Everyone’s watching Nvidia’s earnings report on Tuesday to see whether revenue can beat $109 billion. I think that question no longer matters.
Last week, SemiAnalysis cut its expected average price for Feynman Ultra from about $175,000 to $98,000.
Part of that is because HBM is getting cheaper; the rest reflects a reassessment of pricing power.
Anthropic can choose between TPUs, Trainium, and GPUs from two vendors. OpenAI is developing its own chips.
With backup options in the hands of major customers, the pickaxe sellers can’t just name their price.
However impressive this quarter looks, it still won’t answer the question of what things will look like in 2028.
Nvidia should hurry up and get Jensen Huang back first~~
$NVDAB $NVDA
#英伟达 #AI芯片 #黄仁勋 #JensenHuang
Last week, SemiAnalysis cut its expected average price for Feynman Ultra from about $175,000 to $98,000.
Part of that is because HBM is getting cheaper; the rest reflects a reassessment of pricing power.
Anthropic can choose between TPUs, Trainium, and GPUs from two vendors. OpenAI is developing its own chips.
With backup options in the hands of major customers, the pickaxe sellers can’t just name their price.
However impressive this quarter looks, it still won’t answer the question of what things will look like in 2028.
Nvidia should hurry up and get Jensen Huang back first~~
$NVDAB $NVDA
#英伟达 #AI芯片 #黄仁勋 #JensenHuang
