Local inference
Source reports covering Local inference. Stories will appear when analysis and community reactions are ready.
1 reports · page 1 of 1
Connected topics are temporarily unavailable. You can still browse stories below.
Source reports
OpenAI CEO Sam Altman says it's "quite terrible" for your PC to have RAM — Sam Altman says data centers offer better utilization
news
5 min read
OpenAI CEO Sam Altman said distributing large amounts of DRAM across personal computers is inefficient while memory supplies are constrained, arguing that cloud data centers can use the hardware more intensively. In an interview with YouTube channel Cheeky Pint, Altman described local AI inference and home computers holding DRAM that may be used only 5% of the time as “quite terrible” from a global allocation perspective. Altman said maximum cloud utilization is preferable when DRAM is supply-constrained, challenging expectations that people will increasingly run open-source AI models on PCs they already own or build. OpenAI’s main business relies on closed-source models operated in data centers, although the company also has open-source models. Local AI could allow some users to rely less on subscriptions to cloud-hosted models if their own hardware is powerful enough to run comparable software. The interview also covered recently disclosed PHP vulnerabilities, sandboxing AI models during training, and OpenAI’s mathematics research.