
Anjney Midha, an investor in Anthropic and Mistral, argues that GPU utilization, not just access, is the real bottleneck in AI scaling. He notes that xAI runs at sub-10% Model FLOPs Utilization while best-in-class reaches 60-70%. Midha now builds AMP's independent compute grid to address this systems problem.
Tap to vote and see what everyone thinks.
Summary by ByteBrief
Apple Faces a No-Win Pricing Dilemma. Should Investors Be Worried?