802.11ax Access Points: What AX1500, AX1800 and AX3000 Change on a Floor

What does the AX number on an access point actually measure?

It measures the combined physical-layer rate of both radios, each at the widest channel it supports, added into one figure. It is not the speed a laptop will see, and it is not a per-user guarantee. The rest of the datasheet — antenna chains, port speed, consumption — decides what that aggregate becomes on a floor.

A single client attaches to one band at a time, so its ceiling is that band alone. After protocol overhead and half-duplex operation, a healthy client test usually lands at roughly 50 to 65 percent of that band published rate, which is an engineering convention rather than a product specification. The gap between 1500 and 3000 therefore shows up first as capacity for many clients, and only second as speed for one.

Where do the 1500, 1800 and 3000 figures come from?

Each figure is a tier label that tracks how many radio chains and how wide a channel the chipset runs. On our ceiling line the published steps are 1500Mbps, 1800Mbps and 3000Mbps, plus 3600Mbps on the 802.11be model. A step up often moves antenna chains and port speed along with the headline rate, so read the whole row rather than the first column.

Which step actually changes what people feel?

Antenna chains first, then the uplink port, then placement. A two-chain 5GHz radio serves one 2x2 laptop about as well as a four-chain radio does; the fourth chain earns its keep when a room transmits at once, or when most of the room is 1x1 handsets that need scheduling.

The uplink is the harder ceiling. A gigabit port carries roughly 940Mbps in both directions combined, so a 3000Mbps radio behind it is capped by the port at the busiest moment of the day. That is why CF-E391AX publishes a 2.5G WAN alongside a 1G LAN: the wired side stops being the bottleneck instead of the radio number being pushed higher.

Does a higher AX number also raise the PoE bill?

It can, and the published consumption answers it before the order. Units rated under 12W or under 13W sit inside the 802.3af class, where the switch port supplies 15.4W; the under-24W class only fits 802.3at, where the port supplies 30W. Eight units under 12W come to under 96W, which a switch with a 150W budget still serves with room to spare.

The trap in the ladder is that CF-E393AX and CF-E395AX publish the same 3000Mbps as CF-E593AX while consuming under 24W against under 12W — the same rate for twice the budget. Count the 24W class port by port and hold 20 percent headroom at the switch; that is the only safe way to size this line.

What should you ask for when the datasheet runs out?

Ask for the three things a product page usually leaves out: the per-band split and channel widths behind the aggregate, the MIMO chain count per radio, and the firmware features that decide roaming. A product page publishes what the vendor chose to publish; the datasheet and the firmware release notes answer the rest.

• Per-band rate and supported channel widths (20, 40, 80 or 160MHz)
• MIMO chains per radio rather than the antenna count on the housing
• Maximum associated clients and recommended active clients
• Consumption at full load and the matching PoE class
• Whether current firmware supports 802.11k, 802.11v and 802.11r
• IP rating and the mounting kit: ceiling, pole or wall box

Is the 802.11be step worth it on a normal floor?

Only where the clients and the switch are already there. CF-E373BE publishes 3600Mbps across 802.11a/b/g/n/ac/ax/be, 2.5G WAN and 2.5G LAN ports, and under 24W; it suits a design studio, an edit room or a 4K meeting floor that already has 2.5G switching behind it. On a normal office floor, two 1800 units placed well beat one 3000 unit.

Send the floor area, the client mix and the switch you already own, and the quote follows the tier that matches rather than the tier with the biggest number.