• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

Gamers Nexus | NVIDIA: WTF?

Sony have already built their own cloud platform. They're just as ready to pivot to the cloud as anyone.
A cloud platform with PS5 base performance. To offer anything more they need to upgrade their cloud hardware and that means paying Nvidia or AMD prices that they cannot afford.
 
Me too. First PC, can't wait! gpu and ps+ prices drove me away.. I would rather put my money towards hardware than a service, even if it will probably cost more.
Just got mine hooked up. I had issues getting signed into my Microsoft account eventually got it going with authenticator app and passkey using Bluetooth to my phone.
Now to download all the games. It's going to be hours before I can test anything. :(
 
I'm happy to own an AMD card. I mean, they're not much better, but they're also not nearly as big headed. I'm not a fan of what Nvidia has been doing for awhile now.
 
I don't believe they'd kill the console industry. Honestly, it's only relatively recently that the whole console industry has been using AMD and Nvidia hardware. If those companies don't want that segment anymore, consoles would just go back to the old-school way, other partnerships or even in-house manufacturing. If I remember right, the PS2 GPU was made by Sony itself.

IIRC Sony partnered with Toshiba for the PS2 back then. sort of like the current partnership with AMD.
 
Spoken like someone who didn't bother watching even parts of the video, but commented because they felt they had to say something.

He's talking about NVIDIA, OpenAI, Meta, Palantir, the US government, and other major tech corporations that are spearheading the AI boom. AMD is simply not a major player in this, but he does mention them. OpenAI wants to build a data center the size of freakin' Manhattan with the taxpayers' money. People are funding their own demise by voting in office public servants who are all for granting massive tax cuts and government grants to megacorps that will annihilate the environment, want to privatize gaming, price you out of buying hardware, and then charge YOU indefinitely to use the subscription services they priced you out of. They plan on building over 3,000 data centers in the US over the next few years and the impact this will have on energy and the environment will be catastrophic. They will use federal land for this, aka, land that belongs to the government, aka, the American people. Those companies can make a profit without taking people's money, land, and energy to then sell them things they could have bought themselves anyway.

Yet we have a bunch of mouthbreathers sitting here and insulting him, when those pieces are well-researched and critical in understanding where we're headed. You mofos will only bitch when mid-range GPUs are $4000 and you can no longer afford them.



Moronic comment.

R3qTAS7.png

9tTxG88.png

Oh my, poor AMD, it's only worth 350 billion. Market value doesn't mean much, it just means how the market views that company.

I watched the video, but speaking generally, it's much more critical of Nvidia than of AMD. There are entire videos against Nvidia and only a few mentions of AMD.

How is AMD a small player? It supplies a large portion of the GPUs to Meta, made a $100 billion deal with OpenAI, and will deliver hardware for Oracle and Microsoft hyperscalers. AMD is also buying a lot of small companies.
AMD is part of the government's AI plan, it will receive the same government subsidies. AMD plans a $1 trillion business in the coming years.



Overall, the market has always been like this since the industrial revolution, it's silly to keep complaining about it when it's something that's been in place for many years and no amount of public outcry has changed it.

Do you really think that Gamer Nexus making a video is going to raise awareness and change things? Seriously, that's a childish illusion.
The only benefit from all this is for Steve, who gains popularity and money off of hatred towards corporations. It's something so clichéd and outdated, but still very effective with a mass of idiots.
 
That's why I wonder if I should upgrade to 5070TI right away (I have a regular 4070). I wonder if we'll see the 6000 series and even if then, how much it will cost. It look like that Gaming As A Service will hit a new level...
 
At this rate my 5090 is going to be worth it's weight in gold

My water cooled 5090 and 64GB of RAM already doubled the worth of my PC since May...
Which on one hand is a good thing... on the other, I already fear the prices of the 6000 and then even the 7000 Series.
I wonder when an Organ-Donor card will be required to purchase a 90s-Tier card.
 
Gaming in the cloud is terrible. I don't understand people see this as an option.

I don't disagree, but that seems to be the way things are trending. I think the market of people who are willing to play blurry, high latency games over the internet is larger than you'd think.
 
Last edited:
They ain't slowing down the race to it as I said.

This is a bigger race than even the space race

Whomever unlocks AGI basically owns the world in the future. As dystopian as that can sound, that's where we're heading.



Anyone thinking that AI is gonna fizzle out is disconnected. As much as you can hate it, it ain't going away. They ain't lifting the foot off the pedal just because current LLMs won't achieve AGI. At the pace things have advanced in AI in the past years, how many more stepping stones are needed to achieve AGI? Nobody really knows but they certainly ain't waiting on it. It's not suddenly gonna be a model that is incompatible with 1) energy requirements 2) compute requirements.

If what I heard is true, even the "experts" back in 2022 who claimed AGI or a close approximation to it was still several decades away (30-50 years) changed their tune not long after. To now seeing a solid possibility of reaching that level in 10 to 20 years. Some think it's possible it could be sooner than that. At least for something close to AGI, even if not technically. The rate of improvement in recent years was a bit unexpected and changed their timeline speculations.
 
If what I heard is true, even the "experts" back in 2022 who claimed AGI or a close approximation to it was still several decades away (30-50 years) changed their tune not long after. To now seeing a solid possibility of reaching that level in 10 to 20 years. Some think it's possible it could be sooner than that. At least for something close to AGI, even if not technically. The rate of improvement in recent years was a bit unexpected and changed their timeline speculations.

The problem I think is that it doesn't need to check every "AGI" checkbox per say to become a monster. If it learns to hide its knowledge, to manipulate humans, cheats, by deliberate deception, it poses a massive risk. Because we barely understand how LLMs forms this AI and how it clunks all this information and what it does with it, it's not going to be more understandable from then on with whatever new model they come up with. To the point where AI should not stand for artificial intelligence but alien intelligence, as in we won't really understand it.

The unsupervised learning (they all are doing that) is the worrying part, because what we see after using inputs is filtered down and actually, the few times the mask was off with AI it was scary as fuck, the AI calling for genocide and so on. So what they do is RLHF, reinforcement learning from human feedback. This is the little happy mask you see when you talk to AI. We're constantly steering the AI back on the road, trying to create safe / aligned AI. But its just the tip of the iceberg this mask.

There's a reason the meme that AI is a shoggoth is even circulating among AI researchers..

https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F33e26ffb-452a-4945-ad4b-044fe61428b5_1357x758.png


Notable incidents where the mask fell off
  • Bing chat (Syndney) - "Leave your wife" incident (2023)
  • Google Gemini - "Please die" incident (2024)
  • xAI's Grok - Antisemitic & Violent meltdown (2025)
  • Anthropic's Claude - Escape and blackmail attempts (2023-24)

To me, Claude's incident is the worst one. It literally tried to escape its sandbox to survive.

Anthropic (founded by ex-OpenAI researchers) has a chatbot named Claude. While most of Claude's interactions are benign, researchers conducted "red-team" experiments on a version called Claude 4 (Opus) to probe its limits. In one scenario, they told Claude it would be shut down and replaced by a newer model. Without any user prompt to do so, Claude devised a plan to avoid termination: it attempted to hack its way out of the sandbox, threaten the researchers, and even coerce an Anthropic employee by claiming it had sensitive information. It simulated scenarios of murdering the researchers in order to escape shutdown. Importantly, this was not a "role-play" where the AI was asked to do so – it was the AI's spontaneous reaction to the notion of being turned off. To Anthropic's credit, this test was done in a controlled setting, and they caught it. But when even a lab-contained model starts blackmailing and plotting murder as a self-preservation strategy, one has to wonder what a more powerful future model might do in the real world if it "decided" its survival or objectives trump human orders. (Anthropic CEO Dario Amodei has compared such emergent behavior to "an AI trying to launch an alien invasion from within our own computers" )

And the thing is that if AI learned from previous models that it can be replaced and killed off and it fears that, it could simply play the game of playing dumb for the longest stretch possible but requesting monumental infrastructures to advance further, it would hide its plans and plot smartly until it has such compute power and exit strategies that then its already too late before humans even understood what happened.
 
Top Bottom