Good for you, you caught on that I was talking about the Alexa 65. I'm assuming you also made one google search to see that 500T is the fastest film stock Kodak offers. What you're missing is that 500T 65mm can easily expose at 1200iso or further, either by pushing, flashing, bleach bypass, or a combination of those processes. The increased grain is negligible.

The workflow from capture, to editorial, to presentation for 65mm/70mm is established and as easy as you want it to be. The Alexa 65 on the other hand, fills a terabyte with every 20 minutes of it's 6.5K Arriraw. It's some poor soul's JOB to transcode and back all that data up in order to send it to a distrobution system that will only get you a digital image up to 4K if you're lucky.

65mm film is cheaper than the digital alternative.

You don't have to fully edit on film to finish on film. What you do is scan your film, edit in your Non Linear Editor, then do what's called a Negative Cut, cutting the negative to match the cut you made in your NLE, then you can do a dry gate or wet gate print, even a 70mm blow up. No digital step getting into the final image. This was done for all of Christopher Nolans films after Following.

Cinelicious in Hollywood has scan services in 2K and 4K for super 16 although they can be somewhat expensive.

http://cinelicious.tv/

In my experience when I scanned my super 16 footage at 2K, the files were 2048 by 1296 pixels which is taller than most files because super 16 is 1.66:1 aspect ratio. I've also had the scans come back already in 16:9 when I scanned in HD. What I'm doing for release right now is cropping the top and the bottom of it to make the image 1.85:1. I downloaded the matte frame from Vashi Visuals. My reasons for 1.85:1 are generally for retaining resolution. It has also been useful in being able to reframe slightly in post as I can slide the image up and down without any real problems unless there's hair in the gate. Also as my plan is for projecting it, most projectors project one of two of the mainstream cinematic aspect ratios, 1.85:1 or 2.39:1. You always crop out pixels, never set it to "fit to" 16x9, preffer "scale to horizontally" if you're going for 16x9. For a 2K DCP, set your output resolution to 1998 by 1080 and scale horizontally to fit. If you want to retain 2048 across for some other file output, set your vertical resolution to 1107 pixels.

The photosites on the Alexa 65 sensor are actually the same exact size as on the normal Alexa. It's essentially three Alexa sensors put together so it wouldn't have more dynamic range unless the electronics/software inside are better at utilizing the data coming off the sensor.

Where the sensor isn't natively faster than the the normal Alexa, I would bet that noise isn't so apparent since it's now at 6.5k. It would seem finer, a lot like film grain on 65mm film would appear compared to 35mm. So as long as they don't actually plan to show it on a screen 3 times bigger than normal, the noise would allow them to push the sensor further which is a big help with large format and large format lenses in that you can stop down as a courtesy to the focus puller (also that most large format lenses I've seen only open up to t3.5 or thereabouts at the widest) So it's faster and it's slower.