Technology

Add as a preferred source on Google

Altman Calls for Fuller Cloud Memory Use as DRAM Supply Tightens

He said personal-computer memory was used 5% of the time and argued that local open-source AI would not eliminate the need for data centers.

Updated 1 min read
A technician inspects a memory module in a data-center maintenance bay in Santa Clara, California. / TokenPost.ai (macro)
A technician inspects a memory module in a data-center maintenance bay in Santa Clara, California. / TokenPost.ai (macro)

OpenAI CEO Sam Altman urged cloud operators to use scarce dynamic random-access memory (DRAM) more intensively, saying local open-source AI would not remove the need for data-center computing.

Altman said personal-computer memory was used 5% of the time. He did not explain how he measured the figure.

“You want maximum utilization in the cloud if you're trying to deal with this very supply constrained environment,” Altman said.

He said local AI use still leaves a role for data centers and framed fuller cloud memory use as a response to constrained supply.

TrendForce has forecast cloud providers' 2027 capital spending could exceed $1.5 trillion, with memory shortages among factors pushing up AI server costs, in its outlook on cloud investment and component constraints. That forecast concerns investment in future capacity, while Altman's remarks centered on making fuller use of cloud memory during current supply constraints.

Related articles

View all