- cross-posted to:
- usa@lemmy.ml
- cross-posted to:
- usa@lemmy.ml
https://archive.ph/ZQGtt
https://www.ft.com/content/dc47c5f3-9bd4-4da0-a5cb-c795efd14c9c
You must log in or register to comment.
Nvidia's hardware is so important for scientific computing that them getting diverted by the jingling keys of AI is kind of wild. They could make billions encouraging research into climate mitigation as well!
I think I know of get it. For tech billionaires, AI is the perfect product because it's such a floating signifier. It means nothing and so it can be anything. And the only main thing attaching all AI products is running obscene amounts of calculations on mostly Nvidia GPUs.
I'm so excited for the crash. This might be the last shell game big tech can play. All the interest-free investment money has dried up, the product is fundamentally shit, and the amount of data required to make it any better is so massive that they've run out of organic content to farm without poisoning themselves with AI slop. Beyond it, what do they have? They haven't released a worthwhile innovation since the smartphone 20 years ago and the VR headsets that are almost as old. There's no next big thing to drive another hype cycle and no engineering capacity to generate one. And all of the worst tech demons are hedging their entire futures on AI.
You can say that, but they'll find some old, unused because it was non-viable at-scale product like the block-chain or distributed ledgers, dust it off and then hype it up for the next go-around, all while we get passed up by companies and countries that actually understand how to nvent and engineer things. This IS the market cycle for tech now.
It's hard for me to gauge based on two things:
-
They've spectacularly failed three times in the past few years. Cryptocurrency, then NFTs, now AI. That's against a backdrop of every other tech thing either stagnating, being underwhelming, or being an outright scam like Tesla. The technofuturism of the 2010s is increasingly passe as things get worse and the people most obnoxiously pushing it are Malthusian freaks that have spent the past decade becoming popular targets for hate. During that boom cycle they invested everything in stock buybacks, the latest hype cycle, and going public at the expense of ratfucking their service for its users. The sector is facing layoff waves even before the crash. They're starting this cycle in a much worse position even if they're in "too big to fail" territory.
-
They have to go bigger. In 2007 it was "wow you can be totally connected to everything at all times", in 2015 it was "wow we can stop climate change and every car can drive itself", in 2018 it was "wow we can reinvent economics and achieve the Ron Paul Utopia", in 2020 it was "wow we can reinvent creativity and democratise art", and in 2023 it was "wow we built a superbrain that is about to replace most human workers". Each time they've had to go bigger to maintain the same hype. How do you go bigger than "we put god in a machine" without it sounding implausible from the people who couldn't put god in a machine or replace money or solve climate change or do something new with a smartphone in over a decade? Blockchain won't do it, wearable tech won't do it, internet of things won't do it, decentralised <X> won't do it. Unless they have some completely new buzzword in the pipeline, everything they could latch onto next that's even bigger than AI like fusion power is already poisoned by the same hype economy. Whatever they do have, we can be almost certain that it will create the same ecosystem of grifters and scammers who immediately ruin its reputation.
-
Just goes to show all these tech hype are nothing more than capitalism boom and bust cycle in techonolgy space with shiny presentations.
It's corporations profit chasing on new hype tides
fwiw telecoms (or i guess its direct predecessor in many cases?) is so ubiquitous that it's virtually an appliance. It's unlikely that you even know anyone who doesn't use telecommunications at least five times a day. probably a hot take but: AI sucks, but so did the steam engine three years after it's debut. It was a gimmick that broke all the time and, even when it was working in full, could barely be finagled into doing a very small array of specific tasks. And then it got better. Check back on AI in ten years.
The reason for this bubble-looking graph is that all of the capitalization of low-hanging fruit gets completed and then the company has to start operating with slimmer margins. Cisco has almost a hundred thousand employees and 57 billion dollars of revenue, they just aren't growing quite as fast any more.
Open and free models have been released that are almost entirely in lockstep with proprietary ones as far as I know
What do you mean?
Like the competitiveness of open models versus Op*nAI and the like. I don't see a reason to pay myself. The few times I've asked something to spit out the husk of a program in a language I'll never learn or use again, the commercial options were just as competent as the open one that ran locally.
Image generators are pretty much on par whether proprietary or open from what I can tell. Haven't really messed with voice or anything else myself.
What did Cisco do back then? Aren't they still one of the biggest network equipment manufacturers?
If you have like 3 hours, Bobby Broccoli's 2 part documentary on Nortel does kinda go into the absolute madness that telecoms were back then. It was a race to buy up every last little startup with the biggest amounts of money. Dude claims that his new startup does some miraculous thing that it doesn't really do, doesn't matter, here's a bunch of money. Once they kinda faltered and didn't meet dividend projections by like a cent or two the bubble burst for them.
ft.com link
Fess up, who's the liberal here paying $75/month for this?
