Sam Altman argues global DRAM supply belongs in data centers

Sam Altman, the chief executive of OpenAI, is pushing back against the trend of running AI models on personal devices. During an appearance on the Cheeky Pint podcast, the executive argued that the global supply of DRAM is better utilized when concentrated in cloud servers rather than distributed across consumer hardware that sits idle for most of the day.

The case for cloud consolidation

Altman characterized the desire for local inference as a miscalculation of resource efficiency. He noted that while developers and users often push for open-source models that can run locally, this approach creates an imbalance in global allocation. From his perspective, keeping high-performance memory in a data center ensures maximum utilization of a constrained resource, contrasting this with personal PCs that frequently use their onboard memory at only a fraction of their capacity.

Global supply chain pressures

This tension comes as companies like Samsung and SK Hynix navigate long-term agreements with major cloud operators. Projects like Stargate have locked up a significant portion of future memory output, creating a scenario where tech consumers pay a premium for hardware while industrial demand dictates manufacturing priorities.

Why this debate matters

The Scale Report understands that this friction highlights a growing ideological split in the tech industry. As AI models scale, the debate centers on whether the future of computing resides in edge-based sovereignty or in the massive, centralized infrastructure controlled by the few companies capable of paying for global chip production at scale.