bandwidth
Think of a link as a pipe and the data as water. Bandwidth is how wide the pipe is — the maximum rate at which it can move data, measured in bits per second (bps). A link advertised as 100 Mbps can, at best, push 100 million bits across itself each second. It is a ceiling on speed, a property of the link, fixed regardless of whether you are using it.
Bandwidth is measured in bits per second and its multiples: kbps, Mbps, Gbps. (Watch the units: network rates use bits, but file sizes use bytes, and there are 8 bits in a byte — so a 100 Mbps link downloads at most about 12.5 MB per second.) The word comes from signal theory, where bandwidth literally meant the width of a frequency band in hertz; a wider frequency band can carry a higher data rate, which is why the term carried over to mean capacity. In everyday networking, 'bandwidth' just means the link's bit-rate capacity.
The crucial thing to internalize is that bandwidth, throughput, and latency are three different quantities. Bandwidth is the pipe's width (capacity). Throughput is how much water actually flows through (the rate you really achieve, always at most the bandwidth, often less). Latency is how long one drop takes to travel end to end. Adding bandwidth widens the pipe and can raise throughput, but it does NOT shorten the pipe — it cannot reduce latency. A 'fast' connection that means high bandwidth can still feel laggy if latency is high.
Two links rated 1 Gbps have the same bandwidth, but if one crosses the room and the other crosses an ocean, the long one has far higher latency — same pipe width, very different travel time.
Capacity (how wide the pipe is), not speed-of-travel and not the rate you actually get.
More bandwidth does not reduce latency. Bandwidth widens the pipe; latency is set by distance and queuing. Marketing that calls a high-bandwidth link 'fast' conflates two different things.