If you have some kind of Integrated Graphics chipset (iGPU), it very likely reserves some system memory as an alternative to using dedicated Video Memory (VRAM).
So the story goes like this: AMD Accelerated Processing Units (APUs) and Intel HD Graphics contain essentially cut-down versions of a graphics card, mounted onto the same package as the CPU. In recent years, both AMD and Intel have started pushing these iGPUs into almost every desktop processor sold. Historically, these chips were on the motherboard, but they started moving them into the CPU. You can still buy processors without them, but the vast majority now have them, even on systems with dedicated graphics cards that are much more powerful.
If you have drivers installed for your iGPU, chances are that the operating system is technically able to use it. In order to use it, though, it needs to reserve some of your system memory to use for storing graphics data (textures, framebuffer, etc). This is because iGPUs don't come with their own dedicated VRAM, because there isn't enough space on the processor die to fit it. Well, usually; the Intel "Iris Pro" graphics chip contains 128 MB of eMMC; but I digress...
On the other hand, if the only graphics chipset your system has is the iGPU, then of course you have no alternative than to accept that some of your system memory is going to be used by this iGPU, or else you will have no graphics display at all.
For some iGPUs, you can configure in the BIOS how much RAM is reserved for the iGPU, usually from a minimum of 64 MB up to a maximum amount of about 1 GB, though the exact figures vary.
In theory, the iGPU isn't the only possible system component that could reserve some system RAM for a special purpose, but it is certainly the most common. I have also heard of some dedicated graphics cards using a hybrid of both system RAM and VRAM, which means that there is a smaller amount of VRAM built into the graphics card itself, but the driver also asks for an amount of system RAM to set aside to supplement the VRAM as additional space for graphics data.
If you want the gory details, look up something like "Unified Memory Architecture" (not to be confused with a similar term such as NUMA, used to distinguish normal systems from distributed systems), or look for the open hardware specifications for Intel's GMA or HD Graphics chipsets. Since the Linux graphics drivers for these chipsets are open source, you can dive as deep as you want by grabbing the spec documents or reading the Mesa or Linux DRM (Direct Rendering Manager) source code. Here you go -- have fun.