I think that this was part of the larger trend of IBM losing the leadership of the “PC”, especially after the PS/2 and the MCA debacle (although the PS/2 line did introduce many new features which became de facto PC standards).
Looking at graphics specifically, after the VGA, IBM developed the 8514/A and then the XGA. For most consumers these had several disadvantages: they were expensive, they targeted high-end 1024×768 monitors (fixed-frequency in some cases), and they are accelerators. They were great for CAD and similar tools which could use specific drivers and benefited from the accelerated functions, but they didn’t have any significant advantage for most PC users. The lack of support for 800×600 might have mattered too: while higher-end 13-14” monitors of the era supported 1024×768, in practice that was too fine for most users.
I’m not sure that it’s accurate to say that NEC wrested PC graphics standards away from IBM. For a start IBM wasn’t holding on to them ;-). But NEC didn’t really end up writing the next chapter either... In the late 80s and early 90s, PC graphics was a mess of competing implementations, and any software which wanted to work on a variety of graphics adapters, using their capabilities beyond VGA, ended up having to support them all separately. See Fractint for one example of this. In the late 80s, there wasn’t much use for anything beyond VGA anyway, for most users:
After VGA, I’d say the next significant PC graphics standard was Windows 3 drivers; that’s what brought unification back. NEC’s SuperVGA “standard” did help by providing a fall-back, but that was only really significant with Windows 3 anyway (and its SuperVGA driver). Most graphics adapters provided their own, specific Windows 3 drivers... In terms of graphics standards, VBE only made an impact with version 1.2 in 1991, and even then, it ultimately ended up being a common API for games running on top of SciTech’s drivers.
That’s not to say that NEC’s endeavours weren’t significant; the VESA Local Bus was a huge improvement, but that’s not a graphics standard, and it was developed later (1992). It does however explain why VESA itself ended up being successful. As for why NEC started VESA to drive and publish these standards, it’s probably because it was too small a player to change the industry on its own, and consortia were popular at the time (see also EISA, in 1988).