o AI companies are stockpiling not only GPUs for data centers but also small, high-performance computers that can be used to develop local AI agents.
* The content of this article is largely based on reports and industry rumors, so the numbers should be taken as a reference only
o OpenAI has reportedly purchased tens of thousands of Apple Mac mini and Mac Studio computers in recent months.
- The purpose is estimated to be reinforcement learning, Computer-use Agent, and agent workloads
o In other words, the computing demand of AI companies is expanding from existing GPU data centers to local computers for developers and agents.
o Interestingly, AI companies are not only looking for NVIDIA GPUs but also purchasing Apple Silicon in large quantities.
- The combination of Apple's Unified Memory and Thunderbolt 5 provides an optimal environment for driving on-device agent operations
If agents become widespread, all developers and companies will need to be able to run AI independently.
"AI Agent Proliferation → Inference Demand Increase → Local AI Increase → AI PC/Workstation Increase → High-Capacity Memory Demand Increase"
This could be another signal that the AI semiconductor cycle may be longer than expected...ㅎㅎ
On the other hand, it also signals that RAM will not be cheap for several years to come. ㅜㅜ
▶ Original source: https://wccftech.com/openai-hoarding-tens-of-thousands-of-apple-mac-mini-and-mac-studio-devices-as-asus-and-msi-burn-through-their-entire-first-batch-of-nvidia-rtx-spark-chip-and-beg-for-more/
