Your situation is "D: None of the above".
-- Your GPU is spiking in its utilization, from the way you're describing it. Even the best GPU might hit the occasional situation where it can spike to a very high utilization level (which "96-99%" qualifies for), but the rest of time will be OK at a much lower utilization (& 70% is much lower than 96% in these situations).
-- You said your CPU is staying at 50-60% utilization consistently. That's middle-of-the-road. It's not "low" (which would be 30-40% or lower), & it's definitely not high (that would be 80% or higher).
-- Since you're neither high nor low on your CPU utilization, & your GPU is not consistently on high utilization, you're not really seeing a "bottleneck" in the traditional sense.
That doesn't mean there might not be something else going on, or that there's maybe not some sort of other "glitch". One of the issues (that I don't know has ever really been fixed, if it even can be) with the GTX 970 was the whole "3.5GB/0.5GB" fiasco: although technically equipped with 4GB of VRAM, nVidia manufactured them with 3.5GB of "high-speed" VRAM & the remaining 0.5GB on a "low-speed" bus (it runs about 7 times slower). So, any time the GPU has to use more than 3.5GB of VRAM, you start to see some performance issues.
What you should do is see if there's any particular pattern to the glitches you're seeing. For example, see if your monitoring software can tell how much VRAM is in use, & see if the VRAM exceeds 3.5GB whenever the utilization spikes up.