In the desktop market, I don't think that there will be a lot of change. AMD has the R5 and R3 G series chips that will be a step up from Intel in the i3/i5 range, but won't really make a noticeable difference in Nvidia's current market on the bottom end (they don't really compete there, save for certain specialty uses). In the mid level market, you're still targeting PCIe cards, and Nvidia is well covered in that market. This also includes the high end. The AMD APUs will offer some flexibility in the low end of the market for OEMs and custom builders, but that was never a focus for NVidia in the first place.
In the laptop market, you've got the sub $350 market for low end thin and light or traditional form factor units. There, Intel and AMD compete with bottom of the stack parts, like the A series APUs, Atom core celerons and pentiums, and the occasional end of life i3. AMD will have a product refresh there with the R3-U series, but the performance there will be highly hobbled by single channel memory, vastly fewer CUs, etc. In the $350-$600 segment, Intel has a bunch of i3 and i5 products, a few with very low end Nvidia products. AMD has higher end APUs in the A series. The R3-U will be present here, as well as a few R5-U products. The R5-U will be able to replace a good chunk of intel-Nvidia low end dGPU products, but not for a while as the eco-system there is built up and the channel gets filled. This is at least a year from being an issue. Nvidia can derail a lot of that by just lowering the price or increasing the performance of their products in that segment as those products are already engineered and built and there is momentum in that space. As you move up market, there is still room between $600-$900 for Nvidia to sell a lot of product into. AMD has the RX550 mobile for that space, but Nvidia can decide to offer more value there at will. The R7-U APU can't get performance that high in any form and would only be used where compactness is preferred and bleeding edge GPU performance isn't a concern. While that is still a non-trivial portion of the market, it's not going to have a lot of effect. Above that, you're getting into low volume, high end products. There, Nvidia will compete with the Vega-M HBM dGPU products and Intel KL-G. For larger form factors, Nvidia will still have better top end performance. It's the smaller form factors where the improved power profiles and smaller physical size of AMDs products will have an impact. Imagine having high end GPU performance in something as thin and light as a surface book or Mac Powerbook? Will buyers choose to spend a bit more to get what is essentially a large thin and light that has the GPU muscle to play AAA titles at 2K or 4K? I'm sure some will, but, I don't know if the whole market would shift in that direction. Nvidia can also provide another performance/power bump for their existing mobile high end GPU product to make it more attractive.
Where AMD could have the biggest impact on the market is by making the R7-U available in the $500-$700 laptop price range with nicely specced laptops that can basically do most everything that most users need. It needs to have a thunderbolt port that can use an external GPU enclosure that can be used by a user that might want to have the best available game or professional graphics performance only sometimes, but not necessarily need it while on the go (or can tolerate passable, but not spectacular performance while mobile). Not everyone needs a thin and light, but a decently compact laptop that offers that kind of flexibility in a reasonable price range would get a LOT of buyers.