Wi-Fi 7 marketing leads with a number: up to 46 Gbps. That figure is real in the sense that the arithmetic supports it, and misleading in the sense that no deployment on earth will see it. It assumes 16 spatial streams, a 320 MHz channel, and a modulation scheme that needs near-laboratory signal conditions.

The interesting parts of 802.11be are underneath the headline. One of them — Multi-Link Operation — breaks an assumption that has held since 802.11a. And nearly every gain the standard offers is gated by something the spec sheet never mentions: the antenna, the feedline, and the physics of 6 GHz.

What Wi-Fi 7 Is

Wi-Fi 7 is the Wi-Fi Alliance's brand name for IEEE 802.11be, formally titled Extremely High Throughput (EHT). The Alliance launched its Wi-Fi CERTIFIED 7 program in early 2024, and certified silicon has been shipping in volume since.

It is best understood as the completion of what Wi-Fi 6E started rather than a clean break. Wi-Fi 6E extended 802.11ax into the 6 GHz band and stopped there — same channel widths, same modulation, same one-radio-at-a-time association model. Wi-Fi 7 keeps the 6 GHz band and rebuilds what happens inside it: wider channels, denser modulation, smarter use of fragmented spectrum, and multiple simultaneous links.

Five features do most of the work.

320 MHz Channels

Wi-Fi 7 doubles the maximum channel width from 160 MHz to 320 MHz. Double the bandwidth, roughly double the throughput — the simplest gain in the standard, and the most constrained.

320 MHz channels exist only in 6 GHz. There is no room for them anywhere else. The 6 GHz band spans 1200 MHz (5925–7125 MHz) where regulators have released all of it, which yields three non-overlapping 320 MHz channels. That is the ceiling, and it assumes full-band access.

Full-band access is not the global reality. Roughly 97 countries have opened some portion of 6 GHz for unlicensed Wi-Fi, but the allocations diverge sharply:

• Full 1200 MHz: United States, Canada, Saudi Arabia, South Korea
• Lower half only (5925–6425 MHz): European Union, United Kingdom, UAE, Qatar, and much of APAC — enough for one 320 MHz channel, with no room to plan around it
• None: China, which allocated 6 GHz to IMT for 5G/6G instead

Power class narrows it further. Under Standard Power with Automated Frequency Coordination (AFC) — the mode that gives you outdoor range and real transmit power — the US permits just one 320 MHz channel and Canada two. Low Power Indoor gets the full three, but LPI is exactly what its name says.

The practical read: 320 MHz is a targeted tool, not a default. Deploy it where a specific set of clients needs the throughput. Trying to blanket a multi-AP site with 320 MHz channels means co-channel interference, because there is nothing left to plan with.

4096-QAM (4K-QAM)

Wi-Fi 6 encodes 10 bits per subcarrier with 1024-QAM. Wi-Fi 7 pushes to 4096-QAM and 12 bits, adding MCS 12 and 13 to the rate table. The gain is about 20% higher peak PHY rate.

The cost is signal quality, and this is where Wi-Fi 7 stops being a firmware conversation and becomes an RF engineering one.

Constellation density and noise tolerance trade against each other directly. Packing 4096 points into the same constellation means the points sit closer together, and the receiver needs a much cleaner signal to tell them apart:

An 11 dB jump over Wi-Fi 6. On the transmit side, the IEEE tightens the EVM requirement to −38 dB for 4096-QAM — 3 dB stricter than 802.11ax.

In an open environment, 42 dB SNR means the client is within a few feet of the AP. Not across the room — a few feet. Vendors are candid about this; 4K-QAM is an optional feature for Wi-Fi 7 certification precisely because it so rarely engages.

There's a second penalty stacked on top: doubling channel width from 160 MHz to 320 MHz raises the noise floor by 3 dB, which is why 802.11be raises minimum receiver sensitivity requirements by 3 dB for 320 MHz. The two headline features fight each other. The widest channel makes the densest modulation harder to reach.

This is the section that should reframe how you think about antennas. Under 802.11n, a mediocre antenna cost you some range. Under 802.11be, a mediocre antenna doesn't cost you 4K-QAM range — it costs you 4K-QAM entirely. The feature simply never engages, and you paid for silicon that idles at Wi-Fi 6 rates.

Multi-Link Operation (MLO)

MLO is the genuinely new idea in Wi-Fi 7, and the one most likely to matter in real deployments.

Every prior generation tied a client to one link at a time. You associated to 2.4 GHz, or 5 GHz, or 6 GHz. Band steering and roaming existed, but they were reactive — something had to decide to move you, and the handoff cost you.

Wi-Fi 7 introduces the Multi-Link Device (MLD): a single logical association spanning multiple radios simultaneously. The MLD presents one MAC address upward while each affiliated radio keeps its own link-level MAC. Applications get the benefit with no added software complexity — the MAC layer absorbs it.

What this buys you:

• Latency floor. With links live on two bands at once, a busy channel on one doesn't stall the association. Traffic goes out on whichever link is clear. For AR/VR, cloud gaming, and industrial control, the consistency matters more than the peak.
• Reliability. Interference on 6 GHz no longer means a renegotiation on 5 GHz. The link is already up.
• Aggregation. Simultaneous Transmit-Receive (STR) capable MLDs run uplink and downlink concurrently on a link pair.

MLO is why Wi-Fi 7 is worth deploying even where 320 MHz channels aren't available and 4K-QAM never engages. It works in the lower half of the 6 GHz band, and it works in 5 GHz.

The hardware consequence: MLO means multiple radios transmitting at once, in one enclosure, across adjacent bands. Antenna isolation between chains stops being an optimization and becomes a functional requirement. Poor isolation turns your own second radio into your noise source.

Preamble Puncturing and Multi-RU

Wi-Fi 6 had a brittle relationship with wide channels: if any part of the channel was occupied by an incumbent or a neighboring network, the whole channel was unusable. A 20 MHz obstruction wasted 140 MHz of clean spectrum.

Preamble puncturing lets a Wi-Fi 7 device mask off the blocked subchannel and transmit on the remainder. Multi-RU extends the flexibility to resource-unit allocation within OFDMA, letting a device use non-contiguous chunks rather than a single block.

This matters more in practice than the puncturing coverage suggests, because 6 GHz has incumbents — fixed microwave links and satellite services — and AFC exists specifically to protect them. Puncturing is how wide channels stay viable in the spectrum you're actually allowed to use.

16×16 MU-MIMO

Wi-Fi 7 doubles the spatial stream ceiling from 8×8 to 16×16 MU-MIMO. This is where the 46 Gbps figure comes from, and it's also the least likely feature to appear in shipping hardware anytime soon.

Spatial streams are not free. Each one is a physical radio chain and a physical antenna element, and they need spatial or polarization separation to decorrelate. A 16×16 AP is sixteen chains and sixteen elements. Expect this in high-density enterprise APs, not in anything resembling a consumer router — most Wi-Fi 7 clients ship 2×2.

The number to internalize: a 2×2 client with 320 MHz, 4K-QAM, and MLO tops out around 5.76 Gbps in theory. That is the real-world ceiling for the device in your pocket, and it's an optimistic one.

What Wi-Fi 7 Requires From Your RF Hardware

This is where most Wi-Fi 7 explainers stop and where deployments actually fail.

6 GHz-capable antennas — not repurposed 5 GHz antennas. A dual-band 2.4/5 GHz antenna does not become a 6 GHz antenna because 6 GHz is nearby. Outside its designed band, the element's return loss degrades, VSWR climbs, and the radiation pattern distorts. You will get a link. You will not get 42 dB SNR, and 4K-QAM will never engage. Verify the datasheet covers 5925–7125 MHz. [LINK: 6 GHz / Wi-Fi 7 antennas]

Low-loss coax, because loss scales with frequency. Cable attenuation rises with frequency — an RF fundamental with direct consequences at 6 GHz. Coax that was acceptable at 2.4 GHz is meaningfully worse at 6 GHz over the same run. Every dB lost in the feedline is a dB off your SNR budget, and the 4K-QAM budget has no slack. Two responses: specify lower-loss cable, and shorten the run. [LINK: low-loss coaxial cable assemblies]

Connector selection. At 6 GHz, connector quality and mating consistency contribute measurable loss and reflection. Reflections degrade EVM, and 802.11be tightened the EVM spec by 3 dB. A cheap connector at 6 GHz costs more than a cheap connector at 2.4 GHz. [LINK: RF connectors and adapters]

MIMO spacing and polarization diversity. Spatial streams need decorrelated paths. Elements too close together, or co-polarized when they should be cross-polarized, collapse the effective rank — the radio advertises 4×4 and delivers something closer to 2×2. With MLO adding simultaneous multi-band transmission, isolation between chains carries more weight than in any prior generation. [LINK: MIMO antennas]

The pattern across all four: Wi-Fi 7's gains are conditional, and the condition is RF quality. Earlier generations degraded gracefully — a compromised RF path meant somewhat less range. Wi-Fi 7 degrades by feature. Miss the SNR threshold and 4K-QAM silently stops engaging. There's no error message. You just get Wi-Fi 6 performance out of Wi-Fi 7 equipment.

Do You Actually Need Wi-Fi 7?

An honest assessment, by deployment.

Strong case:

• High client density — stadiums, lecture halls, conference venues. MLO and puncturing address contention, which is the actual problem in these environments.
• Latency-sensitive applications — AR/VR, cloud gaming, industrial control, real-time collaboration. MLO's consistency is the whole point.
• Congested RF environments — dense apartment buildings, adjacent offices. Puncturing and 6 GHz access are worth real money here.
• New builds — if you're pulling cable and mounting APs anyway, specify 6 GHz-capable RF now. The incremental cost is small; retrofitting is not.

Weak case:

• Typical residential. If your internet service is 500 Mbps, Wi-Fi 7 solves nothing. Wi-Fi 6 already exceeds your WAN link. The bottleneck is upstream.
• Sparse client counts. MLO and puncturing address contention. Without contention, there's nothing to gain.
• 6 GHz-restricted regions. In markets with lower-half-only allocation, you lose the 320 MHz story. MLO still delivers, but the value proposition is narrower.
• Existing RF infrastructure you can't touch. Wi-Fi 7 APs on 5 GHz antennas and long runs of lossy coax will underperform. Fix the RF path or don't bother with the upgrade.

Wi-Fi 6E vs. Wi-Fi 7

What's Next: Wi-Fi 8

IEEE 802.11bn — Wi-Fi 8 — is tracking toward ratification around 2028. Pre-standard silicon is sampling now, with enterprise APs and prototypes expected before consumer devices arrive near the end of the decade.

The notable thing about Wi-Fi 8 is what it isn't chasing. Where every generation since 802.11n has led with a throughput number, Wi-Fi 8's stated goals are ultra-high reliability, intelligent spectrum management, advanced error recovery, and energy efficiency. The industry has largely conceded that peak PHY rate stopped being the constraint several generations ago.

That's a useful lens on Wi-Fi 7. The features that will still matter in 2030 are MLO and puncturing — the ones that make the link dependable. The 46 Gbps headline will age like every headline rate before it.

One point of continuity worth planning around: Wi-Fi 8 depends on full 6 GHz access just as Wi-Fi 7 does, and the regulatory picture remains unresolved in the EU, UK, and India. The RF infrastructure you specify for Wi-Fi 7 today — 6 GHz-capable antennas, low-loss feedline, proper MIMO geometry — is the same infrastructure Wi-Fi 8 will need. The antennas outlive the access points. Specify accordingly.

Data Alliance supplies 6 GHz-capable Wi-Fi antennas, low-loss coaxial cable assemblies, and RF connectors for Wi-Fi 6E and Wi-Fi 7 deployments.https://www.data-alliance.net/customer-service/