Most enthusiast/aftermarket mobos do not include the circuitry to drive a display (the "graphics card"). If you don't have the circuitry, you won't have any use for the connector (but you usually will want to plug in a monitor, so you will usually want to get a separate card which will include the connector).
There are pros and cons to having a separate graphics card:
- A seperate card allows the end user the choice of various performance levels, with this being the only route for users who want the highest 3d performance.
- For the highest performance solutions, power consumption and the resulting temperature rise are an issue, and a seperate card allows the room to install an effective (weighty, expensive, noisy...) cooling solution.
- Traditionally, the 3d performance of motherboard graphics solutions has been very weak (although recently some exceptions have come to market which have a performance that is only a bit weak) and on board solutions tend to share main memory.
- The total installed cost of on board solutions can hit a lower price point, and, compared to a high performance solution, can cost very much less.
- The power consumption of an on board solution can be lower and is likely to be very much lower than a high performance solution. As a result cooling performance is likely to be obtained with less airflow, fewer fans and possibly less noise.
So, in a sense, its horses for courses, but if you need 3d performance, separate video card has advantages. If you do not need 3d performance, then there is a strong argument for one of the better on-board solutions (and many pre-packaged business solutions do exactly that). Some of the AMD/ATI solutions allow you to combine the on board and the separate card approaches, but I have no evidence that this actually works under Linux (if that's an issue).