Network interface cards (NICs)

A network interface card (NIC) is the part of a device that connects it to a network. It is also called a network adapter or network interface. On old desktop PCs it really was a separate card that you plugged in. Today it is usually a small chip built into the main board, but the name stuck.

💡 In simple terms: a NIC is the device's mouth and ears for the network. The computer decides what to say; the NIC actually says it on the wire (or over the air) and listens for anything addressed to it.

Why a device needs a NIC

Inside a computer, data is just numbers in memory. A cable carries electrical pulses, a fibre carries flashes of light, and Wi-Fi carries radio waves. Something has to translate between the two worlds. That is the NIC's job. It solves three problems:

• Signals: it converts bits (1s and 0s) into the right kind of signal for the medium, and back again.
• Addressing: it gives the device a hardware address, the MAC address, so other devices on the same network can send frames to it.
• Filtering: it checks every frame it hears and only passes the ones meant for this device up to the operating system. This saves the computer a lot of work.

Where the NIC sits in a network

Every end device has at least one NIC: laptops, desktops, phones, printers, servers, smart TVs. Network devices have them too: every port on a switch and every interface on a router is really a network interface. Here is a small home network. Each device has a NIC, and each NIC has its own MAC address.

In an office the idea is the same, just with more NICs. Desk PCs have one wired NIC each. Important servers often have two NICs, each cabled to a different switch, so the server stays online if one cable, NIC or switch fails.

What is inside a NIC

You don't need to know the electronics, but knowing the main parts helps you understand what can go wrong.

Types of NIC

Wired and wireless NICs compared

The MAC address: the NIC's hardware address

Every Ethernet and Wi-Fi NIC is given a MAC address when it is made. It is 48 bits long and written as 12 hexadecimal digits, for example 02:00:00:00:00:aa. Windows writes it with dashes (02-00-00-00-00-AA) and Cisco devices with dots (0200.0000.00aa). They are all the same address.

The first half (the OUI, organisationally unique identifier) normally identifies the manufacturer; the second half is a serial number the maker chooses. Because the address is stored in the NIC's chip, it is often called the burned-in address (BIA). The example above starts with 02, which sets the "locally administered" bit: that marks an address chosen by software rather than by a manufacturer, which is exactly what an example needs.

MAC vs. IP: the MAC address belongs to the hardware and is used to deliver frames on the local network. The IP address is configured in software and is used to reach devices on any network. A NIC with no IP address can still send and receive Ethernet frames, which is exactly what happens during DHCP.

⚠️ Modern phones and laptops often use a random MAC address on Wi-Fi networks for privacy, and may use a different one on each network. If a router's list of devices shows a MAC you don't recognise, this is a common reason.

How a NIC sends and receives, step by step

Sending

1. An application, such as a web browser, has data to send.
2. The operating system wraps it in TCP or UDP and IP headers. This is called encapsulation.
3. The driver (software that lets the operating system control the NIC) passes the packet to the NIC.
4. The NIC builds an Ethernet frame: destination MAC, its own MAC as the source, and an error check called the FCS at the end.
5. The PHY sends the frame as signals: electrical pulses on copper, light on fibre, or radio waves on Wi-Fi.

Receiving

1. The NIC hears signals and turns them back into bits.
2. It recalculates the error check. If it doesn't match the FCS in the frame, the frame was damaged and is silently thrown away.
3. It looks at the destination MAC. It keeps the frame only if it is addressed to its own MAC, to the broadcast address ff:ff:ff:ff:ff:ff, or to a multicast group it has joined.
4. It stores the frame in memory and tells the operating system (an interrupt) that something has arrived.
5. The driver collects it and the operating system unwraps it and hands the data to the right application.

💡 Step 3 is why NICs matter so much. On a shared medium a NIC may hear many frames that are not for it. It drops them in hardware, so the computer's processor never has to look at them. Tools like Wireshark can put the NIC into promiscuous mode to keep everything it hears.

What the NIC puts on the wire

On Ethernet, the NIC adds the outer layer around the IP packet: the Ethernet frame. The destination and source MAC addresses and the FCS all come from the NIC.

For example, when the desktop above (02:00:00:00:00:bb, IP 192.168.1.20) sends a web request to a server on the internet, the frame it sends looks like this:

Speed and duplex

A wired NIC and the switch port at the other end must agree on two things before they can talk:

• Speed: how fast bits are sent, such as 100 Mb/s, 1 Gb/s or 10 Gb/s.
• Duplex: whether both sides can send at the same time (full duplex) or must take turns (half duplex).

Normally they agree automatically using auto-negotiation: each side advertises what it can do and they pick the best setting both support. Problems start when one side is set by hand and the other is left on auto. You can read more in Speed and duplex.

Link lights

Most Ethernet ports have one or two small LEDs. They are the fastest check you can do: no software needed. Colours vary by maker, so check the manual, but the usual meaning is:

Drivers

A driver is the software that lets the operating system use a particular NIC. Windows, macOS and Linux include drivers for most common NICs. A missing or broken driver is a common reason a brand-new adapter "does nothing": the hardware is fine, but the operating system cannot talk to it. Updating the driver also fixes many odd Wi-Fi problems, such as dropped connections after sleep.

Which OSI layers a NIC works at

A NIC covers the bottom two layers of the OSI model: Layer 1 (signals) and Layer 2 (frames and MAC addresses). Everything above that, such as IP and TCP, is done by the operating system. Some server NICs can help with higher-layer work (called offloading), but that is an optimisation, not their main job.

Seeing your NIC

Windows

C:\>ipconfig /all
Ethernet adapter Ethernet:

   Connection-specific DNS Suffix  . : home
   Description . . . . . . . . . . . : Intel(R) Ethernet Connection I219-V
   Physical Address. . . . . . . . . : 02-00-00-00-00-BB
   DHCP Enabled. . . . . . . . . . . : Yes
   IPv4 Address. . . . . . . . . . . : 192.168.1.20(Preferred)
   Subnet Mask . . . . . . . . . . . : 255.255.255.0
   Default Gateway . . . . . . . . . : 192.168.1.1
   DHCP Server . . . . . . . . . . . : 192.168.1.1
   DNS Servers . . . . . . . . . . . : 192.168.1.1

Wireless LAN adapter Wi-Fi:

   Media State . . . . . . . . . . . : Media disconnected
   Description . . . . . . . . . . . : Intel(R) Wi-Fi 6 AX201 160MHz
   Physical Address. . . . . . . . . : 02-00-00-00-00-AA

PowerShell shows the link speed and status in one line per NIC:

PS C:\>Get-NetAdapter
Name      InterfaceDescription                ifIndex Status       MacAddress         LinkSpeed
----      --------------------                ------- ------       ----------         ---------
Ethernet  Intel(R) Ethernet Connection I219-V      12 Up           02-00-00-00-00-BB    1 Gbps
Wi-Fi     Intel(R) Wi-Fi 6 AX201 160MHz            18 Disconnected 02-00-00-00-00-AA     0 bps

Linux

$ ip link show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: enp3s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP mode DEFAULT group default qlen 1000
    link/ether 02:00:00:00:00:bb brd ff:ff:ff:ff:ff:ff
3: wlp2s0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN mode DORMANT group default qlen 1000
    link/ether 02:00:00:00:00:aa brd ff:ff:ff:ff:ff:ff

$ sudo ethtool enp3s0
Settings for enp3s0:
	Supported link modes:   10baseT/Half 10baseT/Full
	                        100baseT/Half 100baseT/Full
	                        1000baseT/Full
	Speed: 1000Mb/s
	Duplex: Full
	Auto-negotiation: on
	Port: Twisted Pair
	Link detected: yes

More commands are covered in Essential Windows network commands and Essential Linux network commands.

When a NIC fails

A faulty NIC, cable or setting usually shows up as one of these symptoms:

💡 Work from the bottom up: light, then link status, then speed and duplex, then IP settings. A link light that won't come on is a Layer 1 problem; no amount of IP troubleshooting will fix it. See Layer 1 and 2 problems.

Common mistakes

• Thinking the MAC address is the same as the IP address. They are separate: one is hardware and local, the other is configured and routable.
• Looking at the wrong adapter. ipconfig lists every NIC, including virtual ones from VPNs and virtual machines. Check the one you are actually using.
• Forcing speed or duplex on one side only. Leave both ends on auto, or set both by hand to the same values.
• Blaming the NIC for a slow internet. A 1 Gb/s NIC is rarely the bottleneck for a home internet plan.
• Trusting a MAC to identify a phone. Random Wi-Fi MAC addresses mean one phone can appear as several devices.

Knowledge check

Where to go next

The next devices in this unit are the ones your NIC plugs into: hubs (the old way) and switches (the modern way). To go deeper on the addresses and frames a NIC handles, read MAC addresses and The Ethernet frame.