uplink. Much of 5G’s channel coding will be reused. The basic radio-frame structure will remain similar to 5G, in part to make dynamic spectrum sharing between 5G and 6G practical. Supported system bandwidths are expected to range from 3 megahertz to as much as 400 megahertz depending on spectrum, while massive IoT is being designed into 6G from the beginning rather than inherited from an earlier generation.
The architecture is simultaneously being simplified. The baseline design assumes standalone 6G rather than repeating 5G’s initial non-standalone model, which tied 5G radio to the 4G core. 3GPP is also retaining flexibility around centralized and distributed RAN implementations: a higher-layer CU/DU split is supported, while a potential lower-layer multi-vendor interface between baseband processing and the radio unit would be defined in detail by the O-RAN Alliance. The goal is not to discard the cloudification and disaggregation work underway in 5G, but to carry it forward without making 6G dependent on a legacy radio architecture.
Where 6G does represent a more meaningful expansion is in what the network is expected to do. ITU’s inclusion of AI and communication and integrated sensing and communication (ISAC) as explicit usage scenarios is significant. Qualcomm now describes 6G as an AI-native platform combining advanced connectivity, distributed compute and wide-area sensing. Nokia similarly argues that AI-native design is becoming the industry direction, with AI used both to operate the network and supported as a workload across devices, edge infrastructure and cloud.
Capabilities including AI/ML, NTN, energy efficiency and sensing have been progressively developed through the later stages of 5G and 5G-Advanced. With 6G, the industry is attempting to treat intelligence, sensing, energy efficiency, resilience and a broader relationship between connectivity and compute as design considerations from the outset. Qualcomm, for example, expects terrestrial and non-terrestrial networking to be integrated from day one, while both Qualcomm and Nokia increasingly position distributed AI workloads as part of the 6G system proposition.
For operators, that evolutionary character may prove as important as the new capabilities. A commercially viable generational transition depends on preserving infrastructure and spectrum investments where possible while creating enough improvement in capacity, uplink performance, energy efficiency and automation to justify new capital.
None of that means the 6G standard is settled. Release 20 is explicitly a study phase, and major questions around spectrum, AI implementation, sensing architecture and commercialization remain open. The takeaway is that the industry is no longer primarily discussing what 6G might be; standards bodies are increasingly deciding what 6G will do.
6G needs substantially more spectrum, but its commercial viability will depend on finding frequencies that deliver wider channels without forcing operators to abandon the economics of the existing macro network
Standards will define what 6G can do, but spectrum will determine where, and economically, whether it can meet the definition. The capacity requirements alone are significant. The GSMA estimates that mobile networks in dense urban areas will require an average of 2–2.5 gigahertz of mid-band spectrum between 2035 and 2040, rising to 2.4–3.3 gigahertz in higher-demand countries. With roughly 1 gigahertz of mid-band spectrum identified for mobile in many markets today, that leaves an additional requirement on the order of 1–2 gigahertz.
That demand is pushing 6G into higher frequencies, but not necessarily the extreme bands that dominated some early visions of the technology. Instead, a broad industry consensus is forming around the upper mid-band or centimeter-wave spectrum as a new wide-area capacity layer.
The World Radiocommunication Conference in 2027 will consider several ranges for potential IMT identification, including 4.4–4.8 GHz, 7.125–8.4 GHz and 14.8–15.35 GHz. The upper 6 GHz band is also increasingly important. The GSMA argues that 200–400 megahertz channels will be necessary for 6G and says 6.425–7.125 gigahertzz already has a harmonized mobile footprint covering more than 80% of the global population.
Setting aside varying regulatory machinations, this is fairly straightforward. Existing low- and mid-band frequencies have favorable propagation characteristics but relatively little unused contiguous bandwidth. Millimeter-wave and eventually (maybe) sub-terahertz frequencies offer enormous bandwidth, but their propagation characteristics make them difficult and expensive to use for ubiquitous wide-area coverage.
Spectrum roughly between 6 GHz and 15 GHz offers a potential middle ground with enough contiguous bandwidth to create very wide carriers, but frequencies low enough that advanced antenna systems could compensate for much of the additional propagation loss. That last point is critical to the business case.
Qualcomm is developing what it calls Giga-MIMO for the 6–8 GHz range. Because wavelengths become shorter as frequency increases, more antenna elements can be packed into an array of similar physical dimensions. Denser arrays enable narrower beams, greater effective radiated power and more spatial multiplexing. Qualcomm’s system-level evaluations indicate an upper-mid-band Giga-MIMO layer could support approximately five times greater network load and three times higher average user throughput while providing coverage comparable to lower mid-band. Those are vendor research results rather than commercial-network measurements, but the intended outcome is to add capacity without adding another layer of cell sites.
Nokia is pursuing the same fundamental objective with extreme massive MIMO, and explicitly identifies 7–15 GHz spectrum as important to cost-effective 6G coverage and capacity. Ericsson likewise expects centimeter-wave spectrum, particularly the lower portion of 7–15 GHz, to support wide-area deployments when combined with massive MIMO and beamforming. That said, the spectrum challenge is not limited to finding new bands. 6G will also have to coexist with 5G for years.
Ericsson is working on multi-RAT spectrum sharing, or MRSS, to allow 5G and 6G to dynamically use the same spectrum across both FDD and TDD bands. The company expects long device lifecycles and limited spectrum below 7 GHz to require efficient 5G/6G sharing for at least a decade. The emerging 6G radio design is being shaped in part around that requirement.
In the U.S., another form of sharing and coexistence hinges on gaining access to spectrum occupied by federal systems. NTIA is examining portions of 2.69–2.9 GHz, 4.4–4.94 GHz and 7.125–7.4 GHz for potential commercial 6G use. Qualcomm argues that at least 600 megahertz of full-power mid-band spectrum should be ready by 2029, emphasizing “full power” because heavily constrained access can diminish the coverage and deployment economics that make mid-band valuable in the first place. That makes spectrum-sharing technology, incumbent relocation and coexistence mechanisms part of the 6G architecture—not peripheral regulatory questions.
Although millimeter wave and potentially sub-terahertz frequencies have a role to play, the defining 6G spectrum question is how much contiguous bandwidth operators can obtain while preserving enough coverage, transmit power and infrastructure reuse to make deployment economically rational. To reiterate: 6G will absolutely require new spectrum, but successful deployment means it cannot require a new network grid.
The killer app for 6G won’t be an app at all
The search for a 6G “killer app” is the wrong approach to understanding the potential future impact of next-generation cellular. Previous mobile generations were often sold around a defining consumer experience: mobile broadband for 4G, then ultra-fast connectivity and new enterprise applications for 5G. The emerging 6G vision is broader. Rather than enabling one breakthrough application, the network itself is being designed to provide new capabilities — AI, sensing, distributed compute and ubiquitous connectivity, among them — that can be assembled into many services.
That goal is already visible in the formal requirements. ITU’s IMT-2030 framework adds AI and Communication and Integrated Sensing and Communication (ISAC) to more familiar scenarios covering immersive communications, massive connectivity and ultra-reliable communications. The associated technical requirements include AI integration, high-precision positioning and sensing functions such as object detection, localization, imaging and mapping.
The distinction with 6G is that AI changes both what networks carry and what applications need from them. Today’s networks were built primarily around predictable, downlink-heavy human traffic. Nokia argues that emerging AI applications will create traffic that is more interactive, bursty and uplink-intensive as devices, cameras, robots and sensors continuously generate context. Qualcomm makes a similar case, identifying uplink performance as a first-order 6G requirement alongside coverage, spectral efficiency and energy efficiency.
But the larger change is architectural. Qualcomm describes 6G around three converging pillars: connectivity, wide-area sensing and distributed compute. Ericsson talks about an intelligent fabric offering connectivity, spatial information, compute and data services. Nokia frames the same transition as the network becoming a “distributed nervous system” able to move information, execute inference and generate environmental awareness.
ISAC, illustrates what “beyond connectivity” can mean. By analyzing radio signals transmitted through the network, infrastructure can detect and locate objects, map environments and monitor movement even when the object itself is not connected. Ericsson is already studying architectures in which base stations and user devices contribute sensing measurements, while higher-level processing and service exposure sit in the core.
Potential applications range from traffic and drone detection to industrial safety, digital twins and robotic coordination. But these remain at different stages of maturity. A sensing architecture defined in standards, a prototype object-detection demonstration and a customer paying for a commercial spatial-information service are three very different things.
Distributed compute presents a similar opportunity. AI inference does not always belong in a centralized data center; factors such as latency, privacy, reliability and device-power constraints can make edge execution preferable. Qualcomm envisions workloads dynamically partitioned across devices, network-edge infrastructure and centralized cloud resources. Ericsson argues that operators could combine localized AI hosting with differentiated connectivity, security and real-time network information to participate more directly in the AI value chain.
The precursor infrastructure is already being tested in 5G. T-Mobile US, NVIDIA and Nokia are piloting physical AI applications on distributed AI-RAN infrastructure, including vision AI agents running at the edge while the same infrastructure supports mobile connectivity. T-Mobile US has been explicit that these experiments are foundational rather than finished 6G systems. Their importance is that they test whether telecom infrastructure can simultaneously function as a radio network and a distributed computing platform.
This idea of a distributed, programmable platform will ultimately matter more than any single 6G application. Ericsson envisions positioning, sensing, timing, compute and network insights becoming consumable services exposed through APIs. In that model, a developer does not necessarily buy “6G.” It invokes a network capability like precise location, guaranteed latency, edge inference or spatial data, for instance, to improve an application.
This is where the use case discussion meets the commercialization problem. Vendors can increasingly show how 6G might support physical AI, autonomous machines, immersive devices and digital twins. The harder question is whether operators can package those capabilities into products for which enterprises and developers will pay. To understand that possible misalignment, consider the past 10-plus years of mobile edge computing as a new line of business.
The 6G killer app, in other words, won’t be an app at all. It will be a programmable network that can sense, compute and communicate, and a commercial model that lets operators monetize those capabilities.
Operators will scale 6G deployment when it lowers the cost of carrying more traffic, creates new revenue, or both
The telecommunications industry can increasingly put a date on 6G, but it has yet to put a date on the business case. 3GPP’s standards timeline points toward implementable specifications in early 2029. Qualcomm and a broad coalition of operators and vendors are targeting initial commercial systems from 2029 onward, while Ericsson expects initial deployments around 2030. But technical availability does not determine when an operator replaces radios, buys spectrum or commits billions of dollars to another network investment cycle. Economics does.
That leaves 6G with two distinct business cases. The first is relatively straightforward and historically grounded: carry more traffic at lower unit cost. The second is more ambitious and has been a mainstay of the 5G cycle: turn capabilities beyond connectivity into new revenue.
The efficiency case may prove the more important one at launch. Mobile data demand will continue growing regardless of whether operators deploy 6G, while AI applications are expected to create more uplink-intensive and variable traffic. Qualcomm has made total cost of ownership a specific 6G design consideration, including improvements in spectral and energy efficiency, network simplification, automation and infrastructure reuse. The company argues that the 4G-to-5G transition did not produce the same decline in cost per bit achieved from 3G to 4G, an increasingly important problem when consumer subscription revenue is relatively flat.
Ericsson makes a similar argument. Its current 6G architecture emphasizes leaner system design, autonomous operation and reuse of existing 5G Standalone and core-network investments. Initial 6G deployments are expected to build on rather than replace those foundations, while multi-RAT spectrum sharing allows operators to migrate spectrum gradually between generations.
The investment question is about whether 6G creates an entirely new consumer experience and more about marginal economics. What the industry has to figure out is when does adding 6G capacity become cheaper than continuing to densify and optimize 5G?
Work underway before 6G could shift that calculation further. Nokia’s commercial AI-RAN platform has already demonstrated more than 20% spectral-efficiency gains through AI-driven radio techniques, according to the company, with substantially larger gains targeted through 2028. Those are Nokia’s performance claims rather than independent measurements, but the trajectory illustrates that operators are already introducing the AI-native, software-driven technologies expected to underpin 6G while extracting additional capacity from existing assets.
That creates a potentially smoother investment path. Cloud RAN, AI-RAN, 5G Standalone, network automation, Open RAN interfaces and programmable cores can all become the architectural foundation for it.
The second business case is revenue growth. Ericsson envisions operators exposing differentiated connectivity, sensing, positioning, compute, data and AI capabilities through programmable platforms and APIs. Nokia similarly positions “Network for AI” as an opportunity for 6G infrastructure to support distributed intelligence and new services. Qualcomm’s 6G architecture combines connectivity, sensing and compute with the explicit goal of enabling new service and revenue models.
The opportunity is credible, but the revenue pool is less certain. Operators have spent much of the 5G era trying to monetize capabilities beyond best-effort connectivity. Network slicing, edge computing, private networks and APIs have produced real commercial deployments, but not yet a new revenue engine comparable to mobile broadband itself. The GSMA’s current network agenda reflects that unfinished work: complete 5G Standalone, scale differentiated services and Open Gateway APIs, apply AI to improve network economics, and carry those commercial models forward into 6G.
An operator that enters 2030 with a mature standalone core, automated operations, programmable network exposure and AI-native RAN has already absorbed much of the architectural transition. One that has not will face a very different 6G investment calculation.
If history is instructive, there won’t be a single 6G launch moment. Operators with new spectrum, acute capacity requirements and mature network architectures may move quickly. Others may selectively introduce 6G where its spectral or energy efficiency produces a clear return. Operators still extracting value from relatively young 5G assets can wait.
That may be the most important lesson the industry can carry forward from 5G. A new generation does not need to be justified by a single revolutionary application. It needs to improve the economics of the network operators already have while creating credible options for growth. Standards will determine when 6G can be deployed. The business case will determine when it actually is.
