Not sure that the link adequately makes its case. The number of people that actually care about Intel Management Engine is extremely small. Intel's real problem is that their products have struggled to be competitive in the marketplace on a price and performance level.
I think this sentence, in isolation, actually kind of nails the problem. I realize this isn’t what you and the article you’re replying to have in mind, but: no one cares enough about ME to buy it. After years of development, Intel has a solution to server management (and client management?) that is arguably technically superior to the crappy, power hungry chips from ASPEED and the management stacks from the likes of Dell and HPE that, until very recently, were unbelievably bad.
But Intel completely failed to execute! ME should have been a headline feature, made friendly to the point where people would buy Intel systems because of ME. Anyone considering AMD or Ampere or POWER would be annoyed because their organization heavily used ME and Intel’s competitors didn’t have it!
But instead, Intel hid the good parts (iAMT) behind annoying SKUs, obnoxious and secretive software and weird proprietary lock-ins. And the result is: no one cares about ME!
(Intel has messed up most of its recent features in almost exactly the same way.)
Saw AMT once about 15 years ago — available for free on an Intel banded desktop motherboard.
Now I can’t even pay money to use it on my modern PCs…
But as you say, Intel hid it away. Like a guilty secret. They somehow forgot to tell even their most Evil™ customers about the Evil Inside™. Which is jolly suspicious. And they've hidden away other "features" that trash security. It may be the number of people who "actually care about Intel Management Engine is extremely small", but those who do matter and they make decisions. Yes performance is important but security factors are going to be gaining more and more influence now.
So Intel were told time and time again that people want legible, well documented products with no surprises, and they ignored that.
But I agree, that's noise. There are much bigger problems with their product line:
* Adding and removing features (especially AVX, but also the whole market-segmentation nonsense with ECC memory) * The marginal and in some cases negative generation-over-generation performance boosts. * The unlocked wattage situation which screams "we're pushing an end-of-the-rope product that we can only get to decent clocks by feeding it an entire mains circuit". (See: FX-9590, Pentium 4 Extreme Edition)
Intel might be able to say "Today's Glysophate Lake CPU is clearly better than the Haswell you bought 10 years ago", but there will be a lot of mumbling if you ask "how much better is it than last year's Paraquat Lake". That doesn't inspire confidence, especially when they have traditionally been the premium brand.
That's not super true. AMD has had good outlier performance in gaming with the X3Ds for a while, but for regular use something like a 13400f has been as good of a deal as any AMD chip.
Arrow lake seems to have definitely made the situation worse for intel though.
Regaining the faith of users will take years if the best Intel has to say is "this SKU is unaffected by this flaw"; less years if they openly switch to more trusted TSMC processes and/or ARM designs.
But AMD has their own version with the PSP.
So we're stuck on Bulldozer hoping for Risc-V or something to mature enough before Bulldozer is completely obsolete.
(With Apple obviously never having been an option to start with, and Microsoft keeping to push for the likes of TPM and Pluton, thankfully still optional for now.)