This.

Culturally it would have been a big shift, even given the talent they have, but they just don't have the courage of leadership it would have taken to do this. It's been 60+ years since IBM had that, when they bet the company on the 360. The PC doesn't count; that was essentially a side project for IBM. They didn't create the hardware or software for it, and the companies future wasn't riding on it.

It's a part of their business, but not a majority of it, even before AI. They've added AI coprocessors to their Telum CPUs for their Z series platform, but it's not a significant player in the world of AI money. More of a checkbox me-too thing that probably will be of use for legacy customer applications, but no one is building data centers full of Telums to compete with OpenAIAnthropticGoogleEtc.

If you're a maintainer, then I suspect whatever you maintain fucking sucks.

I clearly explained the problem- that the quality of LLM-produced output is a function of the money spent to produce it (generally).
This means that the flood of slop PRs are produced by free-to-cheap models. This is only natural.
People who have no idea what they're doing aren't spending $300 for a large commit. They couldn't be bothered to invest the time in learning to code, they sure as fuck aren't going to invest in this.

Meanwhile, over here in real-life, dealing with real problems on maintained things that don't suck, I've got people making quality commits with LLMs on the daily. But they're part of the team, and they're paying real money for those commits.

In short, I think you're probably just full of shit. If you're not, I feel for whatever the fuck it is you maintain. You can't even apply basic logic to a problem.

As someone who is dealing with this problem, it's not that simple.

The problem with your assessment, is that there is no "the LLM".
LLMs come in a fucking vast spectrum of capabilities.
Put pretty simply, very expensive models do very good.
After very expensive, you get a spectrum from "pretty damn alright", to "outright terrible."

There is no "the LLM."
I can tell from your comment that you're talking out of your ass, and not actually dealing with this problem.

Maybe. I doubt it: Intel seems to have figured out that x86 is not it's exclusive future, a change in corporate mentality that long overdue.

Also, this story must be fake news: Tariffs Don't Work! member?

This is the real answer. A lot of people are afraid of the speech Nazi's that ruin the lives of anyone committing wrongthink. They know the social media companies are moderated by privileged tech weirdos that are protected and immune, and readily bury the people they hate with no consequence. Sane people don't engage with stuff like that.

And the case doesn't really rely on the GUID, so it doesn't matter.

If you actually find anyone doing that in the real world, you should point and laugh until they get angry.

Yes, yes, yes. Thing is, there was a clear lack of urgency here. The timelines you cite are for your case, and whatever requirements, budgets and deadlines you suffer. T-Mobile made a bad bet in 2008, and the writing has been on the wall for years now, and viable alternatives have been available at least as long. Were T-Mobile competently managed, they certainly had the means to meet the necessary deadlines. Instead, they made yet another bad bet trying to litigate against pirates.

The correct bet today is container orchestration and open-source based virtualization tools that aren't at the mercy of inveterate rent seekers. There are many ways to skin these cats, and the fact that T-Mobile has slouched into its current unfortunate position is entirely T-Mobile's fault.

Also, the argument that only "two calls" where made and, therefore, a team of 20 people is somehow ridiculous is specious. Support contracts at this scale involve far more than picking up a phone during business hours, and Broadcom will have absolutely no difficulty poking that argument full of holes.

If you're close enough to see an enemy, they already have multiple autonomous weapons coming for you.

Just little old me. This is a big deal. The density increase here is basically Moore's law surviving another decade, with all that that implies. The zdnet puff piece annoys me. ASML is only mentioned in passing. The truth is ASML is right at the heart of this: it's their machine. And that's not me blowing ASML's trumpet: the story is the deeper relationship going on here. The Albany site is basically the US government (successfully) using IBM as their domestic lab operator to facilitate US strategic prerogatives with regard to frontier lithography, which secures US dominance over EUV tech dissemination for many years to come. The reader sees none of this in this zdnet tripe.

When you understand the back story, the future stories make sense. When the US tells ASML to whom they will and won't be selling equipment, and ASML quietly obeys, understanding this stuff means it won't be lost on you why an EU company bends the knee. This is why China can't make iPhone chips or NVidia GPUs, and that this situation is going to persist for years to come because of what's happening right here. Further, it puts the lie to all the yap about the US "falling behind" and failing because "capitalism" and some mythical abhorrence for "public-private" partnership, etc. The US does all of that, and it does this at least as well as everyone else.

A great deal. IBM licenses, partners and consults with semiconductor manufacturers globally, and runs a thriving IP business from their huge R&D facility in Albany, NY. Samsung, Rapidus, AMD, ST, SMIC and others are all paying for IBM tech in recent deals. GlobalFoundries bought out IBM Microelectronics for IBM's 300mm tech. IBM is among the most prolific patent filers in the world.

The real story here is this: ASML has a new machine for a new process node. ASML is obligated to perform much of their R&D in the US due to strict export and technology sharing agreements with the US government. IBM operates huge, world class R&D lab in Albany, heavily subsidized by the state and US government. The new process that this story is about is really IBM working as an R&D partner with ASML to refine the process and get it ready for commercial operation.

In a few years, when they get the yields to something plausible, ASML customers will buy the new machines, and IBM will be in the room, taking their cut for IP, consulting, support etc.

Replaceable? No. Reparable? Depends on the extent, but even that's hard: the wings are full of hardware, and if you have to spend a year and invent a process for dismantling everything to get at the damage, it becomes financial infeasible. Even if you pull it off, you have new inspection requirements, operational limitations, etc.: it's not the same revenue generating plane after something like this.

There is a lot at stake. Emirates operates these with over 500 passengers. If that manifest burns to death on takeoff because a wing folds there will be hell to pay.

I think the end state of all this looks like game cartridges. If people could buy the weights of a frontier quality model in a high density, high speed ROM that ran locally plugged into an PCI-E or M.2 slot, all you would need then is a reasonably fast tensor processor with a little RAM for context. This is possible, and there is even a company (Taalas) with an early product. They have an online demo that is crazy fast.

You would buy one and use it for some time, probably years, and then discard it when the value of some newer version makes sense to you.

Not far from the truth.

The big implication from this is that making these inference engines isn't difficult and they can't be protected. That's the doom of all of AI money: the product is getting commoditized; you'll buy a Claude/ChatGPT/whatever in a box from Amazon in a few years.