Sam Altman argues that it's bad you want DRAM in your computer and not a data center, actually

Sam Altman, CEO of OpenAI, contends that having personal computers with significant amounts of Random Access Memory (DRAM) for local AI computations is suboptimal. In a recent podcast, he argued that it's poor allocation of the scarce global DRAM supply to have individual machines using it only intermittently. Altman emphasized the importance of maximizing utilization in cloud-based AI services.

He noted that many people might believe the opposite, but his perspective is aimed at addressing the inefficiency in the AI industry rather than targeting individual gamers. Despite his concerns, Altman's criticism is somewhat ironic given that OpenAI, the company he leads, contributed to the very memory scarcity he's discussing.

He also pointed out the negative impact of AI infrastructure development on environmental sustainability goals set by big tech companies. However, Altman chose to focus on a more positive angle, expressing satisfaction in the ironic situation where AI systems might be consuming their own resources. He cited examples like Google's AI features potentially squeezing ad revenue from Search, Amazon's data center buildout leading to negative free cash flow, and the recent downward trend in AI stocks, which resulted in a sell-off.

While it's unclear if the AI bubble has fully burst, the situation appears to be getting more problematic.