It started with a simple observation: an abnormally low connection speed. Comparing the network settings of my laptop with those of a cloud instance (Shadow), one detail jumped out — where the instance showed a virtual network card at 10 Gbps, my PC was capped at 100 Mbps over Ethernet. A hundred times slower. Worth looking into.

The suspect: a USB-C adapter sold as "2.5 Gigabit"

My PC — an ASUS Vivobook 15 — has no built-in Ethernet port, like a lot of thin laptops today. So I was going through a USB-C to RJ45 adapter, bought on Cdiscount, newer than an older model I'd used before. On the product page and on the casing itself: a "2.5GBASE-T" marking, a promise of 2.5 Gigabit throughput, USB-C connector, compatible with "iPhone, MacBook, iPad, Surface Go, Galaxy S24 Ultra." Price: €10.39. Average rating: 2.9/5 from 9 reviews — a number I hadn't paid attention to at the time.

But between the label and reality, there was a world of difference.

The technical check: what Windows actually sees

Rather than trusting the marketing, I wanted to verify what the system actually recognized. A network adapter, whatever its casing, identifies itself to Windows through a pair of hardware IDs (Vendor ID / Product ID) and through the capabilities its driver exposes.

Output of the Get-NetAdapter command:

Name                 : Ethernet 3
InterfaceDescription : USB2.0 Ethernet Adapter
LinkSpeed            : 100 Mbps

The driver name speaks for itself: "USB 2.0 Ethernet Adapter", not a single mention of 2.5G. And the hardware ID confirms the picture:

USB\VID_1A86&PID_E396 → QinHeng Electronics (wch.cn)

To be completely sure it wasn't just a throttled setting somewhere, I queried the possible values for this adapter's link speed:

ValidDisplayValues : {Auto, 10MB Half, 10MB Full, 100MB Half, 100MB Full}

No 1000 Mbps option. No 2,500 Mbps option. The chip itself — a QinHeng/WCH, typically used in entry-level adapters costing a few euros — is physically incapable of going beyond 100 Mbps. It's not a misconfigured driver or a setting to unlock: the silicon caps out there, full stop.

A genuine 2.5 Gigabit chipset (a Realtek RTL8156, for example, the market standard for this kind of product) would have shown up in Windows with speeds available up to 2,500 Mbps, and a Realtek ID — not QinHeng.

A non-conformity, not just a "disappointing product"

So this isn't a matter of perceived quality or performance "below expectations": the delivered product simply doesn't match its description. The casing advertises 2.5 Gigabit, the chip inside only does 100 Mbps — 25 times less than what's being sold.

I filed a complaint with the seller through the Cdiscount form, backed by these precise technical details (hardware ID, driver name, available link speeds) and invoking the legal warranty of conformity (Article L217-4 et seq. of the French Consumer Code). Reply received the same day: refund accepted.

What to keep in mind before buying this kind of adapter

A few markers to avoid getting caught out:

The average rating is a signal. 2.9/5 from just 9 reviews for something as simple as a network adapter is a red flag we tend to ignore when the price is low.

USB-C is no guarantee of speed. The connector says nothing about the chip behind it. You can perfectly well have a USB-C port wired to a 100 Mbps Fast Ethernet chip.

The chipset name matters more than the marketing name. For reliable 2.5 Gigabit, the references that keep coming up are the Realtek RTL8156 (or RTL8156B) and, for plain real Gigabit, the RTL8153 or ASIX AX88179. A serious product page usually names the chip it uses; the total absence of that information, on a generic "compatible with everything" product, is suspicious.

You can check for yourself, after purchase. On Windows, a Get-NetAdapter | Select Name, InterfaceDescription, LinkSpeed in PowerShell is enough to see what the system actually detects. If the description mentions "USB 2.0" for a product sold as 2.5G or Gigabit, or if the listed manufacturer isn't a recognized name (Realtek, ASIX, RTL...), there's a good chance the marketing promise and the hardware reality don't line up.

The Ethernet cable matters too, but that wasn't the problem here. An underspec cable (Cat 5, for instance, limited to 100 Mbps) can also throttle a link that could otherwise run at Gigabit or more. In my case, the cable used is a Cat 8 — easily capable of handling 40 Gbps — so no doubt possible: the limit really came from the adapter's chip, not the cable. But it's the first thing to check if chipset and cable are both potentially at fault: better to use a Cat 6 or higher cable so you never have to wonder.

What now?

While I look for a reliable replacement — this time with an explicitly identified chipset — I've temporarily switched to the 5 GHz Wi-Fi on my Livebox 7, which on its own delivers six times more throughput (600 Mbps) than that famous "2.5 Gigabit" adapter. Ironic, but effective.