Point-to-point telecommunications generally refers to a connection restricted to two endpoints, usually host computers.
A traditional point-to-point data link is a communications medium with exactly two endpoints and no data or packet formatting. The host computers at either end had to take full responsibility for formatting the data transmitted between them. The connection between the computer and the communications medium was generally implemented through an RS-232 interface, or something similar. Computers in close proximity may be connected by wires directly between their interface cards.
When connected at a distance, each endpoint would be fitted with a modem to convert analog telecommunications signals into a digital data stream. When the connection used a telecommunications provider, the connections were called a dedicated, leased, or private line. The ARPANET used leased lines to provide point-to-point data links between its packet-switching nodes, which were called Interface Message Processors.
More recently (2003), the term point-to-point telecommunications relates to wireless data communications for Internet or Voice over IP via radio frequencies in the multi-gigahertz range. It also includes technologies such as laser for telecommunications but in all cases expects that the transmission medium is line of sight and capable of being fairly tightly beamed from transmitter to receiver.
In hubs and switches, a hub provides a point-to-multipoint (or simply multipoint) circuit which divides the total bandwidth supplied by the hub among each connected client node. A switch on the other hand provides a series of point-to-point circuits, via microsegmentation, which allows each client node to have a dedicated circuit and the added advantage of having full-duplex connections.