Every Wi-Fi generation for the last thirty years has been a speed story: 54 Mbps, then 600, then a gigabit, then “multi-gigabit.” You bought the new number, ran a speed test, felt good about it, and then went back to complaining about the exact same things: the dead zone in the back room, the video call that glitches when you walk downstairs, the game that stutters at 9 PM when the whole street is streaming.
Wi-Fi 8 is the first generation that looked at that list of complaints and said fine, let’s fix those instead. The standard is technically IEEE 802.11bn, and inside the IEEE the project goes by a name that tells you everything: Ultra High Reliability. Peak speed? Unchanged from Wi-Fi 7. Nobody’s promising a bigger number this time.
The target isn’t the peak — it’s the 95th percentile
The 802.11bn task group’s goals are all about the moments you actually notice, measured where Wi-Fi 7 is weakest: roughly 25% more throughput in poor signal conditions, 25% lower latency at the 95th percentile (that is, the worst 1-in-20 packets — the ones that cause the stutter), and 25% fewer dropped packets, including during roaming between access points. If Wi-Fi 7 was about how fast the freeway goes at 3 AM, Wi-Fi 8 is about rush hour.
The mechanisms behind it are genuinely new, not just “more of everything”:
- Multi-AP coordination — the big one. Today’s access points mostly act like rival radio stations that happen to share a building, politely (or not) staying out of each other’s way. 802.11bn lets neighboring APs cooperate: coordinating transmissions, scheduling airtime, even jointly serving a client. For anyone running multiple APs or a mesh system, this is the standard finally doing in-spec what vendors have been faking with proprietary tricks.
- Non-Primary Channel Access (NPCA) — if the primary channel is jammed, a device can temporarily borrow a free secondary channel instead of sitting there waiting. Less queueing, steadier latency.
- Dynamic Sub-channel Operation (DSO) — finer-grained, more flexible use of the available spectrum.
- Seamless roaming — fewer lost frames when your phone hands off between APs as you walk around the house.
Where the standard actually stands
This is still a draft standard, and it’s worth being precise about that, because products are already shipping. The project was approved in 2023. Draft 1.0 went to letter ballot in mid-2025; Draft 2.0 went to ballot in July 2026 and passed in September with about 77% approval, after the working group chewed through thousands of comments (which produced Draft 2.1). The projected path from here: Draft 3.0 ballot in January 2027, the wider IEEE-SA ballot in May 2027, and final IEEE approval around May 2028 — though different progress reports have cited dates anywhere from Q1 to Q3 of 2028, so treat any specific month as an estimate. One caution from the timeline so far: Draft 1.0 and Draft 2.0 each landed about two months late. Projections slip. (Curecomp’s deep-dive has the most thorough draft-by-draft timeline I’ve found; this standards timeline lays out the milestones in table form.)
Wi-Fi Alliance certification — the logo that means gear from different vendors actually interoperates — is expected around late 2027 to early 2028, roughly ahead of final IEEE sign-off, which is the normal order of operations. (Peaks and Valleys has a clear month-by-month rundown.)
The draft-hardware era is already here
And this is the part that should make your wallet nervous. Chipmakers are building to the draft well ahead of certification: Broadcom has a full Wi-Fi 8 silicon lineup out, MediaTek unveiled its Filogic 8000 family at CES 2026, and Qualcomm announced the FastConnect 8800 mobile chip at MWC 2026 (notably bundling Wi-Fi 8 with Bluetooth 7, Ultra Wideband, and Thread in one package). (Pandromeda’s explainer covers the chip race in detail.)
Routers are following. ASUS has been selling the ROG Rapture GT-BN98 as a draft-802.11bn gaming router since autumn 2026, and at IFA 2026 TP-Link announced the Archer 8 Ultra and Deco 8 Ultra, opening pre-orders on September 30 in select regions — the UK, Germany, Australia, Canada, notably not the US yet. There are no client devices yet: no phone, tablet, or laptop on sale today supports Wi-Fi 8, with the first expected late 2026 into 2027.
Draft-based hardware isn’t automatically a trap — Wi-Fi 7 chips followed exactly this path in 2023–24, shipping on drafts and converging via firmware updates once the spec locked. But “some features arrive later by update,” which is TP-Link’s actual wording for parts of the Archer 8 Ultra, is a promise, not a product. (This September field guide for wireless engineers is the most honest thing I’ve read about buying in this awkward gap.)
The homelab verdict: don’t wait, don’t pay the draft premium
Here’s my honest take, as someone who enjoys tinkering with this stuff as much as anyone:
If your Wi-Fi works, do nothing. Wi-Fi 8’s payoff doesn’t arrive until certified gear is on shelves around 2028, and then until your whole fleet — router, APs, phones, laptops — speaks it. A draft-spec router talking to Wi-Fi 6 phones gets you exactly nothing.
If you need gear now, buy Wi-Fi 7. It’s a finished, certified standard, it’s what the new M6 Macs speak (see last week’s Apple coverage), and you get a certification logo instead of a draft disclaimer. Wi-Fi 8 being real in 2028 doesn’t make Wi-Fi 7 obsolete any more than Wi-Fi 6 made 5 obsolete.
The feature to actually watch is multi-AP coordination. When it matures, it’s the first time the standard itself — not a vendor’s proprietary mesh layer — coordinates your access points. For homelab and UniFi-style multi-AP setups, that’s the part of Wi-Fi 8 that could change how you design a network, not just how fast it benchmarks.
The speed race is over, and honestly, good riddance. The next race is about the network disappearing — no dead zones, no glitches, no thinking about it. That was always the actual goal.