The Hidden Math Behind WiFi Speed Claims: What 9.6 Gbps Really Means
Every WiFi router box advertises a number. WiFi 6 routers claim “up to 9.6 Gbps.” WiFi 7 boxes say “up to 46 Gbps.” Somewhere in your home is a router that claims speeds you have never once measured. There is no deception happening, exactly — the numbers are real — but the gap between the specification ceiling and the performance you experience is built from a stack of assumptions that the packaging does not explain.
The PHY Rate Is Not Your Throughput
The speed number on the box is the PHY rate: the maximum rate at which the radio can push modulated symbols across the channel under ideal conditions. It is a physical layer measurement, not a practical throughput figure. The relationship between PHY rate and the data rate your applications experience involves several unavoidable losses.
Protocol overhead consumes a meaningful fraction of every transmitted frame. The 802.11 MAC header, frame check sequence, inter-frame spacing, and acknowledgment frames together represent overhead that, on legacy OFDM networks, accounts for 20 to 40 percent of channel time depending on packet size and traffic pattern. Small packets — typical of VoIP, gaming, and IoT applications — have proportionally higher overhead because the fixed-size headers represent a larger fraction of each frame. The efficiency gap between PHY rate and useful throughput starts here.
Retransmissions compound the loss. Any frame that arrives with errors must be retransmitted. In a clean RF environment with strong signal, error rates are low and retransmission overhead is minimal. As signal quality degrades — due to distance, walls, interference, or multipath — the error rate rises, retransmissions consume more channel time, and effective throughput falls. A link running at 75 percent packet success rate is not delivering 75 percent of its PHY rate; the retransmission overhead of the missing 25 percent degrades effective throughput more severely than that.
The Assumption Stack in PHY Rate Calculations
The 9.6 Gbps ceiling for WiFi 6 is not arbitrary. It is the product of several multiplied values, each at its theoretical maximum:
Eight spatial streams at 1.2 Gbps each produces 9.6 Gbps. The 1.2 Gbps per stream assumes 160 MHz channel width, 1024-QAM modulation, a 5/6 coding rate, and a short guard interval. Every one of these values is the maximum possible.
Eight spatial streams requires an 8×8 MIMO configuration on both the AP and the client device. No consumer client device ships with more than 4×4 MIMO. Nearly all smartphones and most laptops implement 2×2. The router may have four or eight antennas; the phone you are holding has two. The maximum concurrent spatial stream count is limited by the lesser of the two endpoints. A 4×4 AP and a 2×2 phone negotiate two streams, not eight.
160 MHz channels are available only in the 5 GHz band, and only where two contiguous 80 MHz blocks are available on non-DFS frequencies — which means channels 36–64 and 100–144. Regulatory environments, DFS radar interference, neighboring network congestion, and practical spectrum fragmentation mean 160 MHz operation is available far less often than the 80 MHz default. Most real deployments operate at 80 MHz.
1024-QAM requires an SNR of approximately 35 to 38 dB to function reliably. This SNR is achievable at close range with line-of-sight to the AP. Through one wall, at five meters, the SNR of a typical deployment drops to a range where the radio negotiates 256-QAM or 64-QAM instead. The modulation scheme — and therefore the per-stream data rate — adjusts continuously based on measured signal quality.
The short guard interval shaves approximately 11 percent more throughput by reducing the silence gap between OFDM symbols. It requires low multipath — that is, very little reflected signal arriving at the receiver. In a typical indoor environment with hard surfaces, standard guard interval is safer.
The Real-World Performance Zone
A realistic performance assessment for a WiFi 6 client under good conditions — within 5 meters of the AP, one lightweight interior wall, on a 80 MHz channel, 2×2 MIMO configuration — is approximately 500 to 700 Mbps of actual application throughput. At 15 meters through two walls, this drops to 200 to 350 Mbps depending on construction materials. At 25 meters through a concrete wall, it may be 50 to 100 Mbps or less.
These are still excellent numbers for a wireless connection. 500 Mbps WiFi 6 throughput far exceeds any consumer internet subscription speed available in 2026. The practical bottleneck for home users is the internet connection, not the WiFi link. The speed numbers matter more for local network transfers — a NAS serving video files, inter-device synchronization — where the WiFi hop must carry real data volumes between devices on the same network.
The Multi-Stream Confusion in Router Marketing
Router manufacturers discovered that adding streams multiplies their headline numbers. A router with 4×4 on 5 GHz and 4×4 on 2.4 GHz produces a combined number of theoretical throughput on both bands simultaneously — the so-called AX6000 or AX12000 designations. These numbers assume a client on each band simultaneously hitting peak throughput simultaneously. No single device is receiving this combined number. It is the aggregate ceiling across the entire network, never the per-device rate.
The convention is technically accurate in the same way that “up to 9.6 Gbps” is technically accurate. It describes a maximum that requires simultaneous optimal conditions on all radios, all bands, and multiple ideal clients at once. It is a ceiling for the entire system, not a speed any one device will achieve.
What the Numbers Should Inform
The spec sheet throughput numbers are not useless — they indicate relative capability between products, reflect the generation of the underlying standard, and determine how much headroom exists above realistic use. A WiFi 7 router’s higher ceiling means better performance per client when multiple high-bandwidth users share the AP simultaneously, even if no single client ever approaches the theoretical maximum. The gap between ceiling and floor is where the real comparison lives.
When evaluating hardware, look for tested real-world throughput benchmarks from sources like SmallNetBuilder, Tom’s Hardware, or Ars Technica — these use actual measurement against real client hardware at standardized distances. The difference between a $200 WiFi 7 router and a $400 WiFi 7 router rarely appears in the PHY rate spec; both will claim the same IEEE maximum. It appears in measured throughput at 5 meters and 15 meters, in client count under load, and in firmware quality. That is where the actual product differentiation is.