Why Your Gigabit Ethernet Is Stuck at 100 Mbps

A single broken copper strand inside an Ethernet cable will instantly slash a gigabit connection to one-tenth of its capacity. The drop is not gradual. It is a hard fall from 1,000 Mbps to 100 Mbps, triggered by the physical architecture of network negotiation rather than software throttling or ISP limits.

Gigabit Ethernet (1000BASE-T) requires all four twisted pairs of wire within a standard Cat5e or Cat6 cable to transmit data simultaneously. Fast Ethernet (100BASE-TX), standardized in 1995, uses only two. When a router and a PC attempt to establish a link, they rely on auto-negotiation protocols defined by IEEE 802.3. If even one pin in the RJ45 connector fails to make contact, or if a single wire pair is severed, the hardware immediately abandons the gigabit handshake. It falls back to the highest common denominator that functions over two pairs. That ceiling is exactly 100 Mbps.

This fallback mechanism traps users who assume their internet service provider is underdelivering. Checking the link speed directly exposes where the bottleneck sits. On Windows 11, navigating to Settings > Network & internet > Ethernet reveals the negotiated Receive/Transmit rate. A faster route on both Windows 10 and 11 involves pressing Windows + R, typing ncpa.cpl, right-clicking the active adapter, and selecting Status. macOS users find the same hardware-level metric under System Settings > Network > Ethernet > Details > Hardware.

That number represents the local segment between the network interface card and the switch. It has nothing to do with wide-area internet bandwidth.

Routers are frequent culprits in these unexpected downgrades. Budget switches and older wireless gateways often ship with 10/100 Mbps LAN ports despite advertising high-speed Wi-Fi capabilities. Plugging a flawless Cat6 cable into a 100 Mbps port guarantees a 100 Mbps link. USB Ethernet adapters present another hidden chokepoint; cheap dongles frequently max out at Fast Ethernet speeds regardless of the host machine’s internal bus capabilities.

Printed jacket markings offer limited reassurance. While Cat5e and Cat6 cables support gigabit transmission, the category rating does not guarantee performance across distance. A degraded cable might successfully negotiate 1 Gbps across a three-foot patch but fail over a 50-foot run through a wall due to crosstalk or attenuation. Older Cat5 installations, largely phased out after 2001, sometimes carry gigabit traffic over short distances, but lack the strict twist-rate tolerances required for reliable high-speed signaling.

Isolating the failure requires eliminating variables. Reseating connectors clears oxidation-related contact failures. Swapping in a known-good, short patch cable using the exact same ports determines whether the fault lies in the cable plant or the networking hardware. If the link speed jumps to 1 Gbps with the replacement cable, the original wiring is compromised.

Software tools bypass the ISP entirely to measure raw local throughput. iPerf3 runs a server instance on one wired machine (iperf3 -s) and a client on another (iperf3 -c SERVER_IP) to stress-test the physical layer. Disabling Wi-Fi during this test prevents the operating system from silently routing traffic over wireless bands. Standard multimeters verify electrical continuity but cannot measure signal degradation or bandwidth capacity. Determining actual data-carrying capability requires qualification testers that analyze frequency response up to 250 MHz for Cat6 infrastructure.