Wi-Fi vs wired for editing off a NAS (and when Wi-Fi is fine)

Wired wins for editing off a NAS, and it is not close when you are scrubbing high-bitrate timelines. But Wi-Fi is genuinely fine for a real slice of the work: reviewing cuts, editing on proxies, dropping a render to the share overnight. The honest answer is not "always plug in." It is "wire the seat that scrubs raw media, and stop apologizing for Wi-Fi on the seat that does not." This post draws that line with the two numbers that decide it: how much throughput your codec needs, and how steady the connection has to be to feel instant.

The two numbers that decide it #

Editing off network storage is a real-time job. The link has to deliver enough bits per second to keep playback fed (throughput), and it has to deliver them on a steady cadence so the next frame is always there when the player reaches for it (latency, and its evil twin, jitter). Throughput is the size of the pipe. Latency is how long the water takes to arrive, and jitter is whether it arrives in an even stream or in unpredictable gulps. A player can ride out a slow pipe by buffering ahead. It cannot ride out gulps it did not see coming, which is why a connection that "tests fast" can still stutter the moment you scrub.

Wired Ethernet is good at both because it is full-duplex and the wire is yours alone: your machine sends and receives at the same time with no one to wait for. Wi-Fi is a shared, half-duplex radio medium where every device near you takes turns transmitting and backs off a random interval before each turn (CSMA/CA). That sharing is exactly what makes throughput inconsistent and latency spiky. Cat6 Ethernet jitter typically sits at 0.1-2 ms and almost always under 1 ms; even good uncongested 5 GHz Wi-Fi runs 3-8 ms, and 2.4 GHz can swing 15-40 ms with frequent spikes (jitter.is, checked Jun 2026). Those spikes are what a video player feels as a dropped frame.

What your codec actually needs #

Start with the demand side, because it sets the floor. The numbers below are Apple's published ProRes target data rates (Apple ProRes White Paper, April 2022, checked Jun 2026). One stream at a time, before you add a second angle, audio, or a coworker on the same volume. I have converted Apple's Mbps to MB/s, because that is what your file-copy dialog speaks.

The jump from HD to 4K is the whole story. One stream of ProRes 422 HQ at UHD is roughly 118 MB/s, and 4444 is about 177 MB/s. Now picture the real timeline: two camera angles, a few stacked color nodes pulling extra reads, and you are asking for 250-350 MB/s of steady delivery from a single editor. That is the number Wi-Fi has to beat, not on a speed test, but continuously, for hours, while the rest of the office is also on the air.

Why Wi-Fi's real ceiling is lower than the box claims #

Router boxes advertise PHY rates, not throughput, and the gap is large. Real-world goodput on 802.11 networks runs around 50 percent of the advertised PHY rate once you account for preamble, inter-frame spacing, headers, and acknowledgments. An 802.11ac link rated 1,300 Mbps lands near 650 Mbps of usable throughput in practice (CWNP and 802.11 overhead analyses, checked Jun 2026). Wi-Fi 7 helps a lot here through wider 320 MHz channels and Multi-Link Operation, which lets a client use the 5 GHz and 6 GHz bands at once; one tested setup held about 2.8 Gbps moving a 4K file from a NAS while a video call stayed glitch-free (Wi-Fi 7 vs Wi-Fi 6 testing, checked Jun 2026).

Two cautions before you bank on those Wi-Fi 7 figures. First, your hardware may not reach them: Apple capped Wi-Fi 7 in the iPhone 16 line at 160 MHz rather than the full 320 MHz, and the M4 MacBook Pro, Mac mini, and iMac shipped with Wi-Fi 6E, not Wi-Fi 7 (MacRumors, checked Jun 2026); Wi-Fi 7 with MLO arrived on the Mac side with the M5 MacBook Pro. Second, the headline number is one client, one room, line of sight. Add a wall, a second editor, and a busy 2.4 GHz band, and the shared medium does what it always does: airtime is split, and your steady 300 MB/s becomes a number that visits and leaves.

When Wi-Fi is genuinely fine #

Most of the noise about Wi-Fi being "unusable for editing" assumes you are scrubbing raw 4K. You usually are not. The fix the entire industry already uses is proxies: lower-resolution stand-in media you cut with, then relink to the originals for the final render. ProRes 422 Proxy at UHD is about 24 MB/s, which is well inside what a clean 5 GHz or 6 GHz link delivers all day. DaVinci Resolve and Premiere both build proxies in a couple of clicks, and the editorial industry standard is to cut on Proxy or LT and conform at the end (proxy-workflow guides, checked Jun 2026).

So Wi-Fi is fine when the work is light or buffered rather than hard real-time. Concretely: editing on proxies; reviewing and commenting on cuts; pulling stills and audio; copying a finished render to the share where seconds of variance do not matter; a laptop seat that occasionally touches the NAS but lives in a browser. The common thread is that none of these need a guaranteed, unbroken 250+ MB/s stream for hours. If your codec demand sits under roughly 50-60 MB/s and a stutter would only cost you a re-scrub rather than a recorded take, Wi-Fi is a defensible choice, not a compromise you are hiding.

When only wired works #

Wire the seat the moment the work becomes hard real-time off full-resolution media. Scrubbing multicam ProRes 422 HQ or 4444 at 4K, color grading with stacked nodes, finishing and mastering, or any timeline where a dropped frame is a real problem all want the wire. The reason is not just the ceiling; it is the floor. A 2.5GbE link delivers about 280 MB/s of real-world throughput and a 10GbE link sustains 850-950 MB/s (multi-gig NAS testing, checked Jun 2026), and crucially it delivers that on a flat, predictable cadence that a player can plan around. That steadiness is the actual product. For the full direct-attach-versus-network tradeoff, see Thunderbolt vs 10 GbE for editors, and for picking the link speed itself, 10 GbE on a budget.

One more honest note: the link is only half the path. A single spinning disk barely brushes 2.5GbE and never reaches 10GbE, so a wired seat fed by a slow array or a cache that keeps missing will still stutter. If your wired connection feels slow, the culprit is often upstream of the cable. The companion piece diagnosing slow NAS performance walks the whole chain, and why your NAS feels slow over a VPN covers the case where distance, not the local link, is the problem.

How to decide for your room #

The rule is short. Look up your editing codec's per-stream rate in the first table, multiply by the number of streams a busy timeline actually plays, and add headroom. If that figure clears comfortably with margin to spare on Wi-Fi in your specific room, with your specific gear, after a stutter test on real footage, use Wi-Fi for that seat. If it is close, or the work is hard real-time, run the wire. Test on your own material, not a synthetic benchmark, because the thing that fails on Wi-Fi is rarely the average and almost always the spike.

This is also where a cache-first mount changes the math, and where I will be honest about the limit. JuiceMount keeps a local SSD cache, so once media is warm, scrubbing and looping a region are served from the SSD at local speed and the network only carries the misses, which makes a Wi-Fi seat far more forgiving than raw streaming would suggest. What the cache cannot do is invent bandwidth on a cold pull: the first read of a fresh 4K clip still crosses the link, so for a colorist hammering brand-new full-res media all day, wire the seat regardless of the mount. The cache smooths the warm path; it does not repeal the throughput floor.