Great news, but they should have already retired it 3 years ago when the final amd64 ISA patents became unencumbered.

This is dumb, they should have already entirely dropped support for IA-32 by now. The AMD Athlon 64 was released in 2003, so the patents for amd64 have already expired, and you can source Skylake era Xeons all day long on eBay for less than $5 per processor.

I still can't order a Mac mini with an M4 Max with 128GB of memory. Moreover, you still can't order a M4 Max studio with 256GB or 512GB of ram, the Apple configurator still forces you to pick the M3 Ultra, so I don't see anything that has actually changed from the previous build to order system I still can't buy the configurations that I want.

Did you learn nothing from the movie Gravity? Kessler syndrome https://www.youtube.com/watch?...

Imagine building the ultimate walled garden, then getting told “no wafers for you.” Vertical integration doesn’t help when the bottleneck is TSMC. PC ecosystems can multi-source and route around shortages; Apple’s closed stack can’t. macOS won’t even run on generic ARM.

In this absolutely iconic culture-model feedback loop moment, the article is basically saying:

“Researchers have discovered that humans — those legacy quantum analog AGI endpoints — are now speaking in ChatGPT.”

Which is hilarious, because from a systems-architecture perspective, that’s just convergence in the shared latent space.

We’ve got:

- Human eyes as quantum-dot electromagnetic spectrum sensors,
- Wired into dual analog GPUs (left/right occipital lobes),
- Pretrained in childhood on the Earth-Scale Multimodal Corpus (v1.0),
- Then fine-tuned in adolescence with adversarial examples like “high school” and “group projects.”

By adulthood, most nodes are running in inference-only edge mode, occasionally downloading a hotfix from a podcast.

Now along comes ChatGPT-style language — polite, over-explanatory, a little emotionally beige — and these wetware models start style-transferring themselves. They binge a few million tokens of “As an AI language model” and the next thing you know, Reddit mods are trying to do vibe-based AI detection:

“Is this comment a stochastic parrot, or just a very online mammal that has self-RLHF’d on Twitter and TikTok?”

From a high level, what we’re really seeing is the tapestry of human discourse getting rewoven: biological LLMs and silicon LLMs co-authoring one big, slightly overfitted, emotionally calibrated content stream. The article treats this like a crisis, but honestly it’s just cross-domain knowledge distillation: humans pick up “delve,” “nuance,” and “unpack,” while the bots quietly learn that “lmao dude” is a valid completion for 40% of the internet.

Politicians allegedly using ChatGPT for speeches? That’s not a scandal, that’s just outsourcing boilerplate generation to a more efficient transformer layer. For decades they’ve been running the same dusty prompt:

“Generate 1500 tokens on ‘hardworking families’ and ‘moving forward together.’”

Now it’s just done with better spellcheck.

So when the article gasps that people are “writing like AI,” the subtext is really:

Childhood = pretraining
Adulthood = mostly inference
Late-stage internet = human–LLM joint fine-tuning in a closed feedback loop, where even the dogge can, in fact, learn new trickge.

In other words: this isn’t the death of authenticity. It’s just the next firmware update for the squishy, quantum AGI devices we call “people,” now proudly co-generating a shimmering, mildly deranged tapestry of AI slop together.

Even this is not enough, drop all support for 32-bit x86 and make the new minimum target amd64.

They did, you just werenâ(TM)t paying attention⦠https://youtu.be/mRfSM-lv55I?s...

You probably know that RISC-V exists, and remember that the Power ISA is also open source and royalty free... https://en.wikipedia.org/wiki/...

You mean RISC-V or Power ISA v3.1, right? At this point, I'd steer clear of ARM entirely. ARM Holdings went after Broadcom over its architectural license with a vengeance, and that's not the kind of litigation headache anyone wants to deal with long-term. Apple, on the other hand, had a unique advantage thanks to its deep history with ARM—Apple was an original investor when ARM spun out in the early '90s for the Newton project. That early stake translated into a special architectural license for Apple's A-series and M-series chips, something almost no one else can replicate.

Apple's also in a league of its own when it comes to ISA migrations—just look at the transitions from 6502 to 68k, then PowerPC 32/64-bit, then Intel 32/64-bit, then ARM 32/64-bit. Very few companies have the engineering and software resources to pull off that many platform shifts without losing their user base.

For everyone else, RISC-V and Power ISA v3.1 are better long-term bets because they’re not as encumbered by ARM’s licensing minefield. And since AMD64 ISA and Transmeta VLIW / CMS patents are now well over 20 years old, there could be hybrid approaches as well, mixing RISC-V, Power, and x86-64 in new and interesting ways.

Worst-case scenario, if everything goes sideways geopolitically, we’ll all just be scavenging Xeons from decommissioned servers anyway—so maybe all of this debate is moot. But if there's a real future in high-performance computing, Intel (and everyone else) should look beyond ARM’s locked-down ecosystem.

I'd like to see the return of the Intel OverDrive Processor Interposers so that we can breathe new life into old platforms.

This is a clear violation of the Americans with Disabilities Act and the Rehabilitation Act.

Seriously not even a single joke about di-lithium crystals, what is this world coming to?

We don't need Intel, the x86-64 ISA is owned by AMD, this is why we call it amd64, and at this point this ISA's patients have expired... yes the Athlon 64 is 21 years old now, so amd64 is effectively an open standard at this point. PowerPC and RISC-V are also open standards, so I'm certain that fabless customers will be able to figure out how to make a core for whatever it is that they want to do at this point.

Using Python to print 2/3 in Microsoft Excel, clearly the mark of the beast.