If you have a endpoint with inbuilt MCU capability like, a tandberg 3000mxp or so, and your multipoint conferencing needs are basic, then yes the endpoints alone will suffice. In the above case you might be able to have 3-4 simultaneous users in one call.
If you need multiple conferences with many more users per conference and also need the MCU resources to be managed by other systems then MCU is required. Cisco IP/VC is one of the options and gels well with other cisco products.