Cisco Wi-Fi 6 Access Points Are End-of-Sale — and the Replacement Changes Your Switch

Written and maintained by Haink's network infrastructure team · Updated August 2026 · authorized-channel, serial-verified

On 30 March 2026 Cisco announced end-of-sale and end-of-life for the Cisco and Meraki Wi-Fi 6 indoor access points — the whole Catalyst 9100AX indoor family, from the entry-level C9105AX to the C9130AX. Support runs to 31 December 2031, so nothing stops working, and there is no urgency in the sense of an outage. The urgency is procurement: after the last ship date you buy these on the secondary market or not at all, and Cisco's own migration path points at Wi-Fi 7 hardware that does not drop into the same switch port.

That last point is what most migration write-ups skip. Swapping a C9120AXI for its recommended replacement is not a like-for-like exchange. The access point changes, and so do the power class of the switch port, the power supply sizing of the access switch, the IOS-XE train on the controller, and — if you want the full uplink speed — the cabling. This page works through all four, with the numbers.

The dates

The gap between end of software maintenance (2027) and last day of support (2031) matters more than it looks. From January 2028 you keep TAC entitlement but stop receiving software fixes — including security fixes — for these access points. For a campus estate that is a four-year window of hardware you can run but cannot patch.

What replaces what

Cisco publishes a direct mapping in the bulletin. It is not a single replacement family — the Wi-Fi 7 line splits by role, and two of the old SKUs land on different products depending on whether they were the internal- or external-antenna variant.

Worth noting: the C9120AXI is the volume SKU in most campus estates, and it is the one that maps to the CW9176I — the highest-power access point in the mainstream part of the range. If your estate is mostly C9120AXI, the power section below is the section that will decide your budget.

The part that costs money: PoE class

Every Wi-Fi 6 access point in this bulletin runs at full specification on 802.3at (PoE+, 30 W at the port). Cisco's own power reference puts maximum draw at 21.4 W for the C9115AX, 25.5 W for the C9120AX and 30.5 W for the C9130AX. A standard PoE+ access switch — a C9300-48P, say — powers a full complement of them without anyone thinking about it.

The Wi-Fi 7 replacements do not behave that way. They will come up on PoE+, but they come up degraded, and what you lose is exactly what you paid for.

Read the CW9176I row again. On a PoE+ port you get a 2.5G uplink and a 2×2 radio on 2.4 GHz. On 802.3bt you get a 10G uplink and 4×4 everywhere. That is the difference between a Wi-Fi 7 access point and a Wi-Fi 7 access point you paid for and then throttled at the switch.

What your access switch has to be

Cisco's Catalyst 9300 data sheet is explicit about which models deliver what. UPOE at 60 W comes from the C9300-24U, 48U, 24UX, 48UXM, 48UN, 24UB, 48UB and 24UXB, plus the C9300L UXG models. Standards-based IEEE 802.3bt Type 3 and Type 4 — UPOE+ at 90 W — comes only from the C9300-24H, C9300-48H, C9300X-24HX and C9300X-48HX.

If your access layer is C9300-48P — the PoE+ SKU, and by far the most common one in the installed base — then a straight C9120AXI → CW9176I swap silently downgrades every new access point. Nothing fails, nothing alarms. You just never get the throughput you bought. This is the single most expensive mistake available in this migration, and it is invisible unless someone checks the port power class.

The power budget arithmetic

Per-port class is only half of it. The second half is whether the switch can supply that power across all its ports at once. Catalyst 9300 power supplies come in 350 W, 715 W, 1100 W and 1900 W; on a C9300-48H, dual supplies yield 1,740 W (2×350), 2,105 W (2×715), 2,490 W (2×1100) or 3,290 W (2×1900) of available PoE.

Run the numbers for a fully populated 48-port access switch:

Two conclusions fall straight out of that table. First, a like-for-like AP count on CW9176I adds roughly 650 W per 48-port switch — enough to push a stack that was comfortable on 350 W supplies into 715 W territory, and enough to matter in a rack power budget across a campus. Second, you cannot fully populate a 48-port switch with CW9178I at full specification: at 3,290 W of available PoE the ceiling is 34 access points. High-density Wi-Fi 7 changes the port-count-per-switch arithmetic that campus designs have used for a decade.

The software constraint nobody plans for

Each Wi-Fi 7 access point carries its own minimum IOS-XE release on the Catalyst 9800 controller, and they are not the same release.

This has a practical consequence that catches people out. If you standardise on the CW9174 — and four of the nine migration paths above point at it — you are committing the whole wireless estate to IOS-XE 17.18.2 or later, a considerably newer train than the 17.15 most production controllers are running. A mixed refresh that puts CW9176 in one building and CW9174 in another means the controller has to sit on the newer of the two minimums, for every site it serves.

Order of operations, therefore: qualify the controller release before the access points land, not after. On the CW9178I there is a further detail — Cisco's deployment guide notes that multi-link operation and the full spatial-stream configuration on 802.3at depend on 17.15.3 or later.

Cabling

The CW9176 and CW9178 uplinks are 100M/1G/2.5G/5G/10G multigigabit. Reaching 10 Gbps requires Cat6 or Cat6A; Cat5e tops out at 5 Gbps. Most campus estates cabled for Wi-Fi 5 or early Wi-Fi 6 are Cat5e or Cat6 to the ceiling. That does not block the deployment, but it does cap it, and it is worth auditing before the switch decision rather than after — if the cabling limits you to 5 Gbps anyway, the case for UPOE+ switching in that building weakens considerably.

What to do, in order

1. Inventory by SKU, not by family. The C9120AXE, C9120AXP and C9120AXI are already past end-of-sale; everything else has until 31 December 2026. You need the split before you can plan.
2. Decide whether to buy Wi-Fi 6 now or Wi-Fi 7 later. Where a site is on a five-year hardware cycle that has three years left, extending it with remaining Wi-Fi 6 stock is often cheaper than pulling the switch layer forward. Support runs to 2031; software maintenance ends 2027.
3. Check the port power class on the access switches that serve every AP you intend to replace. This determines whether the migration is an AP refresh or an AP-plus-switch refresh, and it is the difference between two budget lines.
4. Recompute the PoE budget per switch at the new draw, and check the power supplies you actually have installed — not the ones the model supports.
5. Qualify the controller release against the specific Wi-Fi 7 models you have chosen, using the table above.
6. Then order — and order the switch-side hardware on the same schedule as the access points, because it has its own lead time.

What Haink supplies, and what is actually available

We supply both sides of this migration: remaining Cisco Wi-Fi 6 inventory where extending the existing estate is the right call, and the Wi-Fi 7 line — CW9171, CW9172, CW9174, CW9176, CW9178 — together with the Catalyst 9800 controllers and the UPOE and UPOE+ access switches the new access points require. All units are sourced through the authorized channel and serial-verified against Cisco before payment; see gray-market and channel risk for why that check matters more on end-of-sale hardware than on current product.

Two practical notes on availability. Cisco moved the C9120AX dates up because it ran out of materials, which means vendor-side stock of those SKUs is gone — remaining supply sits with distribution and in channel inventory, and it is finite. And on the Wi-Fi 7 side, the switch-layer hardware is frequently the longer lead time, not the access points. Our current Cisco stock and lead-time position is on the Cisco supply page; how to verify an in-stock claim covers what to ask any supplier for before you commit to a delivery date.

For estates in restricted or licence-controlled destinations, the access points and controllers are subject to the same screening as the rest of the portfolio — see export compliance and dual-use.

Frequently asked questions

Can I keep running Cisco Wi-Fi 6 access points after end-of-sale?

Yes. Last day of support is 31 December 2031, so TAC entitlement and RMA continue for holders of a valid service contract. The constraint that arrives sooner is software: end of software maintenance is 31 December 2027, after which these access points stop receiving fixes, including security fixes.

Which Wi-Fi 6 access points can I still buy?

Everything except the C9120AXE, C9120AXP and C9120AXI, whose end-of-sale dates were moved up to July 2026. The rest of the affected line can be ordered until 31 December 2026, with a last ship date of 31 March 2027 — subject to whatever inventory remains in the channel, which for a product Cisco discontinued for lack of materials is not guaranteed.

Will a CW9176I work on my existing PoE+ switch?

It will come up and pass traffic, but at reduced capability: a 2.5G uplink instead of 10G, 2×2 instead of 4×4 on 2.4 GHz, and USB disabled. Full specification requires 802.3bt. If the access layer is a C9300-48P or similar PoE+ switch, plan the switch refresh alongside the access point refresh.

Do I need to upgrade my Catalyst 9800 controller software?

Almost certainly. The minimum is 17.15.2 for the CW9176 and CW9178, but 17.18.2 for the CW9171 and CW9174 and 17.17.1 for the CW9172H. Choose the access point models first, then take the newest minimum across them as the controller target.

How many Wi-Fi 7 access points fit on one 48-port switch?

For the CW9176I at 39 W, all 48 — but the switch needs roughly 1,872 W of available PoE, which means dual 715 W supplies or better. For the CW9178I at approximately 95 W, the ceiling is 34 access points even with dual 1900 W supplies. Port count is no longer the binding constraint; power is.

Is Wi-Fi 7 worth it if my cabling is Cat5e?

Partly. Cat5e caps the multigigabit uplink at 5 Gbps, so the 10G capability of the CW9176 and CW9178 goes unused. The airtime and client-density improvements still apply, but the case for the higher-end access points and for UPOE+ switching in that building is weaker. Where recabling is not on the table, the CW9172 and CW9174 are usually the better economic fit.

Should I move to Meraki instead of Catalyst while I am replacing the hardware?

It is a reasonable moment to ask, because the CW917x access points operate in either mode — managed by a Catalyst 9800 controller or by the Meraki cloud platform. The decision turns on licensing cost over five years and on whether you need the feature depth of the controller-based path. We cover this on the Cisco selection guide.

Related

• How to choose Cisco switches, routers and firewalls — the full selection guide
• Cisco end-of-life guide — every family currently in migration, with the binding dates
• Enterprise wireless access points — what we supply across Cisco, Aruba and Huawei
• Enterprise switches — access, core and data center
• Network infrastructure supplier — how we scope and deliver campus projects
• HPE Aruba buying guide — the main alternative if the switch refresh changes the economics
• Gulf fintech network — 30-site Catalyst 9300/9500 and FortiGate rollout in six weeks
• Research institution network — 52 Cisco switches, 2,400 ports, eight buildings
• Dubai supply hub · Hong Kong supply hub

Sources

• Cisco — End-of-Sale and End-of-Life Announcement for the Cisco and Meraki Wi-Fi 6 Indoor Access Points
• Cisco — Aironet and Catalyst AP power requirements reference
• Cisco — CW9176 deployment guide · CW9178I · CW9174 · CW9172 and CW9171
• Cisco — Catalyst 9300 Series data sheet (PoE budgets, UPOE and UPOE+ model list)