The idea that ASICs will take over and kill GPUs is wrong. Trainium from $AMZN and TPUs from $GOOGL will grow. But by 2030, ASICs will be ~25% of the AI accelerator market. Not dominant. There is no “one size fits all” in computing — as Lisa Su says. The reason GPUs will remain dominant for years is simple. We are still in the early innings of the AI supercycle. At this stage, developers want flexibility to innovate across models and algorithms. GPUs provide that flexibility at a level ASICs simply can’t match. $NVDA and $AMD are here to stay and dominate. They might lose a few points of market share. But in a market growing at ~60% annually, that is nothing.
View on X ↗