$NVDA has once again returned to the $220 range. $MSFT has announced Maia 300, its third-generation custom AI chip, with large-scale deployment set to begin next year. It’s the same electricity, but output has increased by 40%—regardless of whether power supply becomes a bottleneck, that extra output directly boosts gross margins. Chip inference efficiency will be the next battleground. $GOOGL’s TPU, $AMZN’s Trainium, $MSFT’s Maia, $Meta’s MTIA, and $NVDA and $AMD’s GPUs. In the end, the real winners will be $AVGO and $TSM—keep going, Huawei! Keep going, China!
