# Sam Altman Advocates for Centralized DRAM Allocation Over Local AI Inference OpenAI CEO argues local device memory usage is inefficient for global AI scaling. - **Source:** [Gizmodo](https://gizmodo.com/) - **Published:** 2026-10-31 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/1adeb6cc-2a06-4129-8978-612eb6682811 ## Centralization of Global DRAM OpenAI CEO Sam Altman recently argued that individual device owners having access to DRAM is inefficient. During the Cheeky Pint podcast, he suggested that local AI inference wastes limited memory resources. He believes global allocation should prioritize data center utilization rather than allowing hardware to sit idle in personal consumer computers. ## The Efficiency Argument Altman claims that personal computers use their memory capacity only five percent of the time. He posits that in a supply-constrained environment, maximum utilization in the cloud is the only logical path forward. This perspective challenges the growing trend of running open-source models locally on consumer hardware to maintain privacy and control. ## Infrastructure Cost Pressures The massive demand for AI data centers is driving up the market price of DRAM significantly. OpenAI plans an infrastructure rollout costing 750 billion dollars through 2030. This aggressive capital expenditure directly impacts the cost of components for everyday devices like smartphones, making high-performance memory increasingly expensive for the average consumer. ## Platformization and Control The proposed shift suggests a future where users delegate basic computing tasks to OpenAI cloud agents. By moving computation to centralized data centers, the company aims to consolidate platform control. This model assumes that users will eventually prefer paying for cloud-based AI services over maintaining their own high-cost, local hardware configurations. ## Market Value Concerns Investors are expressing growing anxiety regarding the massive AI spending spree currently occurring in Silicon Valley. Alphabet has already lost roughly 500 billion dollars in market value amid these concerns. Furthermore, only one of the four major AI hyperscalers reported positive free cash flow during the most recent round of earnings. ## Hardware Supply Constraints The scarcity of high-speed memory like 32GB DDR5 is keeping prices elevated, with costs reaching 450 dollars or more. These supply constraints are exacerbated by the insatiable appetite of AI data centers. As long as demand from hyperscalers continues to outpace production, consumer hardware prices are unlikely to see any significant downward relief. ## Geopolitical Chip Dynamics Beyond memory, the broader AI hardware landscape faces intense regulatory pressure. Nvidia CEO Jensen Huang has been lobbying the U.S. government to ease restrictions on selling powerful AI chips to China. These geopolitical tensions add another layer of complexity to the global supply chain, affecting both data center expansion and consumer hardware availability. ## The Future of Computing Altman’s vision highlights a fundamental tension between decentralized personal computing and centralized cloud-based AI dominance. While he frames this as an optimization problem, it represents a significant shift in how digital resources are distributed. Whether users will accept this trade-off remains the central question for the future of the personal computing industry. ## Sources covering this story - https://gizmodo.com/sam-altman-all-the-ram-should-belong-to-me-so-i-can-charge-people-to-use-it-2000833412 — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._