I spend my days securing telecom infrastructure, and my evenings answering the same question from neighbors, clients, and one very persistent brother-in-law: should I run cable or just buy a better router? The wired vs wireless home network debate gets framed as a speed contest, and that framing is wrong. Modern WiFi is fast. WiFi 7 on a clean channel will move over a gigabit to a single laptop. Speed is not the reason I still pull Cat6 through attics in July.
The reason is consistency. A cable gives you the same latency, the same throughput, and the same reliability at 2 AM and at 8 PM when every device in the house — and in your neighbor's house — is fighting for airtime. Wireless gives you a shared, half-duplex radio medium that degrades under load, and no amount of marketing changes that physics.
So this article isn't "wired good, wireless bad." It's a sorting exercise. Some devices belong on WiFi. Some devices belong on copper, full stop. And there's a third option — MoCA over your existing coax — that most people have never heard of and that has saved several of my retrofit projects. I'll give you real numbers for all three, because "it depends" without prices is useless.
The Short Version
If you read nothing else:
- Phones, tablets, laptops: WiFi. Always. That's what it's for.
- Cameras, servers, NAS, access points, desktop gaming rigs, the streaming box bolted to your TV: wired. Non-negotiable in my book.
- Rooms you can't reach with cable: MoCA 2.5 over existing coax before you even think about a mesh node on wireless backhaul.
- Building a new house? Run conduit to every room before drywall. It's the cheapest networking decision you will ever make and the one everyone skips.
Now the details.
What WiFi Is Actually Good At in 2027
Modern WiFi is genuinely good. WiFi 6 fixed a lot of the congestion problems of the 802.11ac era with OFDMA. WiFi 6E opened up the 6 GHz band — clean spectrum with no legacy devices squatting on it. WiFi 7 added 320 MHz channels and Multi-Link Operation, which lets a client use multiple bands simultaneously.
Real-world numbers, from my own testing and from what I see on client sites:
- WiFi 6 (5 GHz, 80 MHz channel): 500–700 Mbps to a good client, same room. 200–400 Mbps a room or two away.
- WiFi 6E (6 GHz): 700–900 Mbps close to the AP. The 6 GHz band drops off faster through walls than 5 GHz — great in the same room, mediocre two rooms over.
- WiFi 7 (6 GHz, 320 MHz channel): I've measured 1.5–2 Gbps to a WiFi 7 laptop sitting near a UniFi U7 Pro. Impressive. Also almost never the number you'll see in daily use.
About that "9.3 Gbps" or "BE19000" printed on the WiFi 7 router box: it's theoretical aggregate nonsense. It's the sum of the maximum PHY rates of every radio in the unit combined, achievable by exactly zero real devices, ever. No client on earth connects to all three bands at max modulation simultaneously and sustains it. Marketing departments know this. I wish they'd stop.
The deeper issue isn't peak speed anyway — it's that WiFi is a shared medium. Every device on a band is contending for airtime. Your camera uploading motion clips, your kid's Quest headset, your smart TV pre-buffering — they all take turns on the radio. Latency on a healthy WiFi network is 2–10 ms with occasional spikes to 50+ ms when something hogs airtime. On a cable it's under 1 ms, always, with essentially zero jitter. For loading a webpage, nobody notices. For a competitive shooter or a video call, you notice.
Where WiFi is the right answer: phones, tablets, laptops, e-readers, smart speakers, most IoT junk. These devices move around or don't care about jitter. Wiring a laptop that lives on your couch is solving a problem you don't have.
Where Wired Is Non-Negotiable
Here's my list of things that get a cable in every network I build, and why.
PoE security cameras
This is the hill I'll die on. WiFi cameras are a mistake you will regret, and I say that as someone who reviews camera deployments professionally. Three reasons:
- Power. A "wireless" camera still needs power, so you're running a wire anyway — it might as well be one Ethernet cable that carries both power and data via PoE.
- Reliability. A camera that drops off WiFi during the exact event you needed footage of is worse than no camera, because it gave you false confidence. Sustained 24/7 video upload is the worst-case WiFi traffic pattern: constant, upstream, and competing with everything else.
- Security. WiFi cameras can be trivially knocked offline with a $30 deauthentication tool. A wired camera on its own VLAN can't. In my line of work, an attacker-suppressible camera isn't a security device — it's decoration.
Every camera I install runs back to a PoE switch — a UniFi Switch Lite 16 PoE handles a typical 6–8 camera house with ports to spare.
Access points
The most overlooked one. Your WiFi is only as good as the link feeding each access point. A mesh node on wireless backhaul is a repeater with better marketing — it burns airtime re-transmitting everything, and your throughput at the far node is often half or less of the entry node. Wire your APs and every one of them delivers full performance. This is the paradox of good wireless: it's built on good wiring.
Desktop gaming
Not for bandwidth — a game uses maybe 1 Mbps in play. For latency consistency. Zero jitter, zero airtime contention, zero retransmits when the microwave runs. A wired connection turns "why am I rubber-banding" into a question you never ask again.
Servers and NAS
If you back up your family's photos to a NAS, that box needs a cable — ideally 2.5 GbE or 10 GbE, because a full-machine restore over WiFi is a miserable afternoon. This is also where Cat6A starts earning its keep, more on that below.
The 4K streaming box near your TV
Your Apple TV or Shield sits in one spot, three feet from where the coax or Ethernet already enters the room, streaming 25–40 Mbps of compressed 4K for hours a day. It's the easiest device in the house to wire and one of the biggest airtime consumers if you don't. Wire it and free up the spectrum for the devices that actually need it.
Cat6 vs Cat6A: What to Actually Pull
The perennial question. Both are cheap; the labor to pull them is the real cost, so people reason "future-proof with Cat6A." Here's my honest take.
Cat6 does 1 Gbps and 2.5/5 Gbps at the full 100-meter channel length, and 10GbE to roughly 55 meters — often less in bundles with alien crosstalk, but a typical residential run is 15–30 meters, nowhere near the limit. Cat6A does 10GbE to the full 100 meters. That's the entire practical difference.
The catch is physical. Cat6A is noticeably thicker (~0.30–0.35" jacket vs ~0.25" for Cat6), stiffer, has a larger bend radius, and is genuinely harder to pull through packed stud bays and harder to terminate — the heavier gauge and tighter pair geometry make keystone punch-downs fussier. Cost difference on materials is modest: quality Cat6 riser (CMR) runs ~$0.15–0.25/ft in a 1,000 ft box; Cat6A more like ~$0.30–0.45/ft. On a 100 ft run, that's a $15–20 difference. The DIY frustration difference is larger than the dollar difference.
My opinion: for a typical home, Cat6 is fine, and I say that as someone who runs 10GbE at home. Your runs are short enough that Cat6 will carry 10GbE anyway, and realistically most rooms will never need more than 2.5 Gbps. Where I do spec Cat6A: the runs to a server closet or NAS location, new construction where the walls are open and pulling difficulty is zero, and anywhere I'm bundling many cables tightly. If you're paying a professional and the quote difference is under $200 for the whole job, sure, take the Cat6A. If you're fishing walls yourself, Cat6 will make you hate your life measurably less.
Whatever you pull: solid-copper, CMR-rated cable from a real brand. No copper-clad aluminum (CCA) — it's non-compliant, drops PoE voltage, and is a fire-code problem. If the price seems too good, it's CCA.
Termination: Keystones, Patch Panels, and Doing It Right
How you terminate matters as much as what you pull. The right pattern is the same one used in every commercial building I've ever audited, shrunk down:
- In-wall cable ends at a keystone jack, punched down with a 110 tool ($10–15). Solid-core cable is designed for punch-down terminations, not crimped RJ45 plugs — field-crimped plugs on solid core are the number one cause of flaky DIY runs. Punch the jack, snap it into a wall plate, done. Keystones are $3–6 each.
- Never put a pre-made coupler or an inline splice inside a wall. Every junction is a future intermittent fault in the one place you can't get to. One continuous run, jack to patch panel.
- At the head end, land everything on a keystone patch panel. A 24-port unloaded keystone panel is $25–50 and takes the same jacks as your wall plates. Label both ends. The patch panel isn't cosmetic — it means the permanent wiring never moves again, and all future changes happen with $3 patch cables.
- Test every run. A $15 continuity tester catches miswires; if you want proof of speed, a laptop and iperf3 at each end settles it.
Wall plate strategy: run two when you think you need one
Use 2-gang boxes (or low-voltage mounting brackets) and pull at least two cables to every location, even where you're sure one is enough. The marginal cost of the second cable is ~$20 in materials while the wall is already open; a second trip into that wall later costs a full drop's labor. Behind the TV alone I routinely land four: streaming box, console, TV itself, soundbar or a small switch. The rule from commercial cabling applies at home — the pull is expensive, the cable is not.
MoCA 2.5: The Middle Ground Nobody Talks About
If your house has coax in the walls — and most North American homes built or cabled since the 1980s do — you have a wired network you're not using. MoCA 2.5 adapters turn that coax into a ~2.5 Gbps backbone. Real-world, I consistently see 1.4–2.3 Gbps between two adapters, with latency in the 3–5 ms range. Not as good as Ethernet, dramatically better than wireless backhaul, and it works through the exact walls that block your WiFi.
The practical details:
- Adapters run ~$70–90 each (goCoax and the Motorola/Translite MM1025 are the ones I deploy), and you need one at each end — so ~$150–180 to link two rooms.
- Install a PoE filter ("point of entry" filter, ~$10) at the splitter where coax enters the house. It reflects MoCA signals back inside — which improves performance — and stops your network traffic from leaking down the drop to the street. As a security guy, I consider this mandatory, not optional.
- Works alongside cable internet on the same coax in most setups; check compatibility if you have a legacy satellite install using the same frequencies.
MoCA is my go-to for the office above the garage, the finished basement with no attic access, the plaster-walled 1950s house where fishing Ethernet means demolition. Two adapters and a filter, twenty minutes, done.
Cost Comparison: Getting a Real Connection to One Room
Here's what it actually costs to get solid connectivity to a single room, four ways:
| Option | Upfront cost | Real throughput | Latency/jitter | Reliability |
|---|---|---|---|---|
| Cat6 drop, DIY | ~$60–100 (cable, keystones, plates, tools amortized) | 1–10 Gbps | <1 ms, zero jitter | Excellent — decades of service life |
| Cat6 drop, pro install | ~$150–350 per drop (market and difficulty dependent) | 1–10 Gbps | <1 ms, zero jitter | Excellent, and warrantied |
| MoCA 2.5 (pair of adapters + PoE filter) | ~$160–190 | 1.4–2.3 Gbps | ~3–5 ms, low jitter | Very good |
| Mesh WiFi node (wireless backhaul) | ~$100–300 per node | 200–800 Mbps, varies by hour | 3–30 ms, variable | Fair — subject to interference and congestion |
Read that table twice. A professionally installed Ethernet drop costs about the same as one nice mesh node, performs 5–10x better under load, and outlives every router you will ever own. The cable in your wall is a 30-year asset. The WiFi gear connected to it turns over every 5–6 years.
New Construction: The Cheapest Moment You'll Ever Have
If you are building or gutting to the studs, this section is worth more than everything above it. Structured wiring before drywall costs a fraction of doing anything after, and I have watched too many $800k new builds go up with a single coax jack per room because nobody asked.
The playbook:
- Home-run everything to one central point. Pick a closet or utility-room wall — cool, dry, with power — and pull every cable from every room back to it. No daisy chains, no splices. A small wall-mount rack or structured wiring enclosure lives here, with your patch panel, switch, and router.
- Run conduit, not just cable. 3/4" or 1" ENT — the corrugated blue "smurf tube" — from the closet to each key location, plus one run to the attic and one to the crawlspace or basement. Conduit means the question "what cable will we need in 2040?" doesn't matter; you pull it then. Smurf tube costs ~$0.50–1.00/ft. It is the best dollar you will spend in the whole build.
- Two Cat6 drops minimum per room; four behind the TV and at the desk. Copper is cheap while walls are open.
- Ceiling drops for access points — one per floor, roughly centered, single Cat6 to a low-voltage bracket. A ceiling-mounted, wired AP will outperform any shelf-top mesh kit you can buy.
- Exterior soffit drops for cameras at the corners and entries, even if cameras are a "someday" item. Pulling to a soffit before insulation takes minutes.
Total materials for a thorough 3-bedroom pre-wire — a 1,000 ft box of Cat6, conduit, boxes, keystones, panel — is roughly $400–700. An electrician or low-voltage contractor doing the pulls during rough-in typically adds $1,000–2,500 depending on scope and market. Compare that against paying $150–350 per drop, forever, after the drywall goes up.
Retrofit: Getting Cable Into a Finished House
Existing house, no open walls. In rough order of preference:
- Single-story with attic: the friendly case. Drill down through top plates from the attic, drop cable inside the wall cavity, fish it out at a low-voltage bracket. A patient amateur with a $30 fish-tape and a strong flashlight can do most rooms. Budget a weekend for 4–6 drops; wear a respirator and go in the morning before the attic hits 130°F — ask me how I know.
- Basement or crawlspace: same trick inverted. Drill up through the bottom plate, fish upward. Often easier than attic work.
- Fishing interior walls between floors: hard mode. Fire blocking, insulation, and mystery framing will fight you. Doable, but this is where paying a pro their $150–350 per drop starts looking like a bargain.
- Exterior conduit: underrated. A run of outdoor-rated cable or PVC conduit along the back of the house, punching in where needed, isn't glamorous but it's honest — I've fed detached garages and far bedrooms this way for $50 in materials. Use outdoor/UV-rated cable and drip loops at every entry.
- When to give up and use MoCA: plaster and lath, brick interior walls, three-story townhomes with no chases, rentals where you can't cut drywall. Don't martyr yourself. Two MoCA adapters get you 90% of the benefit for zero holes.
A real example from last fall: a two-story 1990s colonial, four drops needed — office, two AP ceiling drops, and the family-room TV wall. Attic access covered the upstairs AP and the office; the basement covered the family room; the downstairs-ceiling AP drop was the one ugly fish. About $180 in materials, one long Saturday, plus a MoCA pair (~$170) for a bonus-room dead zone we declined to fight. Total: ~$350 and that house now has better connectivity than most small offices I audit.
The Verdict
The wired vs wireless home network question resolves to a simple rule: wire everything that doesn't move, and give everything that does move the cleanest possible WiFi — which itself depends on wired backhaul. Cameras, APs, NAS, gaming desktop, streaming box: copper. Phones and laptops: radio. Rooms you can't reach: MoCA before mesh.
Pull Cat6 unless your runs are long or your walls are open, terminate on keystones and a patch panel, run two cables everywhere you think you need one, and if you're ever standing in a house without drywall — run the smurf tube. Future you, and possibly future me standing in your attic, will be grateful.
Once the physical layer is solid, the interesting work starts: segmenting cameras and IoT onto their own VLANs and locking the wireless side down properly.