$NVDA vs $AVGO & $GOOGL: Why AI ASICs Are Challenging the GPU Empire

$NVIDIA(NVDA)$ GPU vs $Broadcom(AVGO)$ $Alphabet(GOOGL)$ AI ASICs

GPUs handle many AI workloads and are flexible, which is why they are loved by AI start-ups.

However, they contain parts that AI models do not use, making them highly expensive and inefficient when used in certain ways.

ASICs are better at these very specific use cases, as they are designed with those use cases in mind.

Like running a specific type of AI model extremely fast and energy-efficient.

Essentially, $AVGO $GOOGL TPUs are cheaper than $NVDA GPUs at doing exactly what $GOOGL needs to do.

However, the drawback is that $GOOGL can't easily move these chips to do other tasks.

Whilst $NVDA GPUs' capacity can be more easily moved to other tasks within a company.

That's the trade-off.