Wi-Fi 8 (IEEE 802.11bn): Everything Known So Far About the Ultra High Reliability Standard

July 14, 2026

Wi-Fi 8 is the marketing name for IEEE 802.11bn, the amendment being developed under the working title Ultra High Reliability (UHR). It is the first Wi-Fi generation in roughly thirty years that does not lead with a bigger headline speed number. That single fact explains almost everything else about it — the feature set, the silicon marketing, and the confusion among buyers who assume every new generation is a throughput upgrade.

This is a reference on what is actually known as of mid-2026: what the standard does, where it sits in the IEEE process, which chips exist, and which claims are still vendor projection rather than ratified specification.

The core idea: reliability instead of peak rate

Every Wi-Fi generation from 802.11b through 802.11be (Wi-Fi 7) chased the same metric — maximum theoretical throughput, pushed upward through wider channels, higher-order modulation, and more spatial streams. Wi-Fi 7 arrived at roughly 46 Gbps of theoretical peak rate using 320 MHz channels, 4096-QAM, and multi-link operation. In practice, almost nobody encounters that number. What people encounter is a video call that stutters when they walk into the kitchen, a smart lock that drops off the network, and an apartment building where forty neighbouring access points are all shouting over each other.

802.11bn targets that gap. Its Project Authorization Request sets goals that are unusually explicit about the worst case rather than the best case:

At least 25% higher throughput in challenging signal conditions (coverage edge, high interference)
25% lower latency at the 95th percentile of the latency distribution
25% fewer dropped packets, particularly during roaming between access points
Reduced power consumption
Improved peer-to-peer (P2P) operation

Notice what is absent: no target for peak PHY rate. Wi-Fi 8 keeps the physical-layer envelope that Wi-Fi 7 established. Same bands (2.4, 5 and 6 GHz), same maximum 320 MHz channel width, same top-end modulation, broadly the same theoretical ceiling. No new spectrum is being unlocked, and no new speed record is being set. The entire gain is in coordination, scheduling, and how the network behaves when conditions are bad.

The feature set

The technical content of 802.11bn splits into PHY-layer additions and MAC-layer coordination mechanisms.

Multi-AP Coordination (MAPC)

This is the marquee feature and the one that most clearly separates Wi-Fi 8 from everything before it. Historically, Wi-Fi channel access has been fully distributed and fundamentally selfish: each access point contends for airtime independently, and a neighbouring AP is treated as noise to be tolerated. Wi-Fi 6 introduced coordination within a single basic service set (BSS) via OFDMA. Wi-Fi 8 introduces coordination between overlapping BSSs.

The task group’s specification framework describes several MAPC modes:

Coordinated Spatial Reuse (Co-SR) — neighbouring APs negotiate transmit power so that two transmissions can occur simultaneously without colliding, rather than one waiting for the other. This is a significant upgrade on the opportunistic, power-capped spatial reuse in 802.11ax.
Coordinated Beamforming (Co-BF) — APs shape their beams to null out energy toward each other’s clients.
Coordinated TDMA — explicit time-slicing of airtime between APs.
Coordinated OFDMA — frequency-domain division of resources across APs.

The beneficiaries are cell-edge clients, which is exactly where reliability collapses today. The obvious deployment contexts are enterprise campuses, factories, shopping malls, stadiums, dense apartment blocks — and consumer mesh systems, where multiple nodes already share a vendor and can coordinate without cross-vendor interop problems.

Seamless roaming

Handoff between access points is one of the ugliest failure modes in current Wi-Fi. The client decides when to move, the decision is late, and the transition drops packets. 802.11bn adds enhanced roaming mechanisms designed to make the move between APs closer to a cellular handover than a disconnect-and-reconnect. The PAR’s 25% packet-drop reduction target is aimed squarely here.

Non-Primary Channel Access (NPCA)

Under current rules, if the primary 20 MHz channel is busy, the device waits — even if a large portion of the wider channel is idle. NPCA allows temporary use of a non-primary 20 MHz channel when the primary is occupied, which recovers airtime that is presently wasted and improves capacity in congested environments.

Dynamic Subband Operation (DSO)

DSO resolves bandwidth-capability mismatches between an AP and a client. An AP can dynamically allocate frequency resources to a station that is operating outside the AP’s current operating bandwidth, rather than forcing everything down to the lowest common denominator.

Distributed Resource Units (DRU)

Regulators impose power spectral density limits, which in practice cap the transmit power a device can apply to a small resource unit — hurting uplink from low-power devices at range. DRU spreads a station’s uplink transmission across non-contiguous tones so the same total power is delivered under the PSD ceiling. The result is more uplink range and reliability for exactly the class of device that suffers most today: sensors, smart-home controllers, wearables.

Enhanced Long Range (ELR)

A new PPDU format aimed at extending reliable range, particularly for uplink and for low-rate IoT traffic. Together with DRU, ELR is why vendors are claiming substantially wider IoT coverage rather than faster laptops.

Unequal Modulation (UEQM)

Allows different modulation and coding to be applied across spatial streams or resource units, so the link can be tuned to actual conditions instead of being pinned to a single uniform scheme.

Where the standard actually stands

Wi-Fi 8 is not finished. The specification is in active draft development, and the drafting has been noisy.

November 2023 — task group work on 802.11bn begins.
July 2025 — Draft 1.0 is completed, defining the technical scope.
Letter Ballot 291 on Draft 1.0 generates over 8,000 comments — an enormous volume, and a fair indication of how much of the design was still contested.
January 2026 — around 740 comments resolved at the Victoria interim; Draft 1.3 approved.
March 2026 — roughly 60% of comments resolved; Draft 1.4 authorised. Draft 2.0 slips from May to July 2026.
May 2026 — approximately 75% of comments resolved at the Antwerp interim. Draft 2.0 ballot expected in July 2026.
Target ratification: 2028. The working group’s stated target has held at roughly May 2028, with IEEE final approval around March–May 2028.

On the certification side, the Wi-Fi Alliance is expected to finalise its certification test plan around mid-2027 and launch Wi-Fi CERTIFIED 8 around the turn of 2027–2028. Certified, interoperable consumer product in volume is therefore a 2028 story, not a 2026 one.

One further data point worth noting: the IEEE 802.11 Working Group has already approved an AI Offload study group, chartered to produce a PAR for offloading compute-intensive AI inference to Wi-Fi access points and other edge devices. The Wireless Next Generation committee is simultaneously reviewing directions for the generation after 802.11bn — Wi-Fi 9 conversations have begun before Wi-Fi 8 has a Draft 2.0. The cadence is accelerating.

The silicon is already here

The gap between “standard ratified in 2028” and “chips announced in 2025” is not a contradiction; it is the normal Wi-Fi product cycle. Wi-Fi 7 silicon shipped years before the IEEE completed 802.11be. The same pattern is running again, faster.

Broadcom moved first, launching the BCM6718 in October 2025 alongside enterprise parts (BCM43840, BCM43820) and a client/edge chip (BCM43109) for phones, laptops and automotive. At CES 2026 it expanded with the BCM4918 — a unified SoC combining compute, networking and AI acceleration — plus two dual-band radios, the BCM6714 and BCM6719. Broadcom has framed the whole family around a telemetry engine that continuously measures network behaviour, and has said retail product could appear as early as summer 2026. It is also offering the platform under licensing terms to accelerate adoption.

MediaTek unveiled the Filogic 8000 family at CES 2026, spanning gateways, enterprise APs and client devices (phones, tablets, TVs, laptops, IoT), and demonstrated multi-AP coordination features — coordinated beamforming, spatial reuse, multi-AP scheduling — on live hardware. First silicon was expected to reach customers during 2026.

Qualcomm announced its Wi-Fi 8 portfolio at MWC Barcelona in March 2026, including FastConnect 8800 — presented as the first mobile chip to combine Wi-Fi 8, Bluetooth 7, UWB and Thread in one package — plus Dragonwing networking platforms for consumer and enterprise routers. Qualcomm has framed the generation explicitly around agentic AI traffic, citing projections that AI-related traffic reaches 15–20% of total internet traffic by 2030.

ASUS showed the ROG NeoCore, a Wi-Fi 8 concept router, at CES 2026 along with what it billed as the first real-world Wi-Fi 8 throughput test, and committed to shipping Wi-Fi 8 home routers and mesh systems in 2026. Its claimed figures versus Wi-Fi 7: up to 2× higher mid-range throughput, 2× wider IoT coverage, and up to 6× lower P99 latency through smarter multi-AP and multi-client operation. TP-Link has also demonstrated working Wi-Fi 8 links and has hardware moving through FCC.

Two cautions on all of the above. First, these are draft-based products: they implement a specification that is still resolving thousands of comments, and they will depend on firmware updates to reach final compliance — with no guarantee that every announced feature survives to ratification. Second, and more important, the headline benefits of Wi-Fi 8 are coordination benefits. Multi-AP coordination and seamless roaming require both ends, and ideally multiple APs, to speak the same protocol. A Wi-Fi 8 router talking to a house full of Wi-Fi 6 phones and Wi-Fi 5 IoT devices delivers close to nothing that a Wi-Fi 7 router would not. First-generation Wi-Fi 8 hardware is, functionally, a Wi-Fi 7 router with a forward-looking chipset.

What Wi-Fi 8 is not

It is not faster. Peak theoretical rates are essentially unchanged from Wi-Fi 7. Any “2× throughput” claim refers to mid-range or cell-edge conditions, not the number on a speed test standing next to the router.
It does not add spectrum. No new bands. The 6 GHz band remains the newest frontier, and its regulatory status still varies by jurisdiction.
It is not backward-compatible in its headline features. Legacy clients cannot benefit from MAPC, DRU or ELR.
It is not certified. Nothing sold in 2026 or 2027 carries Wi-Fi CERTIFIED 8.

Practical guidance

For anyone specifying networks in 2026: Wi-Fi 7 is the mature standard for this cycle. It is ratified, certified, widely available, and its client ecosystem is filling in. Deferring a network refresh to wait for Wi-Fi 8 means waiting through a draft-hardware period into a 2028 certification window, and then waiting again for client devices to catch up. Anyone still on Wi-Fi 5 or older should upgrade now, to Wi-Fi 7.

The exceptions are the environments Wi-Fi 8 was actually designed for — very high AP density, heavy roaming, deterministic latency requirements, large fleets of low-power uplink devices. Those operators should be tracking 802.11bn closely, budgeting for a 2028–2029 refresh, and treating draft-hardware pilots as evaluation exercises rather than production deployments.

Glossary

802.11bn — the IEEE amendment underlying Wi-Fi 8. Task group formed 2023; ratification targeted for 2028.

UHR (Ultra High Reliability) — the working name for the 802.11bn project, reflecting its focus on consistency rather than peak rate.

MAPC (Multi-AP Coordination) — framework allowing neighbouring access points in overlapping BSSs to coordinate transmissions rather than contend blindly.

Co-SR (Coordinated Spatial Reuse) — MAPC mode in which APs negotiate transmit power to allow simultaneous transmissions.

Co-BF (Coordinated Beamforming) — MAPC mode in which APs shape beams to reduce interference at each other’s clients.

NPCA (Non-Primary Channel Access) — permits use of a non-primary 20 MHz channel when the primary is busy.

DSO (Dynamic Subband Operation) — allows an AP to allocate frequency resources to a station operating outside the AP’s current bandwidth.

DRU (Distributed Resource Unit) — spreads uplink transmission across non-contiguous tones to work around power spectral density limits, improving uplink range.

ELR (Enhanced Long Range) — new PPDU format extending reliable range, aimed at low-power and IoT uplink.

UEQM (Unequal Modulation) — permits different modulation and coding across spatial streams or resource units.

P99 latency — the latency experienced by the worst 1% of packets. The metric that determines whether a call stutters, and the one Wi-Fi 8 is built to improve.

The short version

Wi-Fi 8 is a maturity release. It accepts that raw bandwidth stopped being the binding constraint somewhere around Wi-Fi 6, and that the real problems are density, interference, roaming, and the tail of the latency distribution. The engineering answer is coordination — access points that cooperate instead of compete — plus PHY work aimed at the weakest links on the network rather than the strongest.

The specification is roughly three-quarters through comment resolution on its first complete draft, with Draft 2.0 balloting in July 2026 and ratification targeted for 2028. The silicon arrived early, as it always does, and the marketing arrived earlier still. The technology is real, the timeline is not.