Is Ethernet always faster than Wi-Fi?

Not in raw bandwidth — a Wi-Fi 6E client at close range can outpace a 1 Gbps Ethernet port. But Ethernet is essentially always faster for latency, jitter, and reliability, which usually matter more for the user experience than peak throughput. A 2.5 Gbps Wi-Fi link with 30 ms of jitter is worse for a video call than a 1 Gbps Ethernet link with 0.5 ms of jitter.

Does Wi-Fi 6 fix the latency problem?

It helps significantly — OFDMA and improved MU-MIMO mean the air is sliced more finely so contention costs less, and the higher modulation rates mean each packet spends less time on-air. A Wi-Fi 6 link in good conditions can hit Ethernet-like latency. But Wi-Fi is still half-duplex on a shared medium with interference, and at the edge of range or in a contested apartment-block environment, the underlying physics still wins.

I'm gaming on Wi-Fi and my pings are fine — do I really need Ethernet?

For casual single-player or asynchronous online games, Wi-Fi is genuinely fine. For competitive FPS, fighting games, or anything where you care about input-to-action latency at the bottom percentile, Ethernet usually buys you something measurable — not in average ping but in the worst 1% of frames where a Wi-Fi retransmission would otherwise cause a stutter. Try both and look at the jitter, not the average.

What about powerline adapters?

Powerline (HomePlug / G.hn) can be a reasonable substitute for Ethernet when running a cable is genuinely impossible. The throughput is usually 200-600 Mbps in practice (despite higher advertised numbers), the latency is comparable to wired Ethernet (sub-millisecond), and it doesn't degrade with distance the way Wi-Fi does. Pair quality depends heavily on your house's electrical wiring — older homes with split phases can be much worse.

Wi-Fi vs Ethernet — the physics, the numbers, and when the cable wins

The fundamental difference: dedicated wire vs. shared air

Ethernet on a switched port is full-duplex on a dedicated medium. Every device has its own twisted-pair cable to the switch; one pair carries data in one direction, the other pair carries the opposite direction simultaneously. No device competes with any other for the wire. Collisions, which used to be a defining feature of Ethernet in the hub era, haven't existed since switches became universal in the late 1990s.

Wi-Fi is half-duplex on a shared medium. Every device on the same channel transmits onto the same piece of air. Exactly one device can be transmitting at any instant on a given channel; everyone else has to wait. Worse, devices cannot reliably tell whether their transmission collided with someone else's — the receiver might pick up nothing useful, the sender doesn't know the packet was lost without an ack.

To work around this, Wi-Fi uses CSMA/CA — Carrier Sense Multiple Access with Collision Avoidance. Each device, before transmitting, listens for the channel to be quiet. If it's quiet, the device waits a random small interval, then transmits and hopes nobody else picked the same slot. This back-off is what makes Wi-Fi inherently slower than Ethernet even on an idle channel.

Every source of Wi-Fi latency, listed

The wired-versus-wireless latency gap on a healthy network is usually 1–5 ms. On an unhealthy network it can be 50× worse. The contributors:

Channel contention. Phones, smart bulbs, smart speakers, doorbells, laptops, tablets — every active device on the same channel is one more candidate you wait behind. The contention cost grows roughly with the square of the number of active devices, because each one increases both the chance of needing to wait and the length of the wait.
Interference. Microwaves (legacy 2.4 GHz overlap), Bluetooth devices, the neighbour's Wi-Fi, baby monitors, even poorly-shielded USB 3.0 cables — anything generating RF in the bands you use can corrupt a packet. Each corruption forces a retransmission, which is another back-off cycle on top of the wasted air time.
Distance and signal strength. Modern Wi-Fi adapts its modulation rate to the signal-to-noise ratio. A strong signal can pack 256-QAM into each symbol; a weak signal falls back to BPSK with half the throughput. Same packet, more time spent on air. At the edge of range, your laptop might be transmitting at 1/16 the speed of one at close range — meaning 16× longer to send the same byte, and 16× longer it's blocking everyone else.
Retransmissions. Wi-Fi acks every unicast packet. A missed ack triggers a retransmit, which goes to the back of the queue along with whatever has queued up since. Tail latency under load is dominated by these retransmits — your average ping might be 5 ms, but the worst 1% of packets can take 50–100 ms because they were re-sent twice.
Sleep modes. Battery-powered devices don't keep their radios powered on at full duty cycle. Your phone negotiates a DTIM interval with the AP and only wakes its radio at scheduled intervals to check for traffic. The result is millisecond-level delays on incoming packets to sleeping clients. Most desktops and laptops don't sleep this aggressively, so the issue is mostly mobile.
AP queue depth. The access point has a single per-client queue; if it's deep, your packet waits behind every other packet for that client. Modern APs implement airtime fairness so a slow client can't hog the channel, but it still adds latency for everyone.

The numbers, in realistic conditions

Numbers from a healthy modern home network, hop 1 to the access point or first switch:

Gigabit Ethernet, switched, idle: 0.1–0.5 ms, sub-microsecond jitter, 0% loss. This is the floor — you cannot beat it on any wireless link, ever.
Wi-Fi 6 (802.11ax), 5 GHz, close range, idle: 1–3 ms, < 1 ms jitter, ~0% loss. Genuinely close to Ethernet for many real-world tasks.
Wi-Fi 6, 5 GHz, contested (kids streaming + smart-home traffic): 5–15 ms, 2–5 ms jitter, occasional 1% retransmissions. Noticeable on video calls if you look for it.
Wi-Fi 5 (802.11ac), 2.4 GHz, neighbour interference: 10–40 ms, 5–20 ms jitter, 1–3% loss. The "Wi-Fi is bad" experience most apartment-dwellers recognize.
Wi-Fi at the edge of range: 50–200 ms, double-digit jitter, painful packet loss. Streams stutter, calls drop, games rubber-band.

For most browsing and video streaming, none of this is noticeable — the upstream bottleneck is your ISP, not the Wi-Fi. For video calls, the difference is detectable if you measure but rarely makes calls fail. For competitive gaming, voice work where you're cohosting, and real-time collaboration tools, the difference is the whole experience.

When Wi-Fi is genuinely fine

Browsing the web, scrolling social media, watching pre-buffered streaming video.
VoIP calls in normal conditions — most VoIP codecs handle 30 ms of jitter transparently.
Cloud-saved gaming where the only network step is downloading the save state.
Asynchronous multiplayer (turn-based, MMOs where 50 ms is rounding error).
Cloud development environments where you accept some round-trip cost anyway.

When you want the cable

Competitive online gaming. FPS, fighters, racing games, MOBAs at high MMR — every retransmission is a frame of input delay you can feel. Pros wire in for tournaments not because the average ping is better but because the worst ping is dramatically better.
Hosting a server or peer-to-peer game lobby. Anyone connecting to you experiences the latency of your worst hop, which is usually your Wi-Fi.
Latency-sensitive remote work. Live pair-coding, RDP sessions, real-time audio production, anything where you watch your own cursor move.
4K streaming on flaky Wi-Fi. Buffering is fine but uncacheable (live sports, live concerts) is where Wi-Fi often fails.
Bulk file transfers. Sustained throughput is much higher and more predictable on a wired link, so a 50 GB backup that takes 10 minutes on Ethernet might take an hour on Wi-Fi in the same house.

What to do when Ethernet isn't an option

Use 5 GHz, not 2.4 GHz. 5 GHz has more non-overlapping channels and far less interference from microwaves, Bluetooth, and the apartment next door. The trade-off is shorter range; the win is dramatically lower contention.
Upgrade to Wi-Fi 6 or 6E. OFDMA and improved MU-MIMO genuinely change the latency story. Wi-Fi 6E adds the 6 GHz band, which is empty almost everywhere as of 2026 — phenomenal for latency-sensitive devices.
Use a wired-backhaul mesh. Most mesh systems can be deployed with Ethernet between the nodes instead of using the air for backhaul. Pick this option if any of your nodes can be wired; mesh-with-wireless-backhaul roughly doubles the latency cost.
Enable QoS / WMM. Wi-Fi Multimedia (WMM) prioritizes traffic categorized as voice or video so it cuts the queue ahead of bulk transfers. Critical if you do video calls while someone else in the house torrents.
Try Powerline or MoCA. Both use existing wiring (electrical or coax) as the backhaul. Latency is wired-level (sub-millisecond), throughput is typically 200–600 Mbps real-world, no air contention.
Diagnose with traceroute and MTR. The first hop in your trace is almost always Wi-Fi if you're wireless. If hop 1 shows 30 ms latency or 2% loss, fix Wi-Fi before blaming your ISP. See the traceroute explainer.

The diagnostic recipe

The fastest way to find out if your wireless is the problem:

Plug a laptop into your router with an Ethernet cable. Run a sustained ping to1.1.1.1 (ping -c 50 1.1.1.1). Record the average, max, and stddev.
Unplug, connect the same laptop to Wi-Fi from the same room. Run the same ping. Compare.
If the difference is >10 ms in average or >5 ms in stddev, Wi-Fi is your bottleneck. If the difference is small, look upstream of your router for the problem.

Wi-Fi vs Ethernet — why your ping is worse on wireless