OpenAI CEO Altman: Putting DRAM in personal computers is a waste of resources; it should be prioritized for AI data centers

Add crowdsourced testing promotional copy Add crowdsourced testing promotional copy end Add App-exclusive traffic redirect

Kuai Technology November 1 news: Global artificial intelligence (AI) demand continues to push up memory chip prices, but OpenAI CEO Altman has questioned the practice of keeping a large amount of memory chips in personal computers. He believes that, from the perspective of global resource allocation, letting limited DRAM be scattered across personal computers and sit idle for long periods is not an efficient arrangement; memory chip resources should be concentrated in cloud data centers to improve overall utilization.

In a recent exclusive interview with "Cheeky Pint," Altman discussed local inference of AI models and the demand for memory chips in personal computers. He pointed out that if everyone runs local AI models on their own computers, it could lead to a large amount of DRAM resources sitting idle.

Altman said: "From the perspective of global resource allocation, having everyone leave the world's limited DRAM memory chips idle in their own computers, with a utilization rate of only 5%, is actually very bad."

He pointed out that amid extremely tight memory chip supply, cloud data centers should maximize resource utilization, but many people intuitively have a different view.

As the AI wave drives up demand for memory chips, global DRAM supply is tight, and consumer electronics such as PCs and phones are also facing cost pressure. Although the memory chip specifications used in personal computers, phones, and AI accelerator cards are not all the same, the production of related products will still compete for some upstream wafer capacity.

Altman also discussed personal AI agent products that have recently attracted attention, saying that the market has not yet seen a product that users cannot imagine living without. He expects that successful agents in the future will still have some characteristics of existing products, but must be more reliable and able to handle more complex tasks.

On the business model side, he believes companies should not rely excessively on their own websites or product interfaces to maintain customer relationships, because consumers prefer to have AI complete various tasks on their behalf directly through major chat applications. He believes that companies' true competitive advantage should come from brands, product selection, and fulfillment networks, rather than the website interface itself.

As for how AI agents are charged, Altman revealed that OpenAI had considered following Amazon Web Services (AWS) and switching to task-based pricing rather than charging by token usage, but customers reacted strongly, believing they must clearly know what they are paying for, so they still prefer token as the pricing unit.

[This article ends] If you need to reprint, please be sure to indicate the source: Kuai Technology

Editor in charge: Lujiao article_adlist[

Article content report ]article_adlist Statement: Sina.com exclusive article, unauthorized reproduction is prohibited. tech_web_article_v2015_block_bottom non-targeted 300*250 button 17/09 wenjing begin

non-targeted 300*250 button end