I'm also bummed by this and I think it's a major sad example of how poor measurables can really pervert a market (granted, in combination with proprietariness and network effects). SSDs have long been lumped together and measured almost entirely by a few high level sticker numbers that companies themselves loved to trumpet because they are high but are primarily relevant to specific DC/server/(certain)workstation workloads, not the common consumer case. Essentially sequential read, write, and then some random r/w, but all performed with high queue depths and a single block size. Typical users aren't exactly running at QD32 all the time. Storage performance is an area where use cases and edge cases matter, and the measures the industry settled on didn't do a good job of conveying that. Only a few places (like Anandtech back in the day) did a solid regularly job of checking 90/95/99/99.9[n] percentile latency, low queue depths, varied blocks, and you still had to dig into it.
So as a result Optane didn't look on the marketing or even in typical reviews any different then competing NAND-based solutions, or heck would even look slower. And lore developed on the internet even amongst tech folks that it "only made a difference on servers". But I was fortunate enough to grab a few to use for core storage and wow is it noticeable, I've built lots of regular SSD big arrays and they look great at simple patterns with large blocks and then one gets into regular workloads and they absolutely tank, with occasional noticeable blips when garbage collecting or the like gets hit. Better than spinning rust overall, but surprisingly not by much sometimes. Whereas Optane is rock solid consistent no matter what.
Intel and Micron were really dumb in how they tried to use it and push it, but I think it's too bad as well it never really got much popular recognition in terms of differences with NAND. A lot focuses on "closer to RAM" but it was also a better SSD.