its actually the sole reason why i ended up paying for plex myself. its not because on ME that i ended up using plex, its moreso everyone else that I want to give my server access to with the least amount of hurdles that made me ultimately go that route.
Dudewitbow
for AI inference (generally hardware agnostic, which is different than AI training, which is heavily CUDA based) favors vram. Digits has to compete against both apples m series chips at max ram capacity, as well as the recently announced AMD Strix Halo devices, which come with up to 128 GB shared ram.
A fully kitted out Strix Halo device with 128gb ram is 2000$, a fully kitted out Mac Studio with 128GB ram is just under 5000$. Nvidia claims it starts at 3000$, so finding out the 128gb cost is still in the air.
besides the lower bar of entry due to being free, Midias research has shown that the younger generation prefers online multiplayer, and as you grow older, you start to favor single player games more.
i usually refer to it like email but all the emails are set to public. it doesnt matter which domain you use, you basically see most of it. then explain it like, you dont create a yahoo email to email only yahoo email, or a gmail account to only gmail users. they all share a public api and send shit to eachother with it.
the advantage for pci-e generations isn't the gpu side, but SSD side(USB too). CPUs by design, have a limited amount of PCI-E lanes they can distribute. Modern SSD's primarily use 4x pci-e lanes, but if you have a generation newer, you can allocate half as many lanes to achieve the same speed.
Since at the moment, the value of even faster SSD's are useless, what ACTUALLY becomes off value is having multiple m.2 slots for expansion. for example, instead of having 1 m.2 running pci-e 4.0x4lanes, you can achieve the same speed at pci-e 5.0x2lanes, or pci-e 6.0x1lane. Using less lanes on m.2 slots means you can allocate more bandwidth to other components(more m.2, more USB ports, etc).
this for eaxmple was the reason why experimental gpus like the Asus Dual RTX 4060ti existed, as its a 8 lane gpu, which had an on board to use the other lanes as ssd storage on motherboards that supported bifurcation.