“Faster internet” usually means more bandwidth
Latency
Delay. A packet may take 20 ms to reach a server whether your plan is 100 Mbps or 5 Gbps.
Bandwidth
Capacity. A 5 Gbps connection can carry roughly 50× as much data per second as 100 Mbps.
Throughput
The rate you actually achieve after protocol overhead, Wi‑Fi conditions, hardware limits, server limits, and congestion.
A 5 Gbps plan does not create a 5 Gbps device
Every transfer passes through a chain of components. The slowest relevant component becomes the bottleneck.
If the PC has a 1 Gbps network interface, that individual wired PC cannot receive 5 Gbps through that interface. The service can still be useful because several devices can consume bandwidth simultaneously.
Wi‑Fi is not a cable in the air
Wi‑Fi is a shared, half-duplex radio system. Devices take turns transmitting on the same airtime. Signal quality, interference, channel width, spatial streams, protocol generation, client hardware, and distance all affect real throughput.
2.4 GHz
Longer range and better wall penetration, but fewer clean channels and usually lower real throughput. Often crowded by neighboring Wi‑Fi, Bluetooth, and other 2.4 GHz devices.
5 GHz
Higher capacity and more channel options, but weaker range through walls. Often the practical sweet spot for modern home Wi‑Fi.
6 GHz
Very clean spectrum and wide channels for Wi‑Fi 6E/7 devices, but shorter effective range and weaker penetration. Excellent when close to the access point.
Why “I pay for 5 gig, why don’t I get 5 gig over Wi‑Fi?” is the wrong expectation
The ISP plan is the capacity delivered to the router. A Wi‑Fi client has its own separate radio link to that router.
| Limit | What it means |
|---|---|
| Wi‑Fi generation | A Wi‑Fi 5 laptop cannot gain Wi‑Fi 7 radio capabilities because the internet plan changed. |
| Spatial streams | Many phones and laptops are 2×2 clients. They cannot use all streams advertised by an 8-stream router. |
| Channel width | 20/40/80/160/320 MHz affects theoretical link rate. Wider is not always better in crowded spectrum. |
| Signal quality | As signal-to-noise ratio falls, Wi‑Fi shifts to more robust but slower modulation rates. |
| Airtime contention | Nearby clients and neighboring networks consume shared airtime even when your ISP link is idle. |
| Protocol overhead | A displayed PHY/link rate is not the same as usable TCP/UDP throughput. |
Interference and distance
Things that reduce Wi‑Fi performance
Walls, floors, metal appliances, mirrors, ductwork, low signal, neighboring APs, poorly chosen channels, legacy clients, mesh backhaul, repeaters, and competing traffic all consume link margin or airtime.
What a speed test is really testing
When run over Wi‑Fi, it tests the entire path: device radio → local RF environment → access point → router → ISP → internet → test server. It is not a pure measurement of the ISP circuit.
Ethernet has negotiated speed limits
Your Ethernet devices negotiate a link rate. If Windows reports 1.0 Gbps, a 5 Gbps internet plan cannot push 5 Gbps through that 1 Gbps link.
| Ethernet link | Approximate maximum before overhead | Typical use |
|---|---|---|
| 100 Mbps | 100 Mbps | Old devices, damaged/2-pair cabling, legacy switches |
| 1 GbE | 1 Gbps | Very common PCs, routers, switches |
| 2.5 GbE | 2.5 Gbps | Modern PCs, APs, multigig home networking |
| 5 GbE | 5 Gbps | Higher-end multigig links |
| 10 GbE | 10 Gbps | Workstations, servers, high-end routers/switches |
Cables matter — but not the way internet folklore says
| Cable | Practical guidance |
|---|---|
| Cat5e | Designed for 1 GbE to 100 m and commonly supports 2.5G/5G over suitable runs and equipment. A good Cat5e cable does not need to be replaced merely because someone bought gigabit service. |
| Cat6 | Excellent general-purpose choice. Supports 1G/2.5G/5G and can support 10G on shorter runs, depending on installation conditions. |
| Cat6A | Designed for 10 GbE at full 100 m channel length. Useful when building for reliable 10G structured cabling. |
| Cat7/Cat8 marketing | Often over-sold to consumers. Buying an expensive “Cat8 gaming cable” does not make a 1 GbE NIC operate at 10 Gbps. |
Every wired hop must support the target rate
That 1G LAN port caps this path near gigabit even though everything before and after it is faster.
The endpoint can be the bottleneck
Network interface
A 1 GbE NIC caps wired traffic near 1 Gbps. A Wi‑Fi 5/6/6E/7 client has its own radio capabilities independent of the router.
CPU and software
VPNs, security software, packet inspection, browser overhead, old CPUs, USB Ethernet adapters, and drivers can limit high-speed transfers.
Storage
At multi-gigabit rates, slow disks, decompression, antivirus scanning, or game-launcher patching can become the bottleneck instead of the network.
Why a game downloads at 1 Gbps on a 5 Gbps plan
A game download is not a raw line-rate test. The launcher may download, decrypt, decompress, verify, patch, and write data to storage. The content provider may also rate-limit a single session.
USB adapters and docking stations
A “2.5G Ethernet adapter” can still underperform if it is attached through a constrained USB path, a low-end hub, a dock sharing bandwidth with other devices, or a host with poor drivers. Always inspect the entire connection path, not just the label on the adapter.
How to troubleshoot a speed complaint correctly
Define what is being measured
Is the complaint a speed-test result, Steam download, file transfer, Wi‑Fi link rate, webpage load time, latency, or buffering? These are different measurements.
Identify the test path
Wi‑Fi or Ethernet? Which router port? Which switch? Which cable? Which NIC? What negotiated link rate?
Find the lowest link rate
A PC reporting 1.0 Gbps Ethernet will not produce a 5 Gbps speed test. A router with only gigabit LAN ports cannot deliver multigig to one wired client.
Remove Wi‑Fi from an ISP line test
For a controlled throughput test, use a capable wired client and multigig hardware. Wi‑Fi results are useful, but they measure the RF environment too.
Check the server and application
Try more than one reputable speed-test endpoint. Game launchers and remote servers can be slower than the ISP connection.
Check concurrent traffic
Cloud backups, security cameras, uploads, game updates, streaming, and other users may already be consuming part of the connection.
Compare expectations to hardware
To test 5 Gbps to one PC, the entire path must be capable of more than 5 Gbps: WAN handoff, router routing/NAT performance, LAN port, switch, cable, NIC, system bus, and test server.
Example call: “I have 5 gig but my PC only gets 940 Mbps.”
If it says 1.0 Gbps, ~900–950 Mbps on a well-performing TCP speed test is normal for that link after Ethernet/IP/TCP overhead. The 5 Gbps service is not the immediate bottleneck — the PC’s 1 GbE path is.
Next check whether the PC, switch, and router have 2.5G/5G/10G ports, and whether the cable and drivers support the desired negotiated rate.
Common internet-speed myths
“I pay for 5 Gbps, so every device should get 5 Gbps.”
No. The plan is aggregate access capacity. Each device is limited by its own local link and the path it uses.
“My Wi‑Fi says 2400 Mbps, so my speed test should show 2400 Mbps.”
No. Wi‑Fi PHY rate includes signaling conditions and protocol overhead; usable application throughput is lower and varies constantly.
“A Cat8 cable will make my gigabit PC faster.”
No. A 1 GbE NIC and 1 GbE switch port still negotiate at 1 Gbps. The cable cannot upgrade the electronics on either end.
“If a speed test is slow, the ISP must be slow.”
Not necessarily. The result includes the client, local network, Wi‑Fi or Ethernet link, router, ISP, internet path, and test server.
“More bars means faster Wi‑Fi.”
Signal strength helps, but it is not the whole story. Noise, interference, airtime use, client capability, channel width, and modulation also matter.
“Gigabit means 1 GB per second.”
No. Gigabit internet is measured in gigabits. 1 Gbps is about 125 megabytes per second before overhead.
“Ethernet always means full speed.”
No. Ethernet can negotiate at 100M, 1G, 2.5G, 5G, 10G, and other rates. A damaged cable or legacy port can reduce the link rate.