Nikon 24-70 2.8 VR is no where as good as their non VR lens. Even Sigmas 24-70 OS is not as good as Canons L non IS lens. So is the compromise of reduced IQ worth to photographers over non IS lens with better IQ?
"...Nikon has made a design choice with the new lens, and the design choice wasn’t ‘let’s make it look great for the bench testers’. They’ve given up the absolute best center resolution in exchange for good resolution across the entire image field. So in the center 1/3 of the image, the old lens had better MTF results, but across the remainder of the field the newer lens is far superior. Not just that the resolution is better, but there is very little astigmatism, which the old lens had in spades." [https://www.lensrentals.com/blog/2015/10/nikon-24-70mm-f2-8-ed-af-s-vr-sharpness-optical-bench-testing/]
No quite as clear cut as your sweeping statement presents. Nikon made a choice with the design of the new VR lens' optical characteristics, just one of the many trade-offs they had to make, including increased size, weight and price. Indeed, I seem to remember that size and weight was the reason that Canon gave for not including IS in the the EF 24-70 f/2.8L II, rather than concerns over resolution (following feedback from professional customers, who have the lens hanging around their neck all day). The Sigma lens is a bit of a dud (though not so bad at the long end), so I'm not quite sure what their design choices were, but the earlier (pre-Art) non-OS version wasn't any better. Tamron's 24-70 f/2.8 VC is also superior to the non-VR Nikon away from dead centre.
To summarise, I think that it is too simple to make the statement that adding image stabilisation will inevitably compromise resolution. For sure, adding another variable to the len's design isn't going to make life any easier for the designer, but the outcome will depend upon the balance of priorities that has been set (probably by the marketing team).