The expected answer is that Ethernet is faster. On a Starlink connection that is usually not what changes. The peak number on a speed test tends to stay where it was, and the thing that improves is everything around it.
That distinction is worth understanding before spending money on cable, because the two problems have different fixes.
Quick answer
| Question | Answer |
|---|---|
| Does Ethernet raise Starlink’s peak speed? | Rarely. The satellite link sets that, not the local medium |
| What does a wired link improve? | Consistency, jitter, latency under load, and immunity to interference |
| Which is faster in isolation? | A wired link, by a wide margin, at distances a home needs |
| Is the local link ever the bottleneck? | Occasionally, and then it is worth fixing |
| Does the Mini need an adapter for Ethernet? | No. It has one built-in port |
| What degrades Wi-Fi first? | Distance, walls, neighboring networks, and the number of clients competing for air time |
| What degrades Ethernet? | Almost nothing at these distances |
| How should this be tested? | One variable at a time, same server, comparing averages rather than peaks |
The local link is not where Starlink’s speed comes from
A Starlink connection has three stages: the radio link to the satellite, the path through the ground station and the wider network, and the local link between the terminal and your devices.
The first two set the throughput you see. The third is quicker than both of them, and by a wide margin — a wired link in this class carries far more than a residential satellite service delivers, and the Wi-Fi radio in the terminal does too when conditions are good.
That is why swapping a Wi-Fi connection for a cable often leaves the headline speed test number almost unchanged. The bottleneck is upstream, and no cable on your property can reach it.
What the wired link actually changes
If throughput is not the difference, why does a wired connection so often feel better? Four reasons, and all of them are about variability rather than capacity.
Air time is shared. Wi-Fi is a shared medium. Every device on the network, and every neighboring network on the same channel, competes for the same air time. A cable is dedicated to one conversation.
Interference is asymmetric. Radio links degrade from walls, floors, appliances, and other people’s networks. Twisted-pair cable is largely immune to all of it, which is the entire reason the pairs are twisted in the first place. Our explainer on how Ethernet cables work covers that mechanism.
Retries add jitter. A marginal radio link does not fail cleanly. It retransmits, which shows up as latency spikes and stutter rather than as a lower average. A wired link that is within its specification does not behave that way.
Contention multiplies. One device on a wired link has the medium to itself. Ten devices on Wi-Fi share it, and the effect is not linear.
| Dimension | Wi-Fi | Wired |
|---|---|---|
| Medium | Shared air time | A dedicated twisted pair |
| Interference | Walls, floors, appliances, neighboring networks | Largely immune by construction |
| Behavior under load | Retries and latency spikes | Stable while inside specification |
| Scaling | Contention rises with every added device | One device per port |
| Mobility | Full | None |
| Practical reach | Room to room, degraded by obstacles | The 100-meter channel |
Put together, that is the honest case for wiring a Starlink installation: not a higher ceiling, but a floor that does not move.
Where Wi-Fi genuinely wins
None of this makes Wi-Fi the wrong choice. It is the right one for phones, tablets, laptops that move, and anywhere a cable is impractical. The argument for a wired link applies to the devices that stay put and care about consistency: a desktop, a console, a work machine on video calls, a network video recorder, an access point backhaul.
Note the last one. Using a cable to feed a Wi-Fi access point takes the long, contended hop out of the path without giving up wireless coverage at the far end.
The Mini changes the arithmetic
The Mini is the Starlink terminal that carries an Ethernet LAN port on the unit itself, which means a wired device can connect directly at the terminal instead of through an indoor router. Its power arrives on a separate DC cable, so the data path and the power path are independent. Article on the Mini’s Ethernet port covers what that port does and what has to be true for it to keep the unit’s IP67 rating.
Because the Mini is often outdoors, the practical wiring question is usually about getting a reliable lead from the terminal to wherever the device sits, rather than about which category to buy. At domestic distances, a factory-terminated Cat6 patch lead is more than enough, and the standard 100-meter channel is far beyond anything a house or a small site will use.
How to test this without fooling yourself
Most "Wi-Fi versus Ethernet" comparisons are measured badly. Four rules fix that.
One variable at a time. Same device, same server, same time of day, only the connection changes. Comparing a wired test at 3 a.m. with a wireless test at 8 p.m. measures the network, not the medium.
Compare averages and jitter, not peaks. A single high reading proves nothing. Run the test repeatedly and look at the spread. The wired link’s advantage lives in the spread.
Test under load, not at idle. Run the test while something else is using the connection. That is when contention and retries show up, and it is closer to how the link is actually used.
Do not use a speed test to judge a cable. A throughput test tells you about the whole path. It cannot tell you whether your terminations are good, and a link can pass a speed test and still be out of specification. That is what a certification tester is for, and our guide to test report validity explains what the resulting report does and does not prove.
Scenarios
A home office where video calls stutter while the speed test reads fine. The peak number was never the problem. Wire the work machine, leave everything else on Wi-Fi, and the jitter goes away.
A client that "feels slow" on Wi-Fi three rooms away from the terminal. Two 5 GHz walls is often enough to halve usable throughput. Either move the terminal, add a wired backhaul to an access point, or accept the wireless path for what it is.
A Mini on a roof with a desktop inside. The wired run does the work, and the DC side stays short by the power source. Wire the device that matters and stop paying for wireless on it.
A shed office 60 meters from the house. A single Cat6 run carries it, and a point-to-point wireless bridge is the alternative when trenching is not an option. The Starlink cable guide covers how the terminal’s own cable behaves over distance, which is a different question from how a LAN run behaves.
FAQ
Is Ethernet faster than Wi-Fi on Starlink? A wired link is faster in isolation, but on a Starlink connection the service itself sets the throughput, so the peak number usually does not change much. What changes is consistency.
Will a cable make my Starlink faster? It will not raise the service’s ceiling. It removes variability between the terminal and your device, which is what most people are actually complaining about.
Does the Mini need an Ethernet adapter? No. The Mini has one latching Ethernet LAN port built in.
Should I use Cat5e or Cat6 for a Starlink LAN run? Either carries the service easily at domestic distances. Cat6 is the better choice for new work because it keeps the 2.5G and 5G options open.
How do I know if Wi-Fi is my problem? Test the same device wired and wireless, back to back, on the same server. If the wired average is steadier, Wi-Fi was your problem.
Does a longer patch cable reduce speed? It consumes channel budget rather than speed directly, and at domestic lengths it will not matter. It becomes relevant only on runs approaching the specification limits.
Sources
- Starlink — Mini Specification Sheet: Wi-Fi 802.11a/b/g/n/ac, dual-band 3×3 MU-MIMO; one latching Ethernet LAN port with Starlink Plug; separate DC power path, input rated 12–48 V at 60 W
Starlink is a trademark of SpaceX. ZOGUO Cable is an independent manufacturer, not affiliated with, authorized by, or endorsed by SpaceX.
Related reading on this site: