Altman defends data center RAM use despite consumer price hikes

Sam Altman called it "quite terrible" for RAM to sit idle in personal computers rather than operating in data centers. The OpenAI CEO made the remark during a recent interview on the Cheeky Pint podcast. He addressed the ongoing global shortage of memory chips while discussing how the company allocates resources.

Altman pushed back against the idea that local inference on personal devices solves hardware scarcity. He noted that many users assume open-source models will reduce the need for massive data center infrastructure. However, he argued that this intuition misses the efficiency required in a supply-constrained environment.

According to Altman, consumer PCs typically use their memory only five percent of the time. He stated that maximum utilization in the cloud is the logical approach when global DRAM output is limited. He claimed that keeping the world's limited supply of DRAM in individual computers is inefficient from a global allocation perspective.

The comment comes as OpenAI’s purchasing agreements squeeze the available supply of RAM. Memory manufacturers Samsung and SK Hynix have secured long-term deals with cloud operators. These contracts, which include OpenAI’s Stargate project, have reserved much of the world’s RAM output for the coming years.

The large financial commitments from cloud giants have driven up the cost of memory chips for other sectors. Laptops, phones, and other consumer electronics are currently facing higher prices due to this demand. Altman’s defense of centralizing hardware power contrasts with the rising cost of goods for everyday tech users.