OpenAI CEO Sam Altman Claims That Your RAM Sits Idle 95% of the Time!

Sam Altman speaking at TechCrunch Disrupt San Francisco 2019, San Francisco, California, October 3, 2019.

What is going on?

In the last year, I have gone through not knowing who the hell Sam Altman is, to I have serious questions on the guy’s morals. To be fair to him, I haven’t heard anything about him other than all he is contributing to the AI Datacenters, and global RAM crisis, so my opinion towards him is kind of heavy on one sided. Recently I have watched three different videos talking about an interview Altman did, where he talks about the need for more RAM, and that consumers are just wasting RAM when it could be used in data servers, running cloud AI. Altman claims that your RAM is idle 95% of the time. I don’t know why but I find this very personally attacking. Who the hell does a CEO of a LLM AI Company think he is telling me what to do with my RAM that I’ve spent my hard earnings on! Under no universe is that appropriate. This is where I passionately believe that companies do not care about consumers anymore which doesn’t make any sense. The entire point of making goods is too be distributed to consumers. Furthermore, is Altman an idiot? If all the RAM should be used for not just AI but any cloud use, how is anyone supposed to be use the cloud resources? Cloud resources are only available through the internet, which normally is going to require an internet browser, which normally use a fine chunk of RAM. This is like a CEO looking at a neighborhood going, all of that land should just be mine. It’s not like you’re using your house more than 5% of the time anyway.

Linus Tech Tips Makes His Own LLM Server

So naturally in retaliation to this, because this is a massive blow to those who do home labs or gaming rigs, Linus Tech Tips makes a video about building their own AI server for local use using a LLM that runs locally on the existing hardware. The video itself is mildly interesting, and there are a handful of things to learn from it, but even after I believe two 5090’s, the LLM is still mildly slow. That is 64GB of DRAM alone!!! I knew that AI and LLM uses a lot of GPU and RAM but seeing it in action still is baffling. It does show that Local AI processing is still going to be awhile yet, but there is a possibility, but only if CEOs such as Sam Altman don’t win the race they want to run.

Computer Hardware Just Straight Locked Away

Zach Shutt, a content creator who also owns a business for building gaming PC’s tends to go over these kinds of things, and this issue is no exception. He made two videos on the topic labeled “Your PC is ‘Terrible’ Now” and “ChatGPT Came For Your PC“. Both of these I would recommend if you are looking for more information regarding this whole ordeal, as well as a business owners standpoint on this. Most of what I’m going to be referencing is the second video “ChatGPT Came For Your PC”, where he goes more down the hardware scope of small, business owners are being blocked from making purchases such as a 5090 from Micro Center due to “Policies”. Another user got his Amazon account permanently deleted because he was trying to buy I think 4 5090’s at once. There was also light found that the AI computing power on basically all Nvidia graphics card since I believe the 20’s series is limited 50% what it can actually do, especially the 5090 basically sharing a chip as the RTX 6000 Pro Blackwell.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top