0
Applied AI·September 7, 2026·1 min read

Jensen Huang says AGI has arrived, and 400,000 more GPUs are coming

Share

Calling GPT-6 Astra “AGI” is less about definitions and more about justifying another 400,000 GPUs of demand — the message is that the frontier train isn’t slowing. For operators, the practical takeaway is simpler: expect continued model step-changes and plan your architecture so you can swap in new capabilities without rewriting everything every 12–18 months.

Applied AI

Ryzen AI Max+ PRO 495 mini PCs with 192GB RAM emerge — Acemagic, GMKtek, Framework debut ultra-expensive 'supercomputers' that can handle 300B-parameter

Mini PCs with Ryzen AI Max+ PRO 495 and 192GB RAM claiming to run 300B-parameter models locally show how quickly frontier-class inference is leaking out of the data center. If your product assumes “big models must live in the cloud,” start stress-testing on-prem and edge variants—customers with data sensitivity or latency needs will ask.