I think a digital X-pan is unlikely in the near future, even as a niche product, for a number of reasons, and wouldn't be the same success the film version was.
Firstly, with digital sensors and advances in lens design and manufacturing, it's no longer as necessary.
a) Its prime application would be landscapes and other static, panoramic scenes. Stitching is easy with digital. Even if there are moving components, it's usually possible to isolate each moving component into one individual frame to avoid ghosting when stitching. You no longer need to capture everything in a single frame.
b) Advances in manufacturing make modern lenses much sharper than older lenses could ever be. It's also easier to make a small, sharp lens than a large, equally-sharp lens. A modern full-frame DSLR lens can resolve as much detail as an older lens for a 4x5 camera. An image on a smaller sensor may need to be magnified more than an X-pan image for the same final size, but the smaller lens will probably also be sharper to start with, so it can be magnified more.
c) CMOS sensors are much more sensitive than colour film ever was. You no longer need a 24x65mm sensor just to reduce the noise in the final print. Furthermore, with digital stacking of multiple frames, you can reduce the contribution of random noise to essentially zero.
Secondly, introducing a new system would come with a few big hurdles.
a) Sensor fab. It's no longer a case of drawing 135-format film through the camera two frames at a time. You'd need a new fab plant to handle the sensor - one that could handle much larger wafers. And you'd also need a source of flawless, much-larger wafers to go with it. This may be much easier once they move beyond silicon semiconductors.
b) Lens selection. You'd need to design a new set of lenses from scratch.
That said, I'd still like to see one at some point. A 24x72mm sensor (e.g. two A7rII sensors stuck side-by-side and calibrated to match each other's output) could easily match 617-format film output, both in aspect ratio and image quality. Movable/tiltable sensor for tilt/shift capability and micro-shifting multiple exposure modes, a mirrorless design and rear-mounted, tiltable display for those shots down at ground level or on a tripod above head height. Since this would be a non-action camera with limited intended uses, you'd also only need a few lenses. Not super-fast (it's a landscape camera, after all) but super-sharp. I'd start with a few zooms - 20-50mm/5.6, 50-135/5.6 and 135-400/5.6. Don't really need primes if you're shooting at narrow apertures anyway. Throw in 1.4x and 2x TCs and you'd be covered for every possible landscape from 10-400mm 35mm equivalent. The kit wouldn't even be that large or heavy.
I, too, would welcome a digital XPan!
But besides the uncommon sensor size and ratio, which would drive prices up, I fear the lenses would have to be much larger than the original Fuji lenses, which were using symmetric designs. Digital sensors would have trouble coping with that and Otus-sized telecentric lenses would probably have to be used instead .
With the rumoured 70+ MP full frame sensors on their way, there might be enough resolution to simply crop to a 24:65 ratio when stitching is not possible. That would be the "UPan" (for underpan ) way of doing things. Weak corners of the lens could be ignored.
That being said, I would still miss the ability to see the picture in a 24:65 ratio in the EVF to ease composition. I wonder why it is not possible to define user selected ratios in recent cameras featuring EVFs? Native 3:2 and video 16:9 are standard, with 4:3 and 1:1 being also offered by the most adventurous camera manufacturers. But how about 24:65?
Cheers,
Fabien